Query processing (i.e., searching)

System and method for database retrieval, indexing and statistical analysis

6385611

Abstract

The present invention provides a system and method with the capacity to compare and analyze keywords of a specific area of study. By the use of the methods of the present invention, some sets of keywords will be seen as "warming up" due to their upward trends, whereas other keywords might be seen as "cooling down" due to their downward trends. Given the accepted fact that growing areas of research are the ones that are more likely to produce scientific breakthroughs, the system identifies these emerging ("hot") areas of research would accelerate the scientific advances of their users. Similarly, users will be able to view and shift from non-productive ("cool") areas of research to productive "hot" areas. The invention involves the utilization of a commercially available database program and provides specific keywords associated with the investigated topic. The present invention also provides a method for indexing the keywords using a keyword tree structure database so the data is in the correct format for analysis. The invention also provides a method for analyzing the number of occurrences of keywords along with the analysis of an impact factor associated with the keywords. The formatted data then allows the construction of several charts so a user can easily assess the state and forefront of a specified topic.


Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A method of analyzing research trends, comprising:

(a) selecting plurality of keywords associated with a trend;

(b) searching a database of preselected keywords, wherein the preselected keywords are generated from a plurality of publications;

(c) associating each keyword with a value indicative of the prestige of the publications in which the keywords appear;

(d) determining a quantified value by summing the value indicative of the prestige associated with each unique keyword, thereby creating a prestige total associated with each unique keyword, summing the number of times a keyword appears in the plurality publications, thereby creating a frequency total for each unique keyword, and multiplying each frequency total with the prestige total associated with each unique keyword, thereby creating the quantified value; and

(e) determining the research trends based on the quantified value.

2. The method of claim 1, wherein the database further comprises specific and nonspecific keywords, wherein the method further comprises, separating the specific keywords from the nonspecific keywords.

3. The method of claim 1, wherein the research trends are determined by predicting future growth trends.

4. The method of claim 1, wherein the value indicative of the prestige of the publications is an impact factor.

5. The method of claim 1, wherein the value indicative of the prestige of the publications is a publication cited half-life value.

6. The method of claim 1, wherein the value indicative of the prestige of the publications is an immediacy index value.

7. The method of claim 1, wherein the searching step includes a search of documents based on a search prioritized by the publication date of each of said documents.

8. The method of claim 1, wherein the searching step includes the use of keywords of a specialized area.

9. The method of claim 1 wherein the predetermined topic is related to a field of authors.

10. The method of claim 1 wherein the predetermined topic is related to a field of institutions.

11. The method of claim 1, wherein the searching step, comprises:

identifying a group of specialty journals;

entering the group of speciality journals into a database query, wherein the database query is executed on the database to retrieve a plurality of articles, wherein each article of the plurality of articles contains a plurality of keywords associated with the article;

associating each article title with a corresponding factor value; and

associating each keyword in each article with the factor value associated with the article containing the keyword.

12. The method of claim 11, further comprising the step of removing the non-focused keywords.

13. The method of claim 11, further comprising the step of removing the sub-headings from the plurality of keywords.

14. The method of claim 11, wherein the factor value is an impact factor value associated with a publication.

15. The method of claim 11, wherein the factor value is a publication cited half-life value.

16. The method of claim 11, wherein the factor value is a immediacy index value.

17. The method of claim 11, wherein the factor value equal to the number of times each keyword is cited in the plurality of articles.

18. The method of claim 11, further comprising,

selecting a plurality of top keywords from said plurality of keywords based on the quantified value.

19. The method of claim 18, wherein the research trends are determined by:

producing a chart from the number of occurrences each time a unique keyword appears, the multiplying factor, and the impact factor, thereby creating a chart pattern; and

determining if the chart pattern is of an increasing value, and if so, concluding that a topic related to the keywords is emerging.

20. The method of claim 18, further comprising the step of determining if the chart pattern is of a decreasing pattern, and if so, concluding that the topic related to the keyword is declining.

21. The method of claim 1, wherein the searching step, comprises:

selecting a plurality of keywords from a tree structure in a bibliographic database, wherein the tree structure in a bibliographic database provides a plurality of subcategory keywords;

selecting a plurality of related keywords from the plurality of keywords and the plurality of subcategory keywords; and

entering the plurality of related keywords into a database query, wherein the database query is executed on the database to retrieve a plurality of articles.

22. The method of claim 21, further comprising the step of removing a plurality of non-focused keywords from the plurality of related keywords.

23. The method of claim 1, wherein the searching step, comprises, selecting a plurality of journal titles from a list of specialty journal titles, wherein the selection process is based on an impact factor.

24. The method of claim 1, wherein the searching step, comprises, selecting a plurality of journal titles from a list of non-specialty journal titles, wherein the selection process is based on an impact factor.

25. The method of claim 1, wherein the searching step, comprises, selecting a plurality of journal titles from a list of a plurality of specialty and non-specialty journal titles, wherein the selection of journal titles occurs if, a corresponding impact factor of the selected non-specialty journal is greater than the impact factor of any specialty journals.


Description

FIELD OF THE INVENTION

The present invention generally relates to database retrieval methods, and more particularly, to a search method that provides quantifiable analysis for determining the distribution and evolving or declining topics within a specialized area.

BACKGROUND OF THE INVENTION

Databases are software tools that contain records. These records are arranged in different fields. Some of the fields that can be found in a bibliographic database might include some of the following: title of the book or article, authors, institution(s), source, abstract, keywords, bibliography, etc. This sort of arrangement allows the user to search in a field or a combination of fields (utilizing Boolean terms; and, or, not) to find desired information.

The bibliography field typically contains one or more citations. A citation may acknowledge the source of another document that is cited as a support for a point of view, or as an authority. In the past, Eugene Garfield, Ph.D. at the Institute for Scientific Information.RTM. (ISI) has used the bibliographic field format to produce statistical data. One of these sets of statistical data pertains to what is known as the Citation Index by which the frequency of a citation reflects the impact that such publication has had within its discipline. This concept is known as Impact Factor. ISI publishes on a yearly basis a list containing thousands of journals according to their ranked Impact Factor and their relation to a plurality of specific subjects (Journal Citation Reports.RTM., JCR.RTM.). Other lists provided by ISI categorize journals by immediacy index, citation half-life, total number of citations, etc.

ISI also works in the area known as bibliometrics. As proposed by Pritchard in 1969, bibliometrics has been defined as "the application of mathematics and statistical methods to books and other media of communication". Thus, ISI has been addressing the following questions: What are the largest journals? What journals are the most frequently used? What are the "hottest" journals? What are the "hottest" articles? What journals have the highest impact factor? What publications does a journal cite, and which ones cite it? What is the historical origin of a new topic, etc.?

ISI and others have expanded this technology and are now capable of addressing other questions outside of the bibliographic field, thus giving rise to the science of informetrics. Other questions being addressed include authorship, country, institution and journal analysis, etc. The areas impacted by this type of analysis cover a wide-spectrum of subjects including: broadcasting, ethics, geology, psychology, management, chemistry, biology, medicine, etc.

Despite this progress in the informetrics field a couple of fundamental questions remain unanswered: What is happening in a specific topic; and what is at the forefront of that topic. Prior methods for predicting trends in specialized areas include hiring consultants that provide an opinion on the evolving areas. Some disadvantages with using consultants are that: (1) Consultants are expensive and typically limited to only their specific area of expertise; and (2) Consultants may incorporate their preferences into the opinions without easy detection. Thus, their opinions might be biased. A research facility interested in directing their research efforts at the evolving areas, to maximize their funding for research projects must expend a large amount of money to hire multiple consultants that will predict trends for each specialty within the research facility. Another disadvantages is that even after this large expenditure of money, the research facility still does not know the relative funding to apply between the specialties due to the speculative nature of consulting.

Databases contain a concept tree structure composed of keywords. These keywords are tags assigned to each article in the database so that any user can retrieve the same articles with consistency. Thus, these keywords operate as a means to recognize specific articles related to that topic. Prior art computer retrieval systems have been able to combine specific keywords with a set of journals. However, these systems lack the statistical analysis, depth, integrity, comprehensiveness and completeness required to consider these studies scientific. Furthermore, there are no published studies or systems that address the specific questions of what is happening in a specific topic, or what is at the forefront of that topic by the use of keywords database and/or keyword tree structures. Given these limitations, there is a need for a system of analyzing trends in research in a way that is efficient, unbiased and reproducible. In addition, a system that would allow a user to know the distribution of specific keywords in a given topic does not exist.

SUMMARY OF THE INVENTION

The present invention provides a system and method with the capacity to compare and analyze keywords of a specific area of study. By the use of the methods of the present invention, some sets of keywords will be seen as "warming up" due to their upward trends whereas other keywords might be seen as "cooling down" due to their downward trends. Given the accepted fact that growing areas of research are the ones that are more likely to produce scientific breakthroughs, the system identifies these emerging ("hot") areas of research that may accelerate the scientific advances of users. Similarly, users are able to view and shift from non-productive ("cool") areas of research to productive "hot" areas.

The process involves the utilization of a database program and provides specific keywords associated with the investigated topic. The present invention also provides a method for indexing the keywords using a keyword tree structure so the data is in the correct format for analysis. The process also provides a method for analyzing the number of occurrences of keywords along with the analysis of an impact factor associated with the keywords. The formatted data then allows the construction of several charts so a user can easily assess the state and forefront of a specified topic.

The process involves the input of the name of the journal to be investigated, removal of the none-original articles such as editorials, news, comments, etc from the query built in the retrieval process, limiting the query by the different years to be investigated, and downing the articles according to the years or group of years to be investigated.

In another embodiment, the process is best suited for the study of a keyword or a small set of keywords that do not require a pre-search to find out the specialized area keywords. The purpose of this process is to find additional keywords that other keywords relate with. Thus, this embodiment does not require to select any journals from the database since the user wishes to know which keywords relate to its query regardless where it is published. Input the keyword(s) into the query and remove none-original articles such as editorials, news, and comments from the query, limit it by the different years to be investigated and download the articles according to the years or group of years to be investigated. This process accounts for all of the focused keywords. The process of this embodiment then calculates a correction factor and applies it following pre-indexing process. The process then continues to the index step to sort the keywords. The process then proceeds with the statistical analysis.

This type of investigation provides the user with a tool to know which are the areas related to the keyword or small set of keywords under investigation. One interesting aspect of it is to find out new correlations of this keyword(s) with unsuspected topics. This type of search is particularly appealing to anyone searching for new uses. For instance, very often pharmaceutical compounds have multiple applications. Novel research that applies to a related compound to the one being investigated might be picked up by the user's search since they might be sharing higher hierarchical keywords.

In another embodiment, the process combines the specialized keywords and a selection of all the specialty journals available with the top non-specialty journals with the highest impact factor than the best specialized journal. The question best addressed with this system is "what is happening in a specialized area of research."

The information provided by this method provides managers with a novel tool to establish current needs and anticipate future requirements that will ultimately maximize their efforts and gains. Beneficiaries of this system would include the following: scientists, managers, strategists, venture capitalists, investment bankers, foundations, information and market analysts, publishers, historians, etc. At the institutional level, the beneficiaries include: companies, non-profit organizations, research centers and governments agencies. The present unbiased and quantifiable system and method allows a user to see the reality of past and present topics' distribution and trends within a specialized area. Moreover, by extrapolation of the data a forecast of future trends is made possible.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

FIG. 1 is a block diagram of a computer in accordance with the present invention;

FIG. 2 is a block diagram of a plurality of computing devices of FIG. 1 connected to a network;

FIG. 3 is a flow diagram depicting the three major steps of a database analysis process according to one embodiment of the present invention;

FIG. 4 is a flow diagram depicting one embodiment of a retrieval process of the present invention, wherein the retrieval process analyzes the keywords in a plurality of articles;

FIG. 5 is a flow diagram depicting another embodiment of the retrieval process of the present invention, wherein the retrieval process utilizes a keyword tree structure;

FIG. 6A is a flow diagram depicting another embodiment of the retrieval process of the present invention, wherein the retrieval process utilizes a plurality of specialty journals;

FIG. 6B is a flow diagram depicting yet another embodiment of the retrieval process of FIG. 6A, wherein the process utilizes a non-specialty journals in combination with the specialty journals;

FIG. 7 is a flow diagram depicting yet another embodiment of the retrieval process, wherein the retrieval process utilizes keywords of a pre-search and specialty and non-specialty journals;

FIG. 8 is a flow diagram depicting one embodiment of the index arrangement process, where the index arrangement process provides a correction factor;

FIG. 9 is a flow diagram depicting one embodiment of the index process of the present invention;

FIGS. 10A-10F are data charts generated from the data formatted in the index process;

FIG. 11 depicts one embodiment of an output of the system in the form of a bar chart, wherein the bar chart is formatted to indicate emerging and declining specialty topics according to different groups of years;

FIG. 12 is another embodiment of an output of the system in the form of a bar chart, wherein the bar chart is formatted to indicate what is happening in a field and what is the forefront of a topic; and

FIG. 13 is another embodiment of an output of the system in the form of a bar chart, wherein the bar chart is formatted to indicate emerging and declining specialty topics according to different groups of years.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

The present invention is directed to a method for determining distributions and trends in a specialized area. The method involves counting keywords and organizing the keywords according to a pre-determined keyword-tree structure. The method operates in a computing environment with access to databases containing journals and their respective articles. JCR.RTM. ratings that determine and classify journals by impact factor and their subjects are used to operate the system.

The present invention is a quantifiable database system and method consisting of three sequential steps: keyword retrieval, indexing and statistical analysis. The output of the statistical analysis allows a user to determine the distribution and trends in a specialized area of interest. The area of interest can be a very broad topic such as biomedical research or clinical neurology. Alternatively, the area of interest can be of an intermediate breadth such as schizophrenia; or the area of interest can be a more narrowed topic such as aspirin. The system gives the user an overview of a selected topic and defines the areas that are most relevant. The method of the present invention provides analysis for two questions: (1) what is happening in a specialized area and, (2) what are the emerging topics, referred to as getting "hot," as well as the topics that are "cooling down."

Elements needed to Operate the System

(1) Computing and Network Environment

The present invention may be embodied in a computer program or in another type of program module embodied in a set of instructions encoded on a computer readable medium such as a tape, CD, disk drive or the like or transmitted over a communication link such as the Internet. FIG. 1 depicts several of the key components of a computer 100. Those of ordinary skill in the art will appreciate that a computer 100 includes many more components then those shown in FIG. 1. However, it is not necessary that all of these generally conventional components be shown in order to disclose an illustrative embodiment for practicing the present invention. As shown in FIG. 1, the design of the computer 100 includes read-only memory ("ROM") 103, a central processing unit 102, a video display adapter 109, and random access memory ("RAM") 104. The ROM 103 stores an operating system 111 for controlling the operation of the computer 100. The RAM 104 also includes a medium to store a software application 110 as described in more detail below. It will be appreciated that these software components may be stored on a computer-readable medium and loaded into memory 204 of the client computer using a drive mechanism associated with the computer-readable medium, such as a floppy drive, a CD-ROM/DVD-ROM drive 106, or a hard drive 108. A relational database may also be stored on the hard drive 108 for assisting the process of the software application 110.

An input/output interface 105 may also be provided for receiving input from a mouse, keyboard, or other input device. In addition, standard components such as a video display adapter 109 could be included to provide an output to the user. The computer 100 may also include a network interface unit 107 for allowing communicating with other computing units. The network interface unit 107 is also used for connecting to a LAN or WAN, or for connecting remotely to a LAN or WAN. Those of ordinary skill in the art will appreciate that the network interface unit 107 includes the necessary circuitry for such a connection, and is also constructed for use with the TCP/IP protocol, the particular network configuration of the LAN or WAN it is connecting to, and a particular type of coupling medium. The client computer may also be equipped with a network interface unit 107 capable of connecting to the Internet through a point to point protocol ("PPP") connection or a SLIP connection as known to those skilled in the art.

The ROM 103, RAM 104, network interface unit 107, video display adapter 109, and input/output interface 105 are all connected to the central processing unit 102 via a bus 101. Other peripherals may also be connected to the central processing unit 102 in a similar manner. These components act in concert to implement the methods of the software application 110, described in further detail below.

In another embodiment, the present invention can be implemented on a network of computing devices that includes a plurality of clients 201 and servers 202. FIG. 2 is a portion of an exemplary network 205 on which the invention is alternatively implemented. Although not necessary to realize the advantages of the present invention, network 205 can be part of a wide area network (WAN) in which different geographical locations are interconnected, either by high-speed data lines or by radio links, perhaps interconnecting hundreds of client computers 201 at widely disparate locations. Although the block diagram of FIG. 2 only depicts three servers 205, a network, and three clients 201, it will be understood by those of ordinary skill in the art that the present invention is also useable on other types of network configurations, such as a token ring or star configuration having a different number of clients 201 and servers 202. A web browser may then be used on the workstation to access the server. This type of architecture is well known in the art of network computing.

As shown in FIG. 1, the computer 100, which may be the configuration of the client 201 or server 202, include at least one hard drive on which are stored a plurality of files, including data and applications that can be accessed by the computing devices 201 and 202. As mentioned earlier, the clients 201 may access the servers 202 through network 205. The network may include a local area network, wide area network, or the Internet.

FIG. 2 shows a network environment using a client/server architecture where the database of the present invention resides on a storage device accessible by a retrieval engine in the server 202. Although FIG. 2 shows one configuration of a computing environment that may be used to implement the methods of the present invention, one skilled in the art would recognize that the present invention may be operated in a stand-alone environment as shown in FIG. 1.

A user at one of the client computers 201 may enter their desired database query that is sent over the network 205 to the retrieval engine resident on one of the servers 202. The retrieval engine retrieves the requested information from the database and the server returns the information to the client 201. Typically, the information that is returned is stored in a file on a hard disk on the client 201. Once all the necessary queries have been performed, the user may manipulate the retrieved data files with any application, such as a word processor. The method for manipulating the data in accordance with one embodiment of the present invention is described in more detail below.

(2) Databases

The database 122 as shown in FIG. 1 contains a plurality of journals related to a specific topic area. The database includes several fields for each article including authors, source, institution, title, keywords, abstract, etc. Articles can be retrieved using keywords and these keywords are arranged in a tree structure. Articles and their respective fields can be downloaded and further manipulated with adequate software for analysis. The example used in this patent illustrates a database for use in medicine and biology provided by the National Library of Medicine, known as MEDLINE, and is available through the Internet. However, the present invention may also use a database in other areas such as astronomy, computing, chemistry, engineering, physics, sociology, art, philosophy, etc.

TABLE 1 is an example of an article download from the MEDLINE database showing some of the above-described fields. 

    TABLE 1
    <1>
    Authors
        Saitta A M. Soper P D. Wasserman E. Klein M L.
    Institution
        Center for Molecular Modeling, Department of Chemistry, University
        of Pennsylvania, Philadelphia 19104-6202, USA.
    Title
        Influence of a knot on the strength of a polymer strand.
    Source
        Nature. 399(6731):46-8, 1999 May 6.
    MeSH Subject Headings
        *Alkanes/ch [Chemistry]
        Carbon/ch [Chemistry]
        Computer Simulation
        Models, Chemical
        Molecular Structure
        *Polymers/ch [Chemistry]
        Structure-Activity Relationship
        Support, U.S. Gov't, Non-P.H.S.
        Temperature
    Registry Numbers
        0 (Alkanes). 0 (Polymers). 124-18-5 (decane). 7440-44-0 (Carbon).
        . . .
        . . .


(3) User

It is preferred that the user of the system and method of the present invention be knowledgeable in the subjects being investigated. In order for the user to be able to address a broad question such as "which are the keywords related to clinical neurology?" or an intermediate question such as "which are the keywords related to schizophrenia?", a pre-search to determine the keywords of the subject being investigated is desirable to increase the method's sensitivity and reliability. Furthermore, interpretation of the final data will be maximized if the user of the system and method of the present invention is knowledgeable in the investigated subject.

There may be a few instances where a user may not be knowledgeable in an area being investigated. Similarly, a user may be interested in investigating a simplistic topic, such as a information on a generally known subject such as aspirin, a pre-search or prior knowledge may not be needed. In both instances a user can still utilize the present invention.

(4) Software

Software that enables the user to arrange, tag, count, index and perform the statistical analysis is most desirable since the amount of data that can be handled can very often be overwhelming (hundreds of thousands of articles; millions of keywords). Nevertheless, for a small search (addressing a small topic or one with a small number of articles) a less software oriented method can be applied. The assistance of a word processor that can sort out keywords alphabetically would suffice to pre-arrange the keywords. If patience is in the nature of the user, each keyword will need to be counted by frequency (number of times that the word is encountered). Following counting, keywords should be indexed and classified according to a pre-determined keyword tree structure. An important limitation of an embodiment of the system without sophisticated software is that the user will not be able to tag each keyword in each article with its correspondent impact factor. Similarly, a multiplication of frequency by impact factor will not be possible either. Nor will it be possible to know the average impact factor/frequency of all keywords under study.

In order to speed up the process, allow the user to handle a great number of articles at a time and make it a more refined, the present invention is preferably implemented with a software program that carries out the steps of the methods described in more detail below.

Problem to be Resolved

Two questions can be addressed with the present invention:

1) What is what is happening in a specific topic?

2) What is at the forefront of that topic?

In other words, question number one addresses the status of a specific topic, while question number two identifies the most relevant areas. Derived from these two questions are other sub-questions that can be addressed by the use of the methods of the present invention. In the biomedical example used in this disclosure, the following would be some examples of sub-questions that could be addressed:

What are the most frequently used cells?

What are the most frequently used organs?

Which are the most relevant molecules?

Which are the most relevant biochemical events?

Which are the most relevant genetic events?

Which are the most relevant physiological events?

Which are the top pathological processes?

Which are the most relevant diagnostic techniques?

Which are the most attractive therapeutic approaches?

Which are the most relevant pharmacological compounds?

FIG. 3 is an illustrative routine 300 for the database analysis that determines the status of a specific topic and the forefront of that topic. The process starts at step 301 where the user retrieves the resources of the search. In summary, and as described in more detail below, this step includes the identification of keywords related to the user selected topic, selecting journals related to the topic, and combining the system. Next, as shown in step 303, the process continues where the system indexes the selected keywords. This step involves arranging the data of prior indexing, indexing the current keywords, and into a database keyword tree structure. Next, at a step 305, the process continues where the system runs a statistical analysis according to the set of sub-questions being addressed.

Referring now to FIG. 4, the retrieval step 301 of FIG. 3 is now shown in more detail. The retrieval routine allows a user of the system to identify keywords of a specific topic. The retrieval routine is also referred to as the pre-search process. This process allows a user to identify the keywords of a topic by using the top journals in the topic of question. The retrieval routine allows a user to find the most relevant keywords of a broad or intermediate subject in which the user might not have a great deal of expertise.

The retrieval routine begins at block 401 where the user identifies the top specialty journals. Here the computing device 100 shown in FIG. 1, houses a database 122 which contains and utilizes all journals or a set of journals, e.g. the top 10 journal according to their impact factor, selected from the JCR.RTM. list under the desired topic. Here, the database of journals 122 will be arranged such that the heading of the JCR.RTM. can be searched by the user by entering the desired topic in a database search. In the following illustrative example, the desired topic of clinical neurology will be used. TABLE 2 is an example of the journals associated with the heading of clinical neurology in the JCR.RTM.. TABLE 2 lists the JCR.RTM. top ten journals, ordered by JCR's impact factor. 

     TABLE 2
                                                      Impact
      Rank    Journal Title                           Factor
        1     annals of neurology                        9.513
        2     brain pathology                            5.663
        3     brain                                      5.381
        4     journal of clinical psychopharmacology      5.094
        5     neurology                                  4.526
        6     stroke                                     4.323
        7     journal of neuropathology & exper. neurology      4.253
        8     archives of neurology                      3.779
        9     schizophrenia bulletin                     3.509
       10     pain                                       3.318


The process then proceeds to step 403 where the program of the present invention builds a database query based on the result of the first database search. In this example, the program utilizes the title of the top ten journals. The query built in this example can be in a format that conforms to a standard database program such as MEDLINE database software offered by the National Library of Medicine. Alternatively, other generic database programs such as Microsoft SQL.TM. can be used with methods of the present invention.

Although the present example only uses the top ten journals, step 403 can use all of the journal titles listed in the JCR.RTM.. TABLE 3 is an example the database query built from the list of discovered journal titles found in step 401. 

       TABLE 3
          1       annals of neurology
          2       brain pathology
          3       brain
          4       journal of clinical psychopharmacology
          5       neurology
          6       stroke
          7       journal of neuropathology & experimental neurology
          8       archives of neurology
          9       schizophrenia bulletin
         10       pain


Next, at a step 405, the program modifies the database query by placing the Boolean value,"OR," between each entry. Thus, the search string may be:

11 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10

The process then proceeds to a step 407 where the program enters the keywords to narrow the search query. As a matter of background, journals such as the JCR.RTM. include editorials, news, comments, news, letters, clinical conference notes, interview summaries, and reviews in addition to the main feature articles (original articles). Since the purpose of the present invention is to obtain data to address what is happening and what is emerging or receding in a specific field, the non-original articles should be removed from the database query. Thus, data obtained from this database query will not be tainted with articles related to non-scientific studies. TABLE 4 is an illustrative example of some of the terms that limit the search query. 

               TABLE 4
                  12               editorial
                  13               news
                  14               comments
                  15               news
                  16               letters
                  17               clinical conference
                  18               interviews
                  19               reviews
                  20               anonymous


The search terms used in step 407 should be entered in the database query using the Boolean term "OR". Thus, the following text can be added to the query.

21 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20

In addition, the keywords associated with the non-original articles must be negated. Thus, the search string should also include Boolean term "NOT" to eliminate the non-original articles. Thus, the following text can be added to the query.

22 11 not 21

Next, a step 409, the database query is limited by a user specified time frame. This will limit the search results to specific articles published in specified years so the final analysis of the program will reflect the most relevant trends in the subject matter of interest. In the example shown in TABLE 5, the database query is modified to only articles published within the timeframe of 1990 through 1998. 

               TABLE 5
                  23               limit 22 to yr=1998
                  24               limit 22 to yr=1997
                  25               limit 22 to yr=1996
                  26               limit 22 to yr=1995
                  27               limit 22 to yr=1994
                  28               limit 22 to yr=1993
                  29               limit 22 to yr=1992
                  30               limit 22 to yr=1991
                  31               limit 22 to yr=1990


The keyword thesaurus of any database changes to some extent in a yearly basis with the introduction of new keywords and elimination of outdated ones. In order to obtain all possible keywords during the aforementioned timeframe, the database queries 23 to 31 should be linked with the logical Boolean term "OR." Thus, the following text can be added to the query.

32 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31

As shown at a step 411, once the database query is constructed using steps 401-409, the query is processed in the database program so the database program appropriately retrieves the articles. In the download step 409, it is preferred that the database program be configured to download all of the articles revealed by the database query. It is preferred that the database program be configured such that database search result is formatted as shown in TABLE 6. TABLE 6 illustrates that the database search results include the name of the journal, also referred to as the source, and the subject headings from the MEDLINE, otherwise known as the MeSH subject headings. The MeSH subject headings are used as keywords in the subsequent steps of the present invention. The example of TABLE 6 shows two records of a search result having seven thousand articles. The database search results are then stored in a file on the computer hard drive.

There are two types of keywords in the MeSH subject headings available from MEDLINE. The first type of keywords represent the general subject matter of the article found in the journal. MEDLINE indicates these types of MeSH subject headings by marking them with an asterisk (*). The present invention utilizes these marked headings and refers to them as focused keywords. As shown in TABLE 6, when the articles are received in step 411, the database query result maintains a record of these focused keywords. The second type of MeSH subject heading indicates that an article covers a secondary or collateral issue. These secondary keywords are not marked with an asterisks in the search result file and are referred to as non-focused keywords. 

        TABLE 6
        <1>
        Source
            Nature. 399(6731):46-8, 1999 May 6.
        MeSH Subject Headings
            *Alkanes/ch [Chemistry]
            Carbon/ch [Chemistry]
            Computer Simulation
            *Coronary Artery Bypass/mt [Mortality]
            Models, Chemical
            Molecular Structure
            *Polymers/ch [Chemistry]
            Structure-Activity Relationship
            Support, U.S. Gov't, Non-P.H.S.
            . . .
            . . .
        <7000>
        Source
            Journal of Cardiac Surgery. 13(5):318-27, 1998 Sep-Oct.
        MeSH Subject Headings
            Adolescence
            Adult
            Aged
            Aged, 80 and over
            Child
            Child, Preschool
            Comparative Study
            Coronary Angiography
            Coronary Artery Bypass/mo [Mortality]
            *Coronary Artery Bypass/mt [Methods]
            Coronary Disease/ra [Radiography]
            *Coronary Disease/su [Surgery]
            Female
            Follow-Up Studies
            Human
            Infant
            Male
            Middle Age
            *Polymers
            Prospective Studies
            *Radial Artery/tr [Transplantation]
            Survival Rate
            Treatment Outcome


Next, the process continues at a step 413 where tags are set in the file containing the search results. As shown in TABLE 7, this step involves associating the corresponding impact factor of each journal title. This step can modify the data file in any format as long as there is an association between each journal and its corresponding impact factor. In one illustrative example, TABLE 7 shows the journal titles discovered in the database query of steps 401-411 as modified by the association step 413. 

            TABLE 7
            Journal Title                  Impact Factor
            abdominal imaging                  0.617
            academic emergency medicine           1.042
            academic medicine                  1.033
            . . .                              . . .
            . . .                              . . .
            journal of cardiac surgery           1.325
            . . .                              . . .
            . . .                              . . .
            nature                            27.368
            . . .                              . . .
            . . .                              . . .
            zoomorphology                      0.821
            zuchtungskunde                     0.218
            zuckerindustrie                    0.364


Next, as shown at a step 415, the process continues where each focused keyword(s) associated with each journal article is associated with the journal's corresponding impact factor. Thus, each article contains keywords tagged with an impact factor that is associated with the journal title. For example, if journal X is associated with an impact factor of 20.123, then all of the keywords in that journal will be associated with an impact factor of 20.123.

TABLE 8 is an example of a file containing the search results on the data file, where the keywords are tagged corresponding with their journal impact factor. In this example, the journal Nature and the Journal of Cardiac Surgery have impact factors of 27.368 and 1.325, respectively. Also shown in TABLE 8, step 415 only associates the focused keywords with the impact factors. As described above, the focused keywords are only those MeSH subject headings indicating that the heading is associated with the substance of the article. Thus, in the following example of TABLE 8, the file only associates the impact factor of 27.368 with the keywords: Alkanes, Coronary Artery Bypass, and Polymers. 

        TABLE 8
        <1>
        Source
              Nature. 399(6731):46-8, 1999 May 6.
        MeSH Subject Headings
             27.368 *Alkanes/ch [Chemistry]
                    Carbon/ch [Chemistry]
                    Computer Simulation
             27.368 *Coronary Artery Bypass/mt [Mortality]
                    Models, Chemical
                    Molecular Structure
             27.368 *Polymers/ch [Chemistry]
                    Structure-Activity Relationship
                    Support, U.S. Gov't, Non-P.H.S.
              . . .
              . . .
        <7000>
        Source
              Journal of Cardiac Surgery. 13(5):318-27, 1998 Sep-Oct.
        MeSH Subject Headings
                    Adolescence
                    Adult
                    Aged
                    Aged, 80 and over
                    Child
                    Child, Preschool
                    Comparative Study
                    Coronary Angiography
                    Coronary Artery Bypass/mo [Mortality]
              1.325 *Coronary Artery Bypass/mt [Methods]
                    Coronary Disease/ra [Radiography]
              1.325 *Coronary Disease/su [Surgery]
                    Female
                    Follow-Up Studies
                    Human
                    Infant
                    Male
                    Middle Age
              1.325 *Polymers
                    Prospective Studies
              1.325 *Radial Artery/tr [Transplantation]
                    Survival Rate
                    Treatment Outcome


As an alternative step, the process continues to a step 417 where the non-focused keywords are removed from the file containing the search results. One purpose of the pre-search process shown in FIG. 4 is to determine the most utilized and relevant keywords of the subject under investigation. Therefore, all the keywords lines that are not marked with an asterisk are removed from the data file. TABLE 9 illustrates an example of the data file of TABLE 8 with the non-focused keywords removed from the file. 

              TABLE 9
              27.368 *Alkanes/ch [Chemistry]
              27.368 *Coronary Artery Bypass/mt [Mortality]
              27.368 *Polymers/ch [Chemistry]
              . . .
              . . .
              1.325 *Coronary Artery Bypass/mt [Methods]
              1.325 *Coronary Disease/su [Surgery]
              1.325 *Polymers
              1.325 *Radial Artery/tr [Transplantation]


As another alternative step, the process may also include a step 419 where the sub-headings are removed from the data file. This step allows the program to further distinguish the keywords.

As a matter of background, some of the keywords in the MeSH subject heading will include subheadings. Subheadings are utilized by the cataloguers at the National Library of Medicine to refine the contents of a keyword. They are identifiable because they follow a "/" at the end of the keyword with an acronym of two letters followed by a bracketed explanation of the acronym. For instance, the example of TABLE 9 shows the keyword "*Coronary Artery Bypass/mt [Mortality]" and "*Coronary Artery Bypass/mt [Methods]." In the first keyword, the focused keyword Coronary Artery Bypass was used in an article that was explaining the mortality caused by this surgical procedure. In the second keyword, the term subheading shows that the article of coronary artery bypass is related to the study of surgical methodology. In some instances, the keywords do not contain any subheading such as the second occurrence of the term "polymers."

For the purpose of addressing the question on how to obtain the journal's keywords, it is irrelevant whether keywords contain or do not contain subheadings. To clarify this point the method removes all subheadings allowing process to easily compare all focused keywords. TABLE 10 is an example of the data file of TABLE 9 where the subheadings are removed. 

                      TABLE 10
                      27.368 *Alkanes
                      27.368 *Coronary Artery Bypass
                      27.368 *Polymers
                      . . .
                      . . .
                      1.325 *Coronary Artery Bypass
                      1.325 *Coronary Disease
                      1.325 *Polymers
                      1.325 *Radial Artery


At this point, the process continues to a step 421where two summation parameters are determined. First, the process sums the impact factors associated with identical keywords. Thus, the impact factor sum increases each time a keyword appears in the search. Second, the process sums the total number of times the keywords appeared in the articles. TABLE 11 is an example of a data file that shows the sum of the impact factors based on the keyword data of TABLE 10. Correspondingly, TABLE 12 is an example of a data file that shows the sums of the total number of times the keywords appeared in the articles based on the data of TABLE 10. It is preferred that the summed numbers of TABLE 11 and 12 are stored in a data file on the computer hard drive. 

                             TABLE 11
                  Impact factor summation output
                28.693           Coronary Artery Bypass
                28.693           Polymers
                27.368           Alkanes
                 1.325           Coronary Disease
                 1.325           Radial Artery
                  .              .
                  .              .
                  .              .


                         TABLE 12
                         Freguency Output
                2              Coronary Artery Bypass
                2              Polymers
                1              Alkanes
                1              Coronary Disease
                1              Radial Artery
                .              .
                .              .
                .              .


The process then continues to a step 423 where the present invention calculates additional variables. In this step, for each keyword, the computing device 100 then multiplies the impact factor totals (the figures of TABLE 11) by the frequency total (TABLE 12). An example of these totals are shown in TABLE 13. It is preferred that the values calculated in step 423, TABLE 13, are also stored in a data file on the computer hard drive. 

                             TABLE 13
         Impact factor summation .times. Freguency output
                57.386           Coronary Artery Bypass
                57.386           Polymers
                27.368           Alkanes
                 1.325           Coronary Disease
                 1.325           Radial Artery
                  .              .
                  .              .
                  .              .


The calculation of step 423 also involves building a fourth data file which includes an average impact factor for each keyword. Thus, the summed impact factor for each keyword, the value of TABLE 11, is divided by the frequency total (TABLE 12) for the corresponding keyword. The results of this part of the calculation is shown in TABLE 14. As shown in TABLE 11-14, it is preferred that all data file outputs are sorted in a descending order with respect to impact factor or frequency total. 

                             TABLE 14
             Impact factor summation/Frequency output
                27.368           Alkanes
                14.346           Coronary Artery Bypass
                14.346           Polymers
                 1.325           Coronary Disease
                 1.325           Radial Artery
                  .
                  .
                  .


The examples shown in TABLES 11-14 use small numbers and a limited number of articles for illustrative purposes only. In reality, a search containing thousands of articles from a search on the subject of clinical neurology would look like the example shown in TABLE 15. Thus, to simplify the subsequent calculation steps, the numeral values of the data stored in the data files, TABLE 11-14, may be truncated at the decimal level so each data file stores integer values. 

                             TABLE 15
                  Impact factor summation output
                  6,743           Alzheimer Disease
                  4,057           Parkinson Disease
                  3,468           Cerebellar Artery
                  3,078           Tremor
                  2,983           Central Nervous System
                  2,793           Reflex
                  2,702           Nervous System Neoplasms
                  2,643           Meningitis
                      .           .
                      .           .
                      .           .
                         Frequency output
                  2,491           Alzheimer Disease
                  1,363           Parkinson Disease
                  1,293           Central Nervous System
                  1,254           Nervous System Neoplasms
                  1,183           Cerebellar Artery
                  1,114           Reflex
                  1,062           Meningitis
                    990           Tremor
                      .           .
                      .           .
                      .           .
         Impact factor summation .times. Freguency output
             16,796,813           Alzheimer Disease
              5,529,691           Parkinson Disease
              4,102,644           Cerebellar Artery
              3,857,019           Central Nervous System
              3,388,308           Nervous System Neoplasms
              3,111,402           Reflex
              3,047,220           Tremor
              2,806,866           Meningitis
                      .           .
                      .           .
                      .           .
             Impact factor summation/Freguency output
                 3.1090           Tremor
                 2.9765           Parkinson Disease
                 2.9315           Cerebellar Artery
                 2.7069           Alzheimer Disease
                 2.5071           Reflex
                 2.4788           Meningitis
                 2.3070           Central Nervous System
                 2.1547           Nervous System Neoplasms
                      .           .
                      .           .
                      .           .


The impact factor summation output, as illustrated by TABLE 11, is specially designed to address the question of "what is at the forefront of a specialized area." This data is used for this question because it takes into account the addition of the relevance of the journal in which the keywords where published. In addition, the frequency output, as illustrated by TABLE 12, is best suited to determine what is happening in a specialized area. The data of TABLE 12, is used to make this determination since it only takes into account the number of times that a specific keyword was encountered regardless of journals' impact factor. The data of TABLE 13 may be used as an intermediate output between the data shown in TABLE 11 and 12. The data of TABLE 13 may be the preferred output depending on the type of search query. For example, the query may combine all of the database's keywords or the query may combine all of the journals that the database contains.

The data of TABLE 14 reveals the relative importance of each keyword. More specifically, a keyword with a high number would indicate that that keyword is an important keyword, or that it is very "hot," and that researchers investigating this keyword will likely be able to access very high impact factor journals. In contrast, a keyword with a low number would indicate that the keyword generates little interest and is related to a topic that is on the decline. Scientists investigating topics related to these keywords are in danger of not being read, publishing in low impact factor journals, and more importantly, possibly placing themselves in a difficult situation where it is difficult to raise financing for their research activities.

In another embodiment, the process may not tag the keywords with respect to the JCR.RTM. impact factors, as shown in steps 413 and 415. On the alternative embodiment, the process may tag keywords with a factor known as the immediacy-index. As provided by ISI, the immediacy-index represents the number of times current articles in a specific journal were cited during the year they were published. ISI publishes ranked lists of journals with respect to immediacy-index values. Thus, in the process of steps 413-423, the immediacy-index values are used in place of the impact factor. TABLE 16 is an example of an ISI publication showing the immediacy-index values with the corresponding journal title. 

     TABLE 16
       Rank     Journal Title                Immediacy-Index
         1      nature genetics                   6.892
         2      cell                              6.475
         3      nature                            6.322
         4      new england journal of medicine        5.726
         5      science                           4.722
         6      annual review of immunology        4.065
         .      .                                   .
         .      .                                   .
         .      .                                   .


In yet another embodiment, the process may use of the cited half-life of the journal, total number of citations, and the cumulative percent of cites to articles published in a set of years. Using the same steps 413-423 described above, these other factors may be substituted in place of the impact factor. As with the impact factor, these alternate factors, such as the cited half-life of the journal, are provided by ISI.

Returning to FIG. 4, the process then continues at a step 425 where the keywords are selected in accordance with the data shown in TABLES 11-14. Here, the computing device, or user, can use any one of the data sets as illustrated by TABLES 11-14 to select the most relevant keywords relating to the topic under investigation.

The procedure of selecting the keywords requires a preliminary explanation to distinguish between specifically related keywords, non-specific related keywords, and non-related keywords. Non-topic keywords do not belong in the subject being investigated. For example the keyword "Foot Dermatoses" has nothing to do with Neurological diseases, therefore, the terms should be discarded. Non-specific keywords are those that belong to the topic in question but are also relevant to other topics. For example, if the topic being investigated is neurological diseases, the keyword "radiography" will be considered as non-specific because although it is used in clinical neurology it is also used in many other medical specialties. Therefore, non-specific keywords should also be discarded. The use of the National Medical Library Thesaurus is an invaluable tool in determining which keywords may be ignored.

Since the number of discovered keywords can be excessive, the process can include a cut-off step. During the selection procedure there may be a point where most of the keywords in the lists are non-specific or non-related keywords. When this occurs, the process should identify the non-specific keywords and stop the selection step 425. The values from the calculation steps 421 and 423 select the specific keywords that relate exclusively to the topic and exclude keywords that are non-topic related and the ones that are non-specific related.

The user should not be concerned by the loss of non-specific keywords. Since there may be three to five focused keywords in every article, it is likely that one or more of the focused keywords will be located in the retrieval process 400. Thus, this article will retrieve a theoretically non-specific word, but in this case since it is attached to specific keywords, the aforementioned non-specific keyword will at the end be a specific-one and not be missed despite the fact that it was never included in the initial query. If a keyword is missed in the search, and is not included in the final query, it is very likely that no loss will occur. The other sets of keywords in the query will retrieve the missing keywords. Thus, it will be counted as if it would had been in the search in the first instance. Therefore, the present invention is flexible in that it does not necessarily require that all keywords to be included. For example, in a hypothetical search to identify articles on the topic of cardiovascular diseases, some of the specific keywords that would had been selected would had included *Coronary Artery Bypass, *Coronary Disease, *Radial Artery among hundreds of them. A non-specific keyword that is dismissed would had been *Polymers, since there are many different polymers and they are used in many different medical specialties. Nonetheless, the selection of any of the three specific keywords would suffice to identify the following article in a search for cardiovasculary diseases. Thus, specific-keywords also retrieve many other non-specific keywords. TABLE 17 is one example of a database query result with the non-specific keyword (polymers). 

    TABLE 17
    <7000>
    Source
        Journal of Cardiac Surgery. 13(5):318-27, 1988 Sep-Oct.
    MeSH Subject Headings
                    Adolescence
                    Adult
                    Aged
                    Aged, 80 and over
                    Child
                    Child, Preschool
                    Comparative Study
                    Coronary Angiography
                    Coronary Artery Bypass/mo [Mortality]
              1.325 *Coronary Artery Bypass/mt [Methods]
                    Coronary Disease/ra [Radiography]
              1.325 *Coronary Disease/su [Surgery]
                    Female
                    Follow-Up Studies
                    Human
                    Infant
                    Male
                    Middle Age
              1.325 *Polymers
                    Prospective Studies
              1.325 *Radial Artery/tr [Transplantation]
                    Survival Rate
                    Treatment Outcome


In another embodiment, the present invention may identify keywords related to a topic using a keyword tree structure belonging to a bibliographic database. This process is shown in the flow diagram of FIG. 5, another embodiment of a retrieval process 500. This process starts at block 501 where the user of the system identifies a topic's basic sub-questions that need to be addressed. For instance, the user may want to ask the following questions:

What are the most frequently used cells?

What are the most frequently used organs?

Which are the most relevant molecules?

Which are the most relevant biochemical events?

Other examples of the sub-questions are shown below in relation to the indexing process. Once the basic sub-questions have been identified, the user selects the related keywords from the sub-questions. For instance, if the question is, "what are the most frequently used cells," the user selects the word "cells" to use in the following methods.

Next, as shown at a step 503, the user selects the related keywords from a database keyword tree structure by using the keyword selected from the keywords of the sub-question.

As a matter of background, databases contain a keyword tree structure, sometimes referred to as a thesaurus. A keyword tree structure is typically arranged in alphabetical order and/or by subject headings. Subject headings are arranged in a keyword tree structure. For example, the keywords are arranged in three types of subject headings: major, medium and minor. Major headings are keywords that define a subject in a broad manner. Underneath of major headings keyword tree structure, are medium and minor headings. Minor headings are keywords that define a subject in a narrow manner. Medium headings exist between major and minor headings. They define a subject neither too broadly nor too narrowly. For example, the term central nervous system diseases could be considered as a major heading, where as cerebrovascular disorders may be considered as a medium heading and, cerebral embolism and thrombosis may be considered as a minor heading.

The process selects all of the keywords corresponding to a topic's basic subquestions that need to be addressed utilizing the major, medium and minor headings in the database's thesaurus. This can be achieved by either copying all of the keywords or by using an exploding command that many bibliographic databases posses. MEDLINE is a one known bibliographic databases that works well with the methods of the present invention.

The term "exploding" is a term of art used in association with the MEDLINE database. As known by a user of the database, when a term or heading is "exploded," the system selects that term or heading and all of the keywords that are underneath that term or heading in the tree structure. Therefore, a major heading that is specific for the subject can be exploded, or in other words, selected.

In one previous example, one of the sub-questions of interest addresses the top pathological processes relating to clinical neurology. In searching through the keyword tree structure the keywords or term central nervous system diseases may be found. The term central nervous system diseases will contain in a pyramidal structure, otherwise referred to as a tree structure, having many different keywords underneath it. Thus, the terms central nervous system diseases can be exploded.

This step should allow a user to select several hundred keywords belonging to the topic.

Next, the process continues to a step 505where the selection process is repeated for all other sub-questions related to the topic. Step 505 is carried out in a similar manner as the first selection step 503. Thus, the keywords found in the thesaurus that are associated with the remaining sub-questions are selected. In both steps 503 and 505, the selected, or exploded, keywords are stored in a data file on the computer hard drive.

Next, at a step 507 the non-focused keywords are removed. This step is carried out in a manner that is similar to the step 417 of FIG. 4. Next, the process continues to block 509where the keywords that are found in steps 503-507 are grouped into a database query, where the selected keywords are linked with the "OR" Boolean value.

The process then continues to block 511 where the database query built at a step 509 and is executed to recover a plurality of articles. This step is carried out in a similar manner as the download step 411 of FIG. 4.

In yet another embodiment of the invention, the retrieval step (block 301 of FIG. 3) can be carried out in by a method that combines the steps of the embodiment of FIGS. 4 and 5. In this process, the computing device processes the retrieval routines 400 and 500 and then combines all of the database queries obtained at steps 409 and 509. After the two database queries have been combined, the process then eliminates all duplicate keywords so that a final count of selected subject keywords can be determined. This combination of the two routines builds a database query that can be executed in a database program to retrieve a plurality of journals and articles. The articles are then downloaded in a manner that is similar to steps 411 and 511.

The utilization of the retrieval processes, the embodiments of FIGS. 4 and 5 and the combination thereof, allows a user of the system to choose the most important keywords. In the method of FIG. 5, the user will sometimes encounter words at the top of the lists that will be outside of the major branches of the keyword tree structure relating to the investigated subject matter. In this case, knowledge on the subject matter may be helpful. For example, a pre-search on the term diagnostic imaging may identify the term Nuclear Magnetic Resonance as the third most frequently used keyword. Although this keyword lies outside any major keyword tree structure branches relating to Diagnostic Imaging, it is an important keyword relating to the subject. TABLE 18 illustrates an example of a keyword tree structure related to the term, Nuclear Magnetic Resonance. 

    TABLE 18
    Analytical, Diagnostic and Therapeutic Techniques and Equipment
    Category
        Investigative Techniques
            Chemistry, Analytical
                Spectrum Analysis
                    Nuclear Magnetic Resonance
    Physical Sciences Category
        Physical Sciences
            Physics
                Nuclear Physics
                    Nuclear Magnetic Resonance


Upon completing the list of keywords for the database query, the user may inspect the list to check if the keywords obtained in the final search are adequate. If the user is knowledgeable in the subject, they may be able to identify if the search was successful. Extraneous keywords unrelated to the topic that are found among the top keywords in the final lists should warn the user that a selected keyword will interfere with the final results. At this stage of the process, the extraneous keywords should be removed. Usually, a higher concept in the thesaurus tree structure embraces a desired concept, albeit that some of its minor branches clearly refer to another subject. If both the higher concept and all the minor concepts are used, the final outcome will have a high background noise that will invalidate the search. Elimination of these keywords from the query will provide a more accurate result. Thus, it is important at all times to check if the objective of a clean search is being accomplished. An example of this type of problem can be best explained by the following: Otorhinolaryngology, the medical specialty of ear, nose and throat studies among many of its subjects, neoplasms (cancers). Refer to TABLE 19 for an example of a Head and Neck Neoplasms keyword tree structure. Otorhinolaryngologic Neoplasms are included within the tree structure of Head and Neck Neoplasms. Head and Neck Neoplasms need to be included in the search keyword section because it is a very frequent term being used in Otorhinolaryngologic journals. Besides Otorhinolaryngologic Neoplasms, Esophageal, Facial, Mouth, Thyroid, and Tracheal Neoplasms are also underneath the tree structure of Head and Neck Neoplasms, but they clearly belong to other searches involving Gastroenterology, General Surgery, Maxillo-Facial Surgery, Dermatology, Neurology, Endocrinology and Pulmonology. 

    TABLE 19
    Head and Neck Neoplasms
        Esophageal Neoplasms
        Facial Neoplasms
        Mouth Neoplasms
        Otorhinolaryngologic Neoplasms
            Ear Neoplasms
            Laryngeal Neoplasms
            Nose Neoplasms
                Nasal Polyps
                Paranasal Sinus Neoplasms
                    Maxillary Sinus Neoplasms
            Pharyngeal Neoplasms
            Hypopharyngeal Neoplasms
                Nasopharyngeal Neoplasms
                Oropharyngeal Neoplasms
                    Tonsillar Neoplasms
        Thyroid Neoplasms
        Tracheal Neoplasms


A solution to all these problems is to eliminate these keywords from the final search while maintaining the word Head and Neck Neoplasms together with the tree structure under Otorhinolaryngologic Neoplasms. Thus, the present invention can modify the database query by adding the following keywords to the string: Esophageal Neoplasms, Facial Neoplasms, Mouth Neoplasms, Thyroid Neoplasms, Tracheal Neoplasms. These new terms added to the query will link the terms with an "OR" Boolean value and negated by adding the "NOT" Boolean value to the database query.

As stated above, the pre-search process 400 of FIG. 4 may be beneficial when the user is not familiar with the subject matter of interest. Thus, the pre-search process 400 can be skipped and the embodiment utilizing the thesaurus as illustrated in FIG. 5 can be used alone. If the embodiment utilizing the entire thesaurus is used, the user will be able to understand the broadest question regarding the subject of biomedical feedback. In our example, the database MEDLINE would address the questions "what is happening in biomedical sciences, and what is at the forefront of biomedical sciences." Since the areas that are covered by the database MEDLINE cover the areas of Medicine, Biology, Biochemistry, Molecular Biology, Cell Biology and Pharmacology the aforementioned question can be addressed.

The following is an example of all the MEDLINE keywords:

1-(5-Isoquinolinesulfonyl)-2-methylpiperazine

1,2-Dimethylhydrazine

1,2-Dipalmitoylphosphatidylcholine

1,4-alpha-Glucan Branching Enzyme

. . .

. . .

Abattoirs

Abbreviated Injury Scale

Abbreviations

ABC Transporters

Abdomen

. . .

. . .

Zygosaccharomyces

Zygote

Zygote Intrafallopian Transfer

Zymomonas

Zymosan

In yet another embodiment of the retrieval process, the selection of journals can be used instead of the selection of keywords. This process can be used separately from the keyword retrieval methods describe above, or the journal selection process can be combined with the keyword retrieval methods. The embodiments associated with the journal selection process include five different methods. In one embodiment, the journal selection includes a simple method of selecting all of the specialty journals listed in the ISI publications. In another embodiment, the journal selection can include the selection of specialty journals with the highest impact factor. In yet another embodiment, only non-specialty journals can be selected for the retrieval process. In other methods, the selection of non-specialty journals can be combined with the selection of specialty journals. Alternatively, the method can select all of the journals in the database.

These methods use the journal impact factor published by ISI lists. As described above, ISI publishes an extensive list of journals from multiple disciplines in a yearly basis in Journal Citation Reports.RTM.. The list of journals are ranked by an impact factor which corresponds to prestige. The impact factor is calculated by dividing the number of citations to articles published in the two previous years by the total number of articles published in that journal during the same two years. TABLE 20 is an example of the journals associated with the term Clinical Neurology listed in a ISI publication from 1997, also listing their corresponding impact factor. 

    TABLE 20
                                                       Impact
     Rank   Journal Title                              Factor
       1    annals of neurology                         9.513
       2    brain pathology                             5.663
       3    brain                                       5.381
       4    journal of clinical psychopharmacology      5.094
       5    neurology                                   4.526
       .    .                                             .
       .    .                                             .
       .    .                                             .
      102   aktuelle neurologie                         0.240
      103   nervenheilkunde                             0.226
      104   psychiatry and clinical neurosciences       0.191
      105   neurosurgical review                        0.161
      106   zhurnal nevropatologii i psikhhiatrii imrni korsakova   0.128


The higher the impact factor that a journal possesses the more prestige it has. It is also likely that the journal also possesses more readership. On the other hand, a journal with a small impact factor will suggest that the quality of its articles is probably average and therefore its readership is small. Journals with a high impact factor cover a wide-spread series of subjects. In contrast, journals with small impact factor tend to be very focused on a subject. It is a well established phenomenon that scientists/writers gravitate towards publications with the highest impact factors.

FIG. 6A is a flow diagram depicting one embodiment of the retrieval process of the present invention using the selection of specialty journals found in the ISI publication. The process starts at a step 601 where the user selects a heading under the ISI. Next, at a step 602, the user or computing device selects the top journals that are listed under the selected ISI heading. Thus, in step 602, the journals with the highest impact factor are selected. TABLE 21 illustrates the results of the ranked list (top 10) chosen for the heading of Clinical Neurology in the ISI list published in 1997.

In this embodiment, the method only considers the journal titles under the ISI heading selected by the user. The journal titles under the selected heading are considered to be the specialty journals. Journals titles that are not listed under the ISI heading are considered to be non-specialty journals.

In the illustrative example shown in TABLE 20, the heading of Clinical Neurology is one that could be selected by the user as a topic of choice. All journal titles listed under the heading of Clinical Neurology are considered to be "specialty journals," and all other journal listings in the ISI publication are considered to be "non-specialty journals." The example of TABLE 21 is a list of specialty journals sorted by the priority of their corresponding impact factor. The results shown in FIG. 6A are most suitable for a pre-search to find out the most relevant keywords used in the subject under study. 

      TABLE 21
          1       annals of neurology
          2       brain pathology
          3       brain
          4       journal of clinical psychopharmacology
          5       neurology
          6       stroke
          7       journal of neuropathology & experimental neurology
          8       archives of neurology
          9       schizophrenia bulletin
         10       pain


Although results shown in FIG. 6A use the top ten journal titles, any numbers of journals can be selected by this ranking. For instance, if an analysis needs more journal titles the present invention can be also modified so that the selection step 602 involves selecting all specialty journals associated with the selected heading, instead of only selecting the top ten.

In another embodiment, the journal selection process may include the selection of non-specialty journals. In this embodiment, non-specialty journals with an impact factor higher than the specialty journal with the highest impact factor are selected. FIG. 6B illustrates the selection process using the non-specialty journals 600'. This follows the reasoning that most if not all of the major advances in a field are not published in the journal's field but in journals with the highest impact factor the present selection of journals addresses the phenomenon by which scientists gravitate towards publishing in the most prestigious journals.

The process starts at a step 601' where the user or the computing device selects a heading under the ISI. The step 601' is carried out in the same manner as the step 601in FIG. 6A. The process then continues at a step 602' where the computing device loads the titles of the non-specialty journal with the highest impact factors. TABLE 22 is an example of a list of non-specialty journals under the heading of Science in the ISI list. Although, the heading of Science is used in this example, other generic headings can be used to collect the title names of non-specialty journals.

The process then continues to a step 605' where the computing device loads a list of specialty journal titles. The step 605' is carried out in the same manner as step 602 of FIG. 6A.

The process then continues at a step 607' where the selection step takes place. In this embodiment, only the non-specialty journals with an impact factor greater than the first related specialty journal are selected. In the example of the heading of Clinical Neurology and according to the list previously shown, the process should select all the journals with a higher impact factor than the journal titled Annals of Neurology. TABLE 22 is a list of the top non-specialty journals in the Science ISI list. Using the method of step 607', all of the non-specialty journals shown in TABLE 22 would be selected because they have an impact factor higher than the highest specialty journal, The Annals of Neurology, which has an impact factor of 9.513. 

     TABLE 22
       Rank     Journal Title                  Impact Factor
         1      annual review of biochemistry        40.782
         2      nature genetics                     38.854
         3      annual review of immunology         37.796
         4      cell                                37.297
         5      nature medicine                     28.114
         6      New England journal of medicine        27.766
         7      nature                              27.368
         8      science                             24.676
         9      endocrinology reviews               23.017
        10      annual review of neurosciences        21.952
         .      .                                     .
         .      .                                     .
         .      .                                     .


In the embodiment of FIG. 6B, the method utilizing the impact factor of non-specialty journals, is most suitable to analyze the question of what is at the forefront of a research topic?

In yet another embodiment, the present invention combines the two different embodiments illustrated in FIGS. 6A-6B. In this embodiment, the method of selecting the specialty journals with the highest impact factor is combined with the method utilizing the non-specialty journals with an impact factor higher than the specialty journal with the highest impact factor. Thus, steps 602 and 602' are both processed to find a list of relevant journals.

In some instances, the number of non-specialty journals is rather small. For instance if there are one hundred publications per year, a statistical and trends analysis will be difficult to accurately evaluate. If this occurs, the user should increase the number of publications used in the analysis. In such a case, the user might want to add titles of the top specialty journals to the list of non-specialty journals. In the example involving the heading of Clinical Neurology, and the specialty journal of Annals of Neurology, the journal list would appear as: 

      TABLE 23
        Rank      Journal Title                Impact Factor
          1       annual review of biochemistry        40.782
          2       nature genetics                   38.854
          3       annual review of immunology        37.796
          4       cell                              37.297
          5       nature medicine                   28.114
          6       New England journal of medicine        27.766
          7       nature                            27.368
          8       science                           24.676
          9       endocrinology reviews             23.017
         10       annual review of neurosciences        21.952
          .       .                                   .
          .       .                                   .
          .       .                                   .
          n       annals of neurology                9.513
          n       brain pathology                    5.663
          .       .                                   .
          .       .                                   .
          .       .                                   .


In addition, the method can include a second specialty journal or a third, if the number of publications remains small. In this example, the Journal of Brain Pathology would be added to the journal selection list. If needed, the user or computer program can repeat the same procedure until the system finds that the number of publications/year is acceptable for statistical analysis. An acceptable number of publications for statistical analysis occurs when three hundred or more publications per year are obtained. TABLE 24 is one example of a list adding the specialty journal titles. 

      TABLE 24
        Rank      Journal Title                Impact Factor
          1       annual review of biochemistry        40.782
          2       nature genetics                   38.854
          3       annual review of immunology        37.796
          4       cell                              37.297
          5       nature medicine                   28.114
          6       new england journal of medicine        27.766
          7       nature                            27.368
          8       science                           24.676
          9       endocrinology reviews             23.017
         10       annual review of neurosciences        21.952
          .       .                                   .
          .       .                                   .
          .       .                                   .
          n       annals of neurology                9.513
        n + 1     brain pathology                    5.663
          .       .                                   .
          .       .                                   .
          .       .                                   .


In yet another embodiment, the system and method described above is modified to use additional non-specialty journal titles in the search. Thus, the step 607' of FIG. 6B could be modified to select all non-specialty journals with an impact factor higher than the second or third highest impact factor of the specialty journals. Again, this will allow the process to increase the number of journal titles if the process does not produce enough articles as the process is described in step 607' of FIG. 6B. TABLE 25 is one example of a journal selection list using this embodiment. This embodiment is more suitable to find answer the question of what is at the forefront of a research topic? 

      TABLE 25
        Rank      Journal Title                Impact Factor
          1       annual review of biochemistry        40.782
          2       nature genetics                   38.854
          3       annual review of immunology        37.796
          4       cell                              37.297
          5       nature medicine                   28.114
          6       new england journal of medicine        27.766
          7       nature                            27.368
          8       science                           24.676
          9       endocrinology reviews             23.017
         10       annual review of neurosciences        21.952
          .       .                                   .
          .       .                                   .
          .       .                                   .
          n       annals of neurology                9.513
          .       .                                   .
          .       .                                   .
          .       .                                   .
          m       brain pathology                    5.663
          .       .                                   .
          .       .                                   .
          .       .                                   .


In yet another embodiment, the process of selecting journals can involve selecting the specialty journals together with the non-specialty journals with an impact factor higher than the best specialty journal with the highest impact factor. This is a combination of the methods of FIGS. 6A and 6B. This embodiment is most suitable to find out what is at the forefront of a research topic? 

    TABLE 26
     Rank   Journal Title                         Impact Factor
       1    annual review of biochemistry              40.782
       2    nature genetics                            38.854
       3    annual review of immunology                37.796
       4    cell                                       37.297
       5    nature medicine                            28.114
       6    new england journal of medicine            27.766
       7    nature                                     27.368
       8    science                                    24.676
       9    endocrinology reviews                      23.017
      10    annual review of neurosciences             21.952
       .    .                                            .
       .    .                                            .
       .    .                                            .
       n    annals of neurology                         9.513
     n + 1  brain pathology                             5.663
     n + 2  brain                                       5.381
     n + 3  journal of clinical psychopharmacology       5.094
     n + 4  neurology                                   4.526
     n + 5  stroke                                      4.323
     n + 6  journal of neuropathology & exper. neurology       4.253
     n + 7  archives of neurology                       3.779
     n + 8  schizophrenia bulletin                      3.509
     n + 9  pain                                        3.318
       .    .                                            .
       .    .                                            .
       .    .                                            .
       m    aktuelle neurologie                         0.240
     m + 1  nervenheilkunde                             0.226
     m + 2  psychiatry and clinical neurosciences       0.191
     m + 3  neurosurgical review                        0.161
     m + 4  zhurnal nevropatologii i psikhhiatrii imrni       0.128
            korsav.


In yet another embodiment, the selection process can be as simple as selecting all of the journals in the database. This process produces the most journal articles and just involves, selecting all journal titles by entering an unrestricted database query.

This embodiment is most suitable to find out the question of what is happening in a topic of interest? It could also be used to address the question of what is at the forefront of that topic? If the computing device has an appropriate memory allocation, this embodiment can be used to address broad questions regarding biomedical research. 

      TABLE 27
        Rank      Journal Title                Impact Factor
          1       annual review of biochemistry        40.782
          2       nature genetics                   38.854
          3       annual review of immunology        37.796
          4       cell                              37.297
          5       nature medicine                   28.114
          6       new england journal of medicine        27.766
          7       nature                            27.368
          8       science                           24.676
          9       endocrinology reviews             23.017
         10       annual review of neurosciences        21.952
          .       .                                   .
          .       .                                   .
          .       .                                   .
          n       revista de biologia tropical         0.132
          .       .                                   .
          .       .                                   .
          .       .                                   .


As mentioned above, the embodiments of the retrieving step (step 301 of FIG. 3), involving both the keyword pre-search and the journal selection search, can be combined. Thus, in yet another embodiment, a method can combine the various retrieval methods described above to build a list of keywords and journal titles. FIG. 7 is a flow diagram showing this combination retrieval process 700. Thus, once the keywords for the specific search have been gathered and a selection of a set of journals to best address one of the two questions has been established the user is ready to proceed with the first step 701 where the keywords are entered into the database query.

The following example involves a user selection of the topic of clinical neurology. Following a pre-search using the systems described above, a list of nine hundred keywords that are considered specific for the subject under study are selected. TABLE 28 is an example of these keywords. 

                  TABLE 28
                  Abducens Nerve
                  Accessory Nerve
                  Acoustic Nerve
                  Acoustic Nerve Diseases
                  Acrodynia
                  Action Potentials
                  Adie's Syndrome
                  .
                  .
                  .
                  Waterhouse-Friderichsen Syndrome
                  Werdnig-Hoffmann Disease
                  Wernicke's Encephalopathy
                  West Nile Fever
                  Williams Syndrome
                  Zellweger Syndrome


In some cases the user might want to avoid obtaining secondary articles that might obscure the final interpretation of the data. As explained above, two kinds of keywords exist in some databases: focused and non-focused. By selectively choosing focused keywords, the user is making sure that all the articles that will finally be downloaded will be the ones in which the user is really interested. In the MEDLINE database application, in order to do that, the user needs to input an asterisk in front of the keywords requested and a backwards stroke /. TABLE 29 is an example of this query. in the step 701, the keywords are entered into the database query linked by the "or" Boolean value. 

                  TABLE 29
                  *Abducens Nerve/
                  *Accessory Nerve/
                  *Acoustic Nerve/
                  *Acoustic Nerve Diseases/
                  *Acrodynia/
                  *Action Potentials/
                  *Adie's Syndrome/
                  .
                  .
                  .
                  *Waterhouse-Friderichsen Syndrome/
                  *Werdnig-Hoffmann Disease/
                  *Wernicke's Encephalopathy/
                  *West Nile Fever/
                  *Williams Syndrome/
                  *Zellweger Syndrome/


In some instances the user might be interested in using subheadings to address a more refined search. For instance the user might want to know more about the origins of these diseases and their physiological and pathological events associated with them. To accomplish this purpose, the user might be interested in abnormalities, anatomy and histology, cerebrospinal fluid, congenital, embryology, etiology, genetics, innervation, metabolism, pathology, physiology, physiopathology of all the keywords selected regarding Clinical Neurology rather than searching all the keywords related to the subject. The user would place a backwards stroke / at the end of all the keywords selected and the two letter code of the subheadings chosen by the user. According to the MEDLINE application, the following are the two letter codes for the subheadings previously mentioned: 
                 ab              abnormalities
                 ah              anatomy and histology
                 cf              cerebrospinal fluid
                 cn              congenital
                 em              embryology
                 et              etiology
                 ge              genetics
                 ir              innervation
                 me              metabolism
                 pa              pathology
                 ph              physiology
                 pp              physiopathology


Therefore, the keywords exposed above would look like the following:

*Abducens Nerve/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Accessory Nerve/ab, ah, cf, cn, em, et,ge, ir, me, pa, ph, pp

*Acoustic Nerve/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Acoustic Nerve Diseases/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Acrodyinia/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Action Potentials/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Adie's Syndrome/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

* . . .

* . . .

*Waterhouse-Friderichsen Syndrome/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

Werdnig-Hoffmann Disease/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Wernicke's Encephalopathy/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*West Nile Fever/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Williams Syndrome/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

*Zellweger Syndrome/ab, ah, cf, cn, em, et, ge, ir, me, pa, ph, pp

Referring again to FIG. 7, the process continues to a step 703 where a set of journals are entered into the database query. For example, if a user has decided to use the method of FIG. 6B to select the journals for the query, all the specialty journals together with the non-specialty journals with an impact factor higher than the best specialty journal with the highest impact factor, TABLE 30 would represent an example of a list of journals involved. In step 703, the journal titles are entered into the database query linked by the "OR" Boolean value. 

    TABLE 30
     Rank   Journal Title                         Impact Factor
       1    annual review of biochemistry              40.782
       2    nature genetics                            38.854
       3    annual review of immunology                37.796
       4    cell                                       37.297
       5    nature medicine                            28.114
       6    new england journal of medicine            27.766
       7    nature                                     27.368
       8    science                                    24.676
       9    endocrinology reviews                      23.017
      10    annual review of neurosciences             21.952
       .    .                                            .
       .    .                                            .
       .    .                                            .
       n    annals of neurology                         9.513
     n + 1  brain pathology                             5.663
     n + 2  brain                                       5.381
     n + 3  journal of clinical psychopharmacology       5.094
     n + 4  neurology                                   4.526
     n + 5  stroke                                      4.323
     n + 6  journal of neuropathology & exper. neurology       4.253
     n + 7  archives of neurology                       3.779
     n + 8  schizophrenia bulletin                      3.509
     n + 9  pain                                        3.318
       .    .                                            .
       .    .                                            .
       .    .                                            .


The process then continues to step 705 where the non-original scientific publications are filtered. As described above, this process filters out the general editorials and news articles. Thus, the keywords of TABLE 31 can be entered in the database query. 

                         TABLE 31
                         editorial
                         news
                         letters
                         comments
                         news
                         clinical conference
                         interviews
                         reviews
                         discussions
                         anonymous


In step 705, the filter terms are linked by the "OR" Boolean value and negated by the "NOT" Boolean value, as described above.

Next, as shown in step 707, the query is further limited by years or group of years so that statistical trends can be studied. If the query is searching for articles between 1990 and 1998 the query will use the limits in the database query. An example of this limitation is shown in TABLE 32. 

                             TABLE 32
                        limit to yr = 1998
                        limit to yr = 1997
                        limit to yr = 1996
                        limit to yr = 1995
                        limit to yr = 1994
                        limit to yr = 1993
                        limit to yr = 1992
                        limit to yr = 1991
                        limit to yr = 1990


Next, the process proceeds to block 709 where the articles are downloaded. This step is carried out in the same manner as the download step 411 of FIG. 4.

An alternative method to use this system is to process two different combinations of keywords and journals. One combination would combine all nine hundred keywords for the term Clinical Neurology and all the non-specialty journals with an impact factor higher than the best specialty journal with the highest impact factor. The second combination would combine all the specialty journals with no keywords. The latter would obtain all the articles published in those journals. Since these journals are directly related to the area under investigation there is no need to contrast them against the selected keywords. Thus, the search becomes more refined and accurate, and most probably will contain a few more articles that otherwise would had been missed. Further steps of removing non-related scientific publications and limiting the query to years prior to downloading are necessary.

Indexing: Arrangement of the data prior to indexing

Now referring again to FIG. 3, the indexing step 303 is described in more detail. In one embodiment of the indexing step 303, the process must arrange the data prior to indexing. An arrangement process 800 is depicted in the flow diagram of FIG. 8. The process starts at a step 801 where the process identifies the nonfocused keywords.

As described above, there are two kinds of keywords. One type of keyword represents the gist of the article and are tagged with an asterisk. These are focused keywords. The second type of keyword represents secondary or collateral issues. These are known as the non-focused keywords.

Upon opening with a word processor of a downloaded set with, for example, seven thousand articles, the user will be confronted with file similar to TABLE 33. 

    TABLE 33
    <1>
    Source
        New England Journal of Medicine. 338(26): 1888-95, 1998 Jun 25.
    MeSH Subject Headings
        Acetylcholine
        Calcium/ph [Physiology]
        *Epilepsy
        *Epilepsy/dt [Drug Therapy]
        Nerve Tissue Proteins/me [Metabolism]
        .
        .
        .
        *Vision
    .
    .
    .
    <7000>
    Source
        Nature Medicine. 4(11): 1269-75, 1998 Nov.
    MeSH Subject Headings
        Brain
        *Brain
        *Brain/im [Immunology]
        *Cerebral Hemorrhage/co [Complications]
        .
        .
        .
        Vitamin A/bi [Biosynthesis]


Since the question to be addressed requires a clear cut answer, it is best to remove non-focused keywords that could jeopardize the answer being sought. Knowing that non-focused keywords are the vast majority of the keywords in each article, it is likely that if taken into the statistical analysis they may obscure the important issues to be determined. Therefore, non-focused words should be removed. Given that focused versus non-focused keywords differ by the tagged * a letter like X can be tagged to all words containing an *, thus all keywords with * will look like *X. TABLE 34 is an example of this process. 

    TABLE 34
    <1>
    Source
        New England Journal of Medicine. 338(26): 1888-95, 1998 Jun 25.
    MeSH Subject Headings
        Acetylcholine
        Calcium/ph [Physiology]
        *XEpilepsy
        *XEpilepsy/dt [Drug Therapy]
        Nerve Tissue Proteins/me [Metabolism]
        .
        .
        .
        *XVision
        .
        .
        .
    <7000>
    Source
        Nature Medicine. 4(11): 1269-75, 1998 Nov.
    MeSH Subject Headings
        Brain
        *XBrain
        *XBrain/im [Immunology]
        *XCerebral Hemorrhage/co [Complications]
        .
        .
        .
        Vitamin A/bi [Biosynthesis]


Next, in a step 803, the process calculates a correction factor. This step is carried out by recording how many changes were made by tagging the focused keywords with an X. This will indicate the user the total number of focused keywords per year or group of years. For instance if a search has been downloaded by years the list showing the number of focused keywords might be like the following: 

          TABLE 35
            1998         85,345         focused keywords
            1997         83,759                "
            1996         82,874                "
            1995         80,996                "
            1994         77,721                "
            1993         73,858                "
            1992         68,493                "
            1991         64,832                "
            1990         57,732                "


Since the number of keywords might vary from different years due to the fact that some journals increase with time the number of articles they carry, some journals are new or some journals change their names, a correction factor needs to be determined. This correction factor will compare every year or group of years in an equal manner. If all the conditions remain equal, any keyword that is on the rise will be able to displace other ones. In such a case, a true comparative analysis between different years will be made possible. Consequently, the correction factor is preferably determined by taking the year or group of years with the highest number of focused keywords that had been tagged with *X and divide it by each one of the other years or group of years. Each year will have a correction factor. The following represent the example years, with their correspondent correction factor. Following this procedure, the letter X can be removed since it will be no longer needed.

Alternatively, rather than counting the focused keywords, the process can count the number of articles obtained. 

           TABLE 36
             1998          1.00        Correction Factor
             1997          1.02                "
             1996          1.03                "
             1995          1.05                "
             1994          1.10                "
             1993          1.16                "
             1992          1.25                "
             1991          1.32                "
             1990          1.48                "


The process then continues to a step 805 where the process calculates the sum of the impact factor and the frequency total. This summation step 805 is carried out in the same manner as the summation step 421 of FIG. 4. Next, the process continues to a step 807 where the process calculates the multiplying factors. This calculation step 807 is carried out in the same manner as the calculation step 423 of FIG. 4. As also shown above, TABLE 37 is an example of the results of the calculation steps 805 and 807. 

                             TABLE 37
                  Impact factor summation output
                 6,743           Alzheimer Disease
                 4,057           Parkinson Disease
                 3,468           Cerebellar Artery
                 3,078           Tremor
                 2,983           Central Nervous System
                 2,793           Reflex
                 2,702           Nervous System Neoplasms
                 2,643           Meningitis
                     .           .
                     .           .
                     .           .
                         Freguency output
                 2,491           Alzheimer Disease
                 1,363           Parkinson Disease
                 1,293           Central Nervous System
                 1,254           Nervous System Neoplasms
                 1,183           Cerebellar Artery
                 1,114           Reflex
                 1,062           Meningitis
                   990           Tremor
                     .           .
                     .           .
                     .           .
         Impact factor summation .times. Frequency output
            16,796,813           Alzheimer Disease
             5,529,691           Parkinson Disease
             4,102,644           Cerebellar Artery
             3,857,019           Central Nervous System
             3,388,308           Nervous System Neoplasms
             3,111,402           Reflex
             3,047,220           Tremor
             2,806,866           Meningitis
                     .           .
                     .           .
                     .           .
    Impact factor summation/Frequency output
                3.1090           Tremor
                2.9765           Parkinson Disease
                2.9315           Cerebwellar Artery
                2.7069           Alzheimer Disease
                2.5071           Reflex
                2.4788           Meningitis
                2.3070           Central Nervous System
                2.1547           Nervous System Neoplasms
                     .           .
                     .           .
                     .           .


                         TABLE 37
                  Impact factor summation output
                 6,743           Alzheimer Disease
                 4,057           Parkinson Disease
                 3,468           Cerebellar Artery
                 3,078           Tremor
                 2,983           Central Nervous System
                 2,793           Reflex
                 2,702           Nervous System Neoplasms
                 2,643           Meningitis
                     .           .
                     .           .
                     .           .
                         Freguency output
                 2,491           Alzheimer Disease
                 1,363           Parkinson Disease
                 1,293           Central Nervous System
                 1,254           Nervous System Neoplasms
                 1,183           Cerebellar Artery
                 1,114           Reflex
                 1,062           Meningitis
                   990           Tremor
                     .           .
                     .           .
                     .           .
         Impact factor summation .times. Frequency output
            16,796,813           Alzheimer Disease
             5,529,691           Parkinson Disease
             4,102,644           Cerebellar Artery
             3,857,019           Central Nervous System
             3,388,308           Nervous System Neoplasms
             3,111,402           Reflex
             3,047,220           Tremor
             2,806,866           Meningitis
                     .           .
                     .           .
                     .           .
    Impact factor summation/Frequency output
                3.1090           Tremor
                2.9765           Parkinson Disease
                2.9315           Cerebwellar Artery
                2.7069           Alzheimer Disease
                2.5071           Reflex
                2.4788           Meningitis
                2.3070           Central Nervous System
                2.1547           Nervous System Neoplasms
                     .           .
                     .           .
                     .           .
         Impact factor summation .times. Frequency output
            16,796,813           Alzheimer Disease
            5,529,691            Parkinson Disease
            4,102,644            Cerebellar Artery
            3,857,019            Central Nervous System
            3,388,308            Nervous System Neoplasms
            3,111,402            Reflex
            3,047,220            Tremor
            2,806,866            Meningitis
            . . .                . . .
            . . .                . . .
         Impact factor summation .times. Frequency output
            16,796,813           Alzheimer Disease
            5,529,691            Parkinson Disease
            4,102,644            Cerebellar Artery
            3,857,019            Central Nervous System
            3,388,308            Nervous System Neoplasms
            3,111,402            Reflex
            3,047,220            Tremor
            2,806,866            Meningitis
            . . .                . . .
            . . .                . . .


In order to be able to compare every year or groups of years on an equal basis a correction factor needs to be applied. This step is depicted at a step 809 of FIG. 8. In the following example of the application of the correction factor, the year being studied is 1994. To the keywords found in the 1994 search, a correction factor of 1.10 needs to be applied. After the correction factor has been applied to the data of TABLE 36, the outputs will look like the output shown in TABLE 38. 

                             TABLE 38
                  Impact factor summation output
               7,417             Alzheimer Disease
               4,462             Parkinson Disease
               3,814             Cerebellar Artery
               3,385             Tremor
               3,281             Central Nervous System
               3,072             Reflex
               2,972             Nervous System Neoplasms
               2,907             Meningitis
                 .               .
                 .               .
                 .               .
                         Frequency output
               2,740             Alzheimer Disease
               1,499             Parkinson Disease
               1,422             Central Nervous System
               1,379             Nervous System Neoplasms
               1,301             Cerebellar Artery
               1,225             Reflex
               1,168             Meningitis
               1,089             Tremor
                 .               .
                 .               .
                 .               .
            Impact factor summation x Frequency output
            20,322,580           Alzheimer Disease
             6,688,538           Parkinson Disease
             4,962,014           Cerebellar Artery
             4,665,582           Central Nervous System
             4,098,388           Nervous System Neoplasms
             3,763,200           Reflex
             3,686,265           Tremor
             3,395,376           Meningitis
                 .               .
                 .               .
                 .               .
             Impact factor summation/Frequency output
               3.1083            Tremor
               2.9766            Parkinson Disease
               2.9315            Cerebellar Artery
               2.7069            Alzheimer Disease
               2.5077            Reflex
               2.4888            Meningitis
               2.3073            Central Nervous System
               2.1555            Nervous System Neoplasms
                 .               .
                 .               .
                 .               .


Indexing

The aim of this system is to gather the information necessary to address the question of "what is happening in a specialized area?" and, what are the topics-items that are getting "hot" as well as the ones that are "cooling down?" As a consequence, the following are some of the sub-questions that can be addressed:

What are the most frequently used cells?

What are the most frequently used organs?

Which are the most relevant molecules?

Which are the most relevant biochemical events?

Which are the most relevant genetic events?

Which are the most relevant physiological events?

Which are the top pathological processes?

Which are the most relevant diagnostic techniques?

Which are the most attractive therapeutic approaches?

Which are the most relevant pharmacological compounds?

. . .

. . .

Referring now to FIG. 9, the indexing process 900 starts, at block 901, by separating the keywords from the sub-questions. For the purpose of demonstrating an illustrative example, the list of keywords in TABLE 39 relate to the sub-question of, "what are the most frequently used cells?" This search will be narrowed to a group of keywords in the time frame from 1996 till 1998. 

              TABLE 39
                601              Astrocytes
                 31              Blood Cells
                 95              Hair Cells
                423              Interneurons
                 53              Lymphocytes
                 82              Microglia
               1,084             Motor Neurons
                353              Neuroglia
               4,964             Neurons
                823              Neurons, Afferent
                 67              Neurons, Efferent
                268              Oligodendroglia
                257              Purkinje Cells
                274              Pyramidal Cells
                328              Retinal Ganglion Cells
                173              Schwann Cells
                 23              Stem Cells
                 17              T-Lymphocytes


Next, at a step 903, the process continues where it indexes keywords according to a database keyword tree structure. This process can be carried out before or after the separation step 901. Following the selection of one of the data sets of TABLE 38, either the 1st, 2nd, or 3rd set of data, the software executed on the computing device is used to input the data into a template file containing the keyword tree structure. If one of the sub-questions being addressed is, "what are the most frequently used cells used in Clinical Neurology," the process should use two files. The first file should contain the keywords and their respective numbers (impact factor, frequency or impact factor multiplied by frequency; outputs number 1, 2 or 3). TABLE 39 is an example representing the list of cells found in our search using output number 2. The second file is the template file containing the keyword tree structure. The following is an example on how neuroglia and neuron cells are arranged according to the National Library of Medicine thesaurus keyword tree structure used by MEDLINE. 

        TABLE 40
        Neuroglia
               Astrocytes
              Microglia
              Neuropil
                 Neuropil Threads
              Oligodendroglia
                 Myelin Sheath
              Schwann Cells
                 Myelin Sheath
                         Neurilemma
                         Ranvier's Nodes
               Neurons
              Dendrites
                 Neurites
              Growth Cones
              Interneurons
              Lewy Bodies
              Nerve Fibers
                    Adrenergic Fibers
                         Sympathetic Fibers, Postganglionic
                    Autonomic Fibers, Postganglionic
                         Parasympathetic Fibers, Postganglionic
                         Sympathetic Fibers, Postganglionic
                    Autonomic Fibers, Preganglionic
                    Axons
                         Neurites
                         Presynaptic Terminals
                               Mossy Fibers, Hippocampal
                    Cholinergic Fibers
                         Autonomic Fibers, Preganglionic
                         Parasympathetic Fibers, Postganglionic
                    Nerve Fibers, Myelinated
                         Myelin Sheath
                               Neurilemma
                               Ranvier's Nodes
               Neurofibrils
                    Neurofibrillary Tangles
               Neurons, Afferent
                    Hair Cells
                         Hair Cells, Inner
                         Hair Cells, Outer
                         Hair Cells, Vestibular
                    Olfactory Receptor Neurons
                    Photoreceptors
                         Photoreceptors, Invertebrate
                         Photoreceptors, Vertebrate
                                Cones (Retina)
                               Rods (Retina)
                                     Rod Outer Segments
                    Retinal Ganglion Cells
               Neurons, Efferent
                    Motor Neurons
                         Anterior Horn Cells
                         Motor Neurons, Gamma
               Neuropil
                    Neuropil Threads
               Nissl Bodies
              Purkinje Cells
              Pyramidal Cells
              Senile Plaques


Next, at a step 905, the process arranges each sub-question according to the National Library of Medicine thesaurus keyword tree structure. This step may need some knowledge of the user to receive a desired result. The following is the arrangement of the previous keywords. To better understand the purpose of this step cells have been categorized as neural cells and non-neural cells. 

            TABLE 41
            Neural Cells
                    353  Neuroglia
                       601 Astrocytes
                         82 Microglia
                       268 Oligodendroglia
                       173 Schwann Cells
                  4,964 Neurons
                       423 Interneurons
                       823   Neurons, Afferent
                               95 Hair Cells
                               328 Retinal Ganglion Cells
                        67 Neurons, Efferent
                             1,084 Motor Neurons
                       257 Purkinje Cells
                       274 Pyramidal Cells
            Non-neural Cells
                    23 Stem Cells
                    31 Blood Cells
                        53 Lymphocytes
                               17 T-Lymphocytes


Next, at a step 907, as the keyword tree is arranged, the lower hierarchical keyword numbers should be added upwards into the hierarchy. The reason for this procedure is that very often an intermediate keyword in the tree structure might be scoreless while underneath it there are keywords with tagged numbers. Therefore, in reality the scoreless keyword should in actuality contain a number. For example the 17 keywords of T-Lymphocytes will be added to the 53 Lymphocytes set making it 70 Lymphocytes. The 70 Lymphocytes will now be added to the 31 Blood Cells making 101 Blood Cells. Finally, the 101 Blood Cells will be added to the 23 Stem Cells to make 124 Non-neural Cells. Following with our example the arranged keyword tree would look like the example of TABLE 42. 

                                 TABLE 42
                                 9,792 Neural Cells
                  1,477 Neuroglia
                         601 Astrocytes
                         82 Microglia
                         268 Oligodendroglia
                         173 Schwann Cells
                  8,315 Neurons
                         423 Interneurons
                       1,246 Neurons, Afferent
                               95 Hair Cells
                               328 Retinal Ganglion Cells
                       1,151 Neurons, Efferent
                             1,084 Motor Neurons
                         257 Purkinje Cells
                         274 Pyramidal Cells
              124 Non-neural Cells
                    23 Stem Cells
                    101 Blood Cells
                         70 Lymphocytes
                               17 T-Lymphocytes


Then, as shown at a step 909, the process continues where the keyword groups are sorted numerically and arranged in a decreasing manner. Our example would look like the data of TABLE 43. 

            TABLE 43
            9,792 Neural Cells
                  8,315 Neurons
                       1,246 Neurons, Afferent
                               328 Retinal Ganglion Cells
                               95 Hair Cells
                       1,151 Neurons, Efferent
                             1,084 Motor Neurons
                         423 Interneurons
                         274 Pyramidal Cells
                         257 Purkinje Cells
                  1,477 Neuroglia
                         601 Astrocytes
                         268 Oligodendroglia
                         173 Schwann Cells
                         82 Microglia
              124 Non-neural Cells
                    101 Blood Cells
                         70 Lymphocytes
                               17 T-Lymphocytes
                    23 Stem Cells


Next, as shown at a step 911, steps 901-909 are repeated for all other sub-questions. questions. Then, as shown at a step 913, steps 901-911 are all repeated in the database query limited by the number of group of years.

An important principle of the present retrieval system is to treat all years under the same conditions. Following the concept of equality shown in the application of a correction factor, correct each keyword with the correction factor described in that section.

Keyword Statistical Analysis According to a set of sub-questions

In relation to the data of TABLE 43, the data table now shows a hierarchical distribution of keyword cells. Using a non-specialized type of software like Microsoft.RTM. Word.RTM. or Excel.RTM. a pie chart can be made. This pie chart can be made by inputting directly the numbers shown in TABLE 43, which will be automatically converted into percentages by the software program. This step can be carried out by hand or automatically by the use of macros. Following the conversion into percentages TABLE 43, hierarchical distribution of keyword cells from the group of years of 1996-1998 will look like the data of TABLE 44. 

                      TABLE 44
                        98.7% Neural