Method and apparatus for proofreading a document using a computer system which detects inconsistencies in style

Abstract

A computer system for proofreading a document in electronic form performs the steps of: identifying elements of the document; interpreting elements of the document; creating known element objects; linking related known elements; and comparing known elements and linked objects to identify inconsistencies in the document. Further, the computer system for proofreading a document has an output device for supplying an identification of the identified inconsistencies.

Claims

What is claimed is:

1. A method for proofreading a document in electronic form using a computer system, which comprises the steps of:

identifying attributes of the document by parsing the document using a parsing rules database of the computer system;

interpreting the identified attributes of the document using an interpretation rules database of the computer system to identify elements of the document, wherein the interpretation rules database includes domain-independent and domain-specific rules;

creating known element objects in the computer system using the identified elements;

linking related ones of the known element objects using a linkage rules database of the computer system; and

comparing the known element objects and linked objects using a proofreading rules database of the computer system to identify semantic inconsistencies in the document.

2. An apparatus for proofreading a document in electronic form, comprising:

attribute identification means for identifying attributes of the document by parsing the document;

element identification means for identifying elements of the document using domain-independent and domain-specific rules according to the attributes identified by said attribute identification means;

style determination means for determining, as significant styles, attributes that are present in at least a predetermined number of the elements identified by said element identification means;

element linkage means for linking related ones of the elements identified by said element identification means, including linking similarly styled elements which share at least one significant style; and

inconsistency identification means for comparing the elements and linked elements to identify semantic inconsistencies in the document.

3. A computer system for proofreading a document in electronic form, comprising:

a. attribute identification means for identifying attributes of the document by parsing the document;

b. element identification means for interpreting using domain-independent and domain-specific rules the attributes of the document identified by said attribute identification means and for identifying elements of the document according to the interpretation of the attributes;

c. object creation means for creating known element objects using the elements identified by said element identification means;

d. element linkage means for linking related ones of the known element objects created by said object creation means;

e. object comparison means for comparing the known element objects and linked objects to identify semantic inconsistencies in the document; and

f. output means for outputting an identification of the semantic inconsistencies identified by said object comparison means.

Description

BACKGROUND OF THE INVENTION

The central problem addressed by this invention is the identification of the structure and significant units of text within an unknown, user document. This problem is complicated by an inability to pre-designate any firm, conclusive rules as to the use of any particular style at a given location in the document and further by the possibility of inconsistent structures and style usage within the same document.

Problems related and auxiliary to this central problem include: the identification and learning of user structuring styles; the identification of the relationship among numbered units of text that may be either nested or parallel and otherwise unrelated; the identification of additional, functional components within the text that function to conceptually link structural document components or pieces of text; the proofreading of document elements once they have been deciphered.

SUMMARY OF THE INVENTION

The subject invention relates to a method for proofreading a document which comprises the steps of: identifying elements of a document; interpreting elements of the document; creating known element objects; linking related known elements; and comparing known elements, thereby proofreading a document. Further, the subject invention relates to a computer system.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1: Process data flow diagram of a process of proofreading a document in accordance with the present invention.

FIG. 2: Block diagram of a system for practicing a method for proofreading a document in electronic form in accordance with the present invention.

FIG. 3: Process data flow diagram of a significant style counter in accordance with the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

This invention relates to a method for proofreading a document which comprise the steps of: identifying elements of a document; interpreting elements of the document; creating known element objects; linking related known elements; and comparing known elements, thereby proofreading a document.

In one embodiment of the subject invention wherein elements of the document are identified the method further comprises the step of identifying attributes of the document.

In one embodiment of the subject invention wherein elements of the document are interpreted the method further comprises the step of interpreting the document based on simple rules. In another embodiment of the subject invention wherein elements of the document are interpreted the method further comprises the step of interpreting the document based on a weighted rule base. In another embodiment of the subject invention the method further comprises the steps of interpretation of the document base on a heuristic rule base. In another embodiment of the subject invention the method further comprises the steps of interpreting the document based on a user guide.

In one embodiment of the subject invention wherein elements of the document are created the method further comprises the steps of deconstruction of the document.

In the subject application "additional significant style" is defined as a designation of a series of attributes found to be attached to occurrences of a particular type of known element. Each such style is determined by the association of these attributes with the known element in a number of occurrences that exceeds a pre-determined minimum number or proportion of occurrences of that known element. "Attribute" is defined as a category label associated with a word. "Conform" is defined as the process of editing an approximate phrase to correspond to its matching master phrase, or vice versa, retaining the grammatical structure of the source phrase as much as possible. "Document" is defined as any form of expression representable in the form of a sequence of characters (eg, a manuscript or a computer program). "Domain" is defined as a particular field of discourse having its own specialized terminology and types of documents, eg, law, software engineering, etc, "Embedded numeration" is defined as a pair of numbered phrases N and N', such that the numbered sequence of words in N' occurs as a subsequence of the numbered sequence of words in N. N' is embedded in N. "Functional element" is defined as a word, or sequence of words, in a document that serve a unified functional purpose in the document, independent of the words' individual meanings. In a particular domain, different types of functional element will exist. For example, in the domain of legal documents, one type of functional element is the "defined term", corresponding to usage of a term defined elsewhere in the document "Global style" is defined as a designation of that series of attributes to be attached to each occurrence of a particular type of known element. The Global Style is determined by the association of these attributes with the known element in a number of occurrences that exceeds a pre-determined minimum number or proportion of occurrences of that known element "Known element" is defined as a component of a document that is known to the invention and is found to possess certain attributes. "Lexical attribute" is defined as an attribute indicating if a particular word has independent meaning, or if its meaning is only in its grammatical function. "Phrasing" is defined as any part of a document conveying meaning, consisting of one or more words. "Semantic attribute" is defined as an attribute categorizing a sequence of words according to its meaning. For example, "Ab Def Inc." may be categorized as a "company name". Semantic categories may be organized in a hierarchy, with higher-level categories including many lower-level categories; a word is considered to have all semantic attributes whose categories it is included in. Note that in contradistinction to lexical attributes, semantic attributes may apply to sequences of several words. "Structural component is defined as a section of a document, a word, or sequence of words, in a document that serves a structural purpose in the document, independent of the word's individual meanings. Structural Components are domain independent, serving overall to sub-divide, structure and organize the phrasing contained in the document. "Syntactic attribute" is defined as an attribute giving a word's syntactic function. For example, in English text, this may be as simple as a part-of-speech label ("noun", "verb", etc.) or a more complex descriptor based on the word's relationship to other surrounding words.

In addition, the present invention relates to a computer system for proofreading a document, comprising a computer containing an executable proofreading document program in which such program imparts conforming functionality to the computer by changing the state of the computers logic unit upon execution. In another embodiment the computer program is stored on the computer hardware, software or in RAM. In the preferred embodiment the computer program is the process of proofreading a document as described herein.

In addition, the present invention relates to a computer system for proofreading a document, comprising: a. a display; b. a memory; c. a microprocessor; d. a storage; e. an input circuit; and f. an output circuit.

In one embodiment the computer system is shown in FIG. 2. In one embodiment the microprocessor includes inter alia the arithmatic logic unit. In one embodiment of the subject application the input means is a modem, mouse, keyboard, stylus and tablet, joystick, microphone, modem, direct memory access, pen, or voice command. Input means are known to those skilled in the art which permit a user to input. In one embodiment of the subject invention the output means is an output device such as a display, a printer and sound modem. Output means are known to those skilled in the art. In one embodiment a display is a CRT or monitor. In another embodiment the memory is RAM, ROM, PROM, EPRO, EEPROM program memory or data memory. In one embodiment the storage is a hard disk, floppy disk, optical disk, or tape.

Further, the sotware or hardware may be implemented in conjunction with any of a number of operating systems, including but not limited to: DOS, Windows, UNIX, etc.

As shown in FIG. 1 the first step, Unknown Element Parsing 2, is centered around the identification of the pertinent attributes of the text. Attribute identification 4 involves the rules-based gathering of certain attributes of units of text. This rules base, termed the "Parsing Rules Base," parses text in order to determine its attributes, as detailed below. Each attribute alone, or in tandem with other attributes, is used in the next step as a basis for identifying the nature and possible significance of these units of text. Only those units of text meeting a sufficient number of the criteria required by this rules base are deemed candidate Unknown Elements, deserving of additional analysis and interpretation in the next step.

The second step, Unknown Element Interpretation, involves the determination of the structural or functional nature of each Unknown Element. This is accomplished via any one of three steps.

The first step involves analysis with reference to an additional rules base, termed the "Interpretation Rules Base 6." If, in the event of excessive ambiguity, the Interpretation Rules Base does not suffice, heuristic processing 10 is used to enable confirmation. If heuristic processing does not suffice to resolve ambiguity, the user may be requested to input additional information to guide the interpretation.

The third step, Creation of Known Elements 18, involves the creation of a series of objects of various types from the formerly Unknown Elements. Each such object is termed a Known Element. Each individual object possesses the attributes specific to it in the text, even though the standard attributes possessed by objects of that type or the specific attributes possessed by other objects of the same type may differ. These differences are resolved in the sixth step detailed below, the Comparison of Known Elements.

The fourth step, Significant Style Counts (FIG. 3), involves the incrementation of a counter for each type of Known Element in order to determine the standard attributes associated with it. In the event that several different sets of attributes are used a significant number of times, several parallel counts of significant styles may be kept. One of these significant styles for each Element may be deemed the overridingly significant style, termed the "Global Style" 46, while the remainder may be retained as otherwise significant styles 48.

The fifth step, Linkage of Related Known Elements 22, rests on an additional rules base, the "Linkage Rules Base," that instructs the invention which Known Elements to link and where to link them. The insertion of these links allows several things, including comparison of Known Elements to one another, as in the sixth step below, as well as movement directly among linked Known Elements.

The sixth step is Comparison of Known Elements, a further rules based process. These steps, based on the "Proofreading Rules Base, 28" guide the invention as to where and in what ways to compare Known Elements. Essentially, this Rules Base examines Known Elements for several things, including: the presence of attributes that differ from the style(s) used in other occurrences of that type of Known Element; the presence or absence of linkages where the Linkage Rules Base mandates the presence of linked elements; and the presence of two or more Elements of a specific type that are either duplicative of, or substantively inconsistent with, other similar Known Elements. To resolve inconsistency in any of these events, the invention may make its own, rules-based determination or may request user input.

Upon the completion of this process, the invention has deconstructed the document into its structural components, has located any functional components that conceptually link different elements of the document, and has linked all related elements. Finally, it has proofread the document, noting inconsistencies among linked elements, enabling any inconsistencies to be remedied either by application of a further rules base or by user instruction.

The present invention addresses several problems associated with analysis of unknown, user-idosyncratic documents. The central problem is the identification and proofreading of the structure, associated significant units of text, and other, functional units of text within the document. This problem is complicated by an inability to pre-designate any firm, conclusive rules as to the use of any particular style at a given location in the document and a resultant need to extract user structuring styles. This is further complicated by the possibility of inconsistent user structures and styles within the same document.

For the purposes of this description, the following sample document is assumed. All of the examples discussed below relate to the issues posed by this document, including both its titles and its numbering styles the pieces of text associate with these numbers, as provided in these examples:

Document

1. Sub-Title 1. The text reads as follows: (a) this is the first section, (b) this is the second section including the following sub-sections (i) the first sub-section, (ii) the second sub-section and (iii) the third sub-section, and (c) this is the third sub-section.

2. Sub-Title 2. This text is ambiguously numbered (a) the first section and (b) the second section (i) a section with an ambiguous relationship to the preceding numbering series and (ii) a second entry in that numbering series.

3. Sub-Title 3. This text is numbered (a) the first section and (b) the second section (i) a new, unseen numbering style, (ii) second in that series and (iii) third in that series, and (c) this is the third sub-section, including a cross-reference, as stated in Section 1(a).

B. Parsing

Unknown Element Parsing involves a rules-based review of a pre-existing document in order to identify preliminarily relevant attributes of the text. The analysis of the document is conducted by way of applying a body of rules, termed the "Parsing Rules Base," to aid in the parsing of the document and the gathering of detailed attribute information. The Parsing Rules Base includes rules that relate to four types of attributes of a piece of text, including length attributes, textual significance attributes, format attributes and ancillary attributes. Each of these is further detailed below. These rules may include both rules that are independent of the domain of knowledge to which the document relates, as well as other, domain-specific rules that may enhance the accuracy of the parsing. The individual rules or groups of rules to be applied to the text under consideration may vary with the position in the document, the successful location of specific text attributes, and other factors. Where the application of the parsing rules base to the attribute-possessing text preliminary indicates the text to be a potential Known Element, the analysis of the text is continued by way of the Unknown Element Interpretation steps included in the second step.

Length attributes refer to the word count, line count or character count of the unit of text under analysis.

Textual significance attributes indicate the substantive significance of a unit of text in a document. These attributes include semantic attributes, lexical attributes, and functional attributes. Semantic attributes are those attributes that categorize a word or sequence of words according to their meaning. Semantic categorization of terms involves the include, for example, types of documents, types of laws, types of entity, and other categories that may transcend domains of knowledge or that vary from one domain of knowledge to another. A further type of semantic categorization is synonym sets, in which different terms are considered to have the same meaning. Lexical attributes refer to the lexical significance of words, i.e. the determination as to whether a word has meaning independent of its grammatical function. Functional attributes refers to the functional significance of a word or sequence of words within a document, i.e. the determination of whether a term carries functional meaning in addition to any semantic or lexical significance.

Format attributes relate to the visual formatting attached to a unit of text. These attributes include style attributes, spacing attributes, and capitalization attributes. Style attributes refer to the font style and other formatting of the text and any "white space" contiguous to such text. Spacing attributes refer to the justification of the text, the insertion of tabs in the text and the insertion of other "white space" contiguous to the text Capitalization attributes refer to the capitalization of part or all of the text.

Ancillary attributes relate to characters contiguous to, but not an integral part of, a unit of text and include non-letter attributes, punctuation attributes and numeration attributes. Non-letter attributes refer to the presence of characters that are neither letters, numbers nor sentence punctuation that serve functional purpose within the document. Punctuation attributes refer to the presence of significant punctuation beside the text. Numeration attributes refer to the attachment of numbers to pieces of text and the numeration styles of these numbers.

Assume the sample text provided above, with the title as follows:

"Document"

Further assume the Parsing Rules Base only requires the assembly of text formatting attributes if the position of the text (P) is the first page and the length of the text (L.sub.text) before any paragraph break is sufficiently short to be potentially indicative of a title, where:

L.sub.text .ltoreq.4 words

As the aforementioned text possesses this attribute, the Parsing Rules Base will continue to gather the attributes of the text, as follows:

Format Attributes (F)=Bolding and Underlining

Capitalization Attributes (C)=All Capitals

Justification Attributes (J)=Centered

C. Interpretation

The second step, termed Unknown Element Interpretation, may involve three separate components, comprising rules based analysis, heuristic processing, and user interface. First, as above, the Interpretation Rules Base may include both rules that are independent of the domain of knowledge to which the document relates, as well as other, domain-specific rules that may enhance the accuracy of the parsing. The individual rules or groups of rules to be applied to the text under consideration may vary with the position in the document, the successful location of specific text attributes, and other factors. The function of these rules is to interpret the assembled attributes of the text and thereby to determine the nature of the units of text, i.e., the Unknown Elements, possessed of these attributes.

1. Simple Rules Based Interpretation

Assume the sample text provided above, with attributes as follows:

P=Page 1

L.sub.text <4 words

F=Bolding, Underlining

C=All Capitals

J=Justification Attributes

Assume the Interpretation Rules Base sets the following interpretation rule.

If:

P=Page 1

L.sub.text <6 words

AND

F=Bolding OR Underlining OR Italicization

AND

C=All Capitals

OR

J=Centered

THEN

Text=Title

As the aforementioned text possesses four of these five attributes, it is determined by the Interpretation Rules Base to be a title.

2. Weighted Rules Based Interpretation

The accuracy of the rules-based interpretation may be heightened by the use of weighting of rules. This weighting may enable the rules to more accurately analyze the structure of the text, as the weights reflect the probabilities associated with occurrence and significance of various text attributes. The weighting of rules may enable the invention to interpret certain otherwise-ambiguous text without further resort to the user in several situations. The significance of weighted rules-based decision making is made clear by the following example with respect to the determination of the nature of ambiguous text:

Assume the sample text, with the title as follows:

"Document"

Further, assume the Interpretation Rules Base contains a two-tiered rules structure, where the first tier contains two rules that must be met to qualify as a title, while the second tier contains the following weighted interpretation rules structure:

Tier 1

If:

P=Page 1

L.sub.text <6 words

THEN: Tier 2

ELSE: End

Tier

    ______________________________________
    F =               Bolding (2)
                      Underlining (2)
                      Italicization (1)
    C =               All Capitals (10)
                      Initial Capitals (4)
    J =               Centered (1)
    ______________________________________

    ______________________________________
    F =               Bolding = 2
                      Underlining = 2
    C =               Initial Capitals = 4
    ______________________________________

    ______________________________________
    Global Style:
            First Embedded Numeration Style: a, b, c
            Second Embedded Numeration Style: i, ii, iii
            AND
            Second Embedded Numeration is further embedded in First
            Embedded Numeration
    ______________________________________

    ______________________________________
    Global Style:
            First Embedded Numeration Style: a, b, c
            Second Embedded Numeration Style: x, y, z
            AND
            Second Embedded Numeration is further embedded in First
            Embedded Numeration
    ______________________________________

    ______________________________________
    Additional Significant
                 First Embedded Numeration Style: a, b, c
    Style:       Second Embedded Numeration Style: i, ii, iii
                 AND
                 Second Embedded Numeration is further
                 embedded in First Embedded Numeration
    ______________________________________

    ______________________________________
    Global Style:
            First Embedded Numeration Style: a, b, c
            Second Embedded Numeration Style: i, ii, iii
            AND
            Second Embedded Numeration is further embedded in First
            Embedded Numeration
    ______________________________________

    ______________________________________
    Sections:  <2. Sub-Title 2. This text is ambiguously numbered
               . . . that numbering series.>
    Section Attributes:
               Sub-Titled.
               Spacing: 0
               Justification: Left
               Length: 39 Words
    Global Style:
               First Embedded Numeration Style: a, b, c
               Second Embedded Numeration Style: i, ii, iii
               AND
               Second Embedded Numeration is further embedded
               in First Embedded Numeration
    Numbering Style:
               <Arabic Numeral>
               Structural Punctuation: <.>
    ______________________________________

    ______________________________________
    Section Sub-Title:
                     Sub-Title 2.
    Sub-Title Attributes:
                     F = Underlining
                     C = Initial Capitals
                     Structural Punctuation: <.>
                     Structural Conjunction: None.
    ______________________________________

    ______________________________________
    Sub-Sections: <Sub-Section 1: (a) the first section and>
                  <Sub-Section 2: (b) the second section (i) a
                  section . . . that numbering series.>
    Sub-Section Attributes:
                  No Sub-Title.
                  Numbering Style: <a, b, c>
                  Structural Punctuation: None.
                  Structural Conjunction: <and>
    ______________________________________

    ______________________________________
    Sub-Sub-Sections:
                   <Sub-Sub-Section 1: (i) a section with an
                   ambiguous . . . series and>
                   <Sub-Sub-Section 2: (ii) a second entry in
                   that numbering series. >
    Sub-Sub-Section Attributes:
                   No Sub-Titles
                   Numbering Style: <i, ii, iii>
                   Structural Punctuation: None.
                   Structural Conjunction: <and>
    ______________________________________

• Inventors
  Razin, Ely Y;
• Assignee
  Expert Ease Development, Ltd. (Jerusalem, IL)
• Published
  Sep-26-2000
• Current US Classes:
  704/10
  715/531
• Application #
  617260
• International Classes
  G06F 017/21
• Field of Search
  707/530 707/531 707/532 707/533 707/540 704/2 704/4 704/5 704/6 704/9 704/10
• Examiner
  Hong; Stephen S.
• Agent
  Maioli; Jay H.
• US Patent References:
  5485372
  5576955
  5598518
  5678053