Method and apparatus for performing broadcast operations6198906Abstract A method and apparatus for performing broadcast operations that enables a programmer to schedule broadcast constituents, such as programs, schedules, commercials, etc. according to relative sequential orderings indicative of the sequence the constituents are played. The corresponding material which manifests select of the constituents and the media which manifests the material are also identified such that to play the broadcast, the structure is accessed to identify the sequence of play and the media to be played. Claims What is claimed is: Description BACKGROUND OF THE INVENTION
p 0x0002 pi 0x0200
d 0x0004 di 0x0400
s 0x0008 si 0x0800
f 0x0010 fi 0x1000
m 0x0020 mi 0x2000
o 0x0040 oi 0x4000
= 0x0101
t 0xFFFF
f 0x0000
This embodiment enables the creation of the converse of any relationship between endpoints by swapping the upper and lower bytes of the number. Since equals is its own converse, it is the only 2 bit relation. To perform a lookup of the endpoint relations corresponding to any given interval relation, the endpoint relations are stored in a hash table whose key is the value of the interval relation encoded above. To look up the composition of two interval relations (needed during the matrix multiplication process), the composed values are stored in a 13.times.13 matrix where each entry is the composition of a single relation to a single relation. In practice, with the above encoding of relations, the proper row/column index is found directly as the bit position of each relation with appropriate adjustment for the equals position, and that not all of the low order bits are used in each byte. Therefore a table lookup is a composition of two simple (non-ORed) relations. When an OR is involved, e.g., compose (=p) with (m o), the algorithm looks up the four combinations of the individual relations and ORs them together to get the composition result. It can therefore be seen that as BCP performs repeated squarings of the interval matrix, using a table look up the square operation is preferably performed exactly as multiplying two numeric matrices except that composition substitutes for multiplication and logical intersection substitutes for addition. Once the relationships have been determined, the relationship matrix is reduced to a sequence of endpoints (SOM and EOM). The sequence is a list of endpoints in the order needed to satisfy the interval relations. Preferably the sequence is generated by referencing a table that provide the corresponding SOM and EOM for two intervals based upon the interval relationship between the two intervals defined in the relationship matrix. A portion of an exemplary table, for relations 1 through 68, used is shown in FIGS. 7a-7c. It should be noted that some members of the sequence of endpoints may be a cluster of endpoints denoting simultaneity. For example, events in auto-follow mode will create EOM/SOM pairs to show that one starts exactly when the other one finishes. Therefore, preferably, the sequence of endpoints is created via a two-step process. First, the interval relation matrix is converted to a matrix of endpoint relations by a look-up process on each entry (one interval relation yields six endpoint relations). The endpoint relations are then sorted to produce the sequence. The sort should take account of possible bifurcation in the endpoint relations which may result in multiple sequences. Preferably, the matrix is topologically sorted to produce the final sequence. To topologically sort, a complete sequence of elements, for which there is only a partial ordering, is arranged. For example, if the endpoint relations are A<B, B<D, A<C, then there are several possible total sequences that satisfy the constraints: ABDC, ABCD, ACBD. This also is an example of how multiple possible sequences exist for a set of endpoint relations. Multiple sequences imply points of departure for alternate schedules where runtime choice is made. It may also be that multiple schedules converge back to a single endpoint which implies a rejoining of the schedules. In practice, as will be apparent to one skilled in the art, this two step process can be reduced to one step, requiring only a single traversal of the interval relations in the matrix. The system and method of the present invention provide a powerful tool for a user to determine a schedule of programming as well as to allocate resources to use to provide programming. One example is the use of the present invention to schedule operations which utilize broadcast equipment such as video tape recorder (VTR) equipment. In the broadcasting field, events can be viewed as something that happens to equipment, people or accounting/tracking systems. For example, an event might be a programmed event which is an event related to an actual programmed commercial break, a cut (A cut is a change in the video signal from one source to another. This change happens between 2 video frames, one from the first source, the second from the other source), a switcher event (a switcher is a device that has as input 2 video sources and transition the video output between the two sources over a period of time, e.g., 1 second), or a GPI trigger (a General Purpose Interrupt (GPI) is an electrical contact closure that causes some other action to occur, such as starting a VTR to play or stop). The above exemplary events can be defined by pairs of interval relations. In the present example, the events are operations defined by the physical operation of the VTR, such as play video, stop video, trigger GPI start switcher action, etc. These actions are directly related to endpoints of an interval. For example, the physical act, start video, is the action that occurs coincident with the starting endpoint of a video clip. Some physical acts imply a relationship among several non-zero duration actions. For example, a switcher event is a non-zero duration action which overlaps portions of two other actions, each being plain video clips. The intervals would be specified as follows: V1 overlaps V2--the first clip (V1) overlaps the second clip (V2) S finishes V1--the switcher event ends V1 S starts V2--the switcher event begins V2. The power of the present invention with respect to the present example is that it completely describes the ordering relationships among the physical actions performed with respect to a video system. For example, play a clip, transition between clips, trigger a clip, choose a clip from a set of alternates, and from the description, deduce all the single actions required to implement the specific relations. Patterns of event relationships can be reused in different schedules and different places in the schedules. For example, a pattern describing a (e.g. commercial) television show can be used many times in a broadcast schedule. There are several very common events in video operations which are describable by fairly elementary interval relations. For example, cuts in editing, or auto-follow in playlists, specify that the end of one clip (A) is immediately followed by the start of another clip (B) with no intervening time. Thus, the interval relationship would be (A) meets (B). For a GPI trigger, a clip of (A) is started by the occurrence of a GPI trigger (G). The duration of a GPI always is less than the duration of a clip; thus, the interval relation would be (A) started-by (G). A switcher event involves two video clips (A), (B) and a switcher action (S) of finite duration which creates desired transition effect. Thus, the interval relations specified are (A) meets (B); (S) overlapped by (A); (S) overlaps (B). The specification (S) overlaps (B) is important since it defines that the switcher event terminates before event (B) terminates. The following is a more complicated example in which alternate programming is defined. In particular, this case illustrates a common situation of a live program of indefinite duration (event L) which is followed by a pair of clips (C1, C2), having different durations and where the choice will be made at run time as to which clip C1, C2 to use. The clips are further followed by a single event (H) scheduled to start at a "hard" time. The end of the live event triggers C1 or C2 and the "hard" time triggers H; thus, the interval relations are: L meets C1, C2=starts, or is started--by C1; C1 precedes or meets H; C2 precedes or meets H. The relation L meets C1, C2=starts, or is started--by C1 specifies that L starts, L ends followed immediately by the start of C1. C2 begins at the same time as C1 but it is not known which one finishes first. In other words, C1 and C2 may be of different lengths and its not known which is longer, if any. It be noted that it is irrelevant whether L is specified to meet Cl or meet C2. The constraint propagation algorithm will deduce the other given a second constraint. Likewise, a second constraint could correctly describe C1's relation to C2 and the algorithm would determine the constraint. The last two constraints define that no matter the run time choice of C1 or C2, the running clip will stop the moment it starts. Note that the running clip may finish leaving black video if the "hard" time start of H has not occurred. The system and method of the present invention also support what is referred to herein as event variables. Event variables are events which can participate in a set of interval relations, but which do not indicate a specific playable entity, such as a clip or trigger. Event variables can act as sequencing space holders pending specification of the particular clip spot, trigger, etc. Event variables are identified herein by a name starting with "?" (e.g., ?X). Variables may also be defined to take the place of a particular type of event, such as a GPI program segment, etc. The convention used herein is to identify such a variable by the addition of a colon and the type name (e.g., ?X:GPI). Thus, a variable is a kind of constituent which "takes the place of" a real one. In terms of implementation, the variable is used just as any other constituent in terms of creating a composition structure. A variable has no substructure and no representation itself, but it can be "bound to" (usage of the term bound to can be liken to usage of "bound" and "binding" in the computer art where variables are bound to values during the execution of the code). When queried, a constituent variable obtains its as-composed duration from the structure of the constituent bound to it and gets its as-represented duration from the representation of the constituent the variable is bound to. For example, take a sequence of 3 elements: A, ?X, and B. Since ?X is not yet bound to a real element, any query directed at ?X will not return a definite value. But the sequence can still be placed into a pod, schedule, program, etc. At some time before the sequence is executed, ?X is bound to element E. Now, due to the binding process, the sequence appears to be A, E, B. Constituent variables support pre-defined patterns of constituents which can be re-used any number of times and only require specification of what real constituents are bound to the variables. This particularly advantageous for repetitive programming in which the content may vary, but the scheduling characteristics remain constant. As noted above, in order to use variables in the specification of intervals, it is necessary to specifically provide a binding for each variable. A binding is established by specifying an interval relationship between the variable interval and a real event interval that is bound at run time. A variety of binding processes are known to those skilled in the art. However, it is preferred that an association specifying an equality relationship is specified such that it participates in the same interval algebra calculation and no further calculation is necessary. The context of the association dictates the event that the variable event is bound. This is discussed in further detail later in this document. The use of variables can be expanded to include more than one variable in interval relationships. These variables are typically distinguished by name. For example, in the interval relation,
?L: Studio1 meets C1, or is started - by C1
C1 precedes or meets ?H;
C2 precedes or meets ?H
The live event "L" and "hard" time event "H" are not yet specified. However, when they are subsequently specified, the live event L must be an event of the type Studiol. By establishing event variables, more complex relations can be specified. In addition, commonly used relations can be specified once and reused repetitively as needed. The following examples illustrates patterns of interval relations which are common in broadcast scheduling. It should be noted that lines of the code shown below that begin with a semicolon (;) denote a comment line commonly used in programming. A "donut" is a pair of clips, commercials or spots (S1, S2), referred to hereinafter collectively as spots, which must be played as a unit but with exactly one other spot between the two.
Interval relation: S1 meets ?X
?X meets S2
S1 and S2 are two specified spots and X is a variable spot. Thus, when scheduled S2 will always be played after S1 and some spot later bound to X. A bookend is a pair of spots (S1, S2), one of which occurs at the beginning of a commercial pod (?P) and the other at the end. A pod is a term used to define a group spots played together as a unit. Typically, the length of a pod corresponds to the length of a commercial break. The interval relation for a bookend is defined as:
Interval relation: S1 starts ?P:Pod
S1 precedes S2
S2 finishes ?P:Pod
The above defines that spot S1 starts the beginning of P which is of the type "pod". It is also stated that spot S1 precedes S2 and S2 finished the pod P. Note that the use of the operator "precedes" between the spots S1 and S2 specifies that there is room for any number of spots between S1 and S2, and that all of the additional spots also occur within the same break. A straddle is a more complex pattern where there are related spots, for example, four spots (?S1, ?S2, ?S3, ?S4), each of which is placed at the beginning of a different break during predetermined period, e.g., hour, of broadcast. For this pattern, we assume 5 program segments (?P1 . . .?P5) in the hour, where P1 is the 90 second lead-in segment that occurs at the top of the hour.
Interval relation: ; program segment relations
?P1 precedes ?P2
?P2 precedes ?P3
?P3 precedes ?P4
?P4 precedes ?P5
; spots start at top of break
?P1 meets ?S1
?P2 meets ?S2
?P3 meets ?S3
?P4 meets ?S4
; other spots can be inserted
?S1 precedes ?P2
?S2 precedes ?P3
?S3 precedes ?P4
?S4 precedes ?P5
As the program segments ?P1 . . .?P5 and spots ?S1 . . .?S5 are variables, this structure can be used repeatedly by binding the variables to different constituents. It should be noted that integration of this with a more formal event compositional structure, such as is found in predicate calculus, can provide a level of modularization which would make this pattern and other patterns even more useful. For example, patterns can be used within patterns to generate even more complex patterns; furthermore, it is contemplated that pattern matching among common patterns can be used to deduce possible bindings for the variables. Another pattern encountered is the "alternates with auto-follow". This pattern is a variation of the live event/alternate/timed-start pattern. The pattern includes a live event (L), two possible alternates (C1, C2), and an element (A) that auto-follows whichever of C1 or C2 is chosen at run time. To solve the problem of specifying the "meets" relationship between C1 and C2 to A, a variable, "G", which represents the alternate choice at run time, is used. The resulting pattern is:
Interval relation: L meet ?G
L meets C1
; provide a path from L to A
L precedes A
; C1 and C2 start simultaneously
C1 starts, equals, or started-by C2
; A auto-follows something
?G meets A
Thus, there is an indefinite period between the end of L and the start of A during which either C1 or C2 will run starting at the end of L. To establish the auto follow relationship between the running element (C1 or C2) and A requires that the choice of C1 or C2 be made by executing one of these rules at run time.
Interval relation: ; one of the next two is chosen at run time
?G equals C1
?G equals C2
Once the link is established, the rules specify the desired effect. Note that the simultaneous start of C1 and C2 also means that the system will know to allocate sufficient resources to play both clips, assuring that a last-moment choice is possible. It is contemplated that such as system would prompt the operator to make such choices at run time in order to meet the interval relationships specified. As can be seen above, interval relationships can be defined for a variety of broadcast programming situations. Furthermore, interval relationships can define usage of resources such as VTRs, satellite feeds and the like. Interval relations provide time-independent (i.e., "uninstantiated") orderings of endpoints of events. Having reduced the problem to a set of endpoints and their relations, temporal constraints are then added to the existing structure by instantiating some or all of the endpoints of the events. Temporal constraints may be specified in a mix of two different ways, hard time constraints, and durations and offsets. Hard time constraints specify exactly when an event starts or stops. Thus, particular endpoints are fully instantiated. An implementation may choose to instantiate the endpoints as absolute time values (12:00:00:012 May 12, 1997) or as an offset time to an externally specified value (30 min from start of playlist). Endpoint differences such as durations or offsets can be stored in with the endpoint or interval relations. An example would be the interval relation: Y starts 5 seconds after X starts. This is illustrated below: ##STR1## Typically, offset values are used to specify start-to-start and end-to-start durations or offsets. As is readily apparent, durations and offsets can also be variable and bound subsequently. By evaluating the endpoint sequences and applying simple arithmetic formulas to the values specified for the endpoints, durations or offsets, the exact structure of a series of timed events can be calculated. Further, any violations of ordering constraints are easily visible as a matrix entry containing a false value immediately identifies an inconsistency. The results of changing any time value can be calculated by following the sorted endpoint list and performing the indicated calculations until either some constraint is violated or some fixed time point is encountered. There are no fixed "ripple rules" which must be encoded into a playlist handler. The "rules" are a combination of the endpoint sequence paths and the local constraints at each endpoint in the list. A local constraint is a property of the ordering edge that connects two endpoints. The edge specifies that, for example, E1<E2. This condition must always hold true. Endpoints may also contain information such as a time value which, when paired with another node in the endpoint in the network, must satisfy the time values as well. In the above example, there is a different kind of inconsistency if E1 has a start time of 1 pm and a duration of 5 minutes and E2 has a start time of 1:03 PM. This illustrates that at some point in time, the time constraints can not fall within the endpoint constraints. The local constraints can be tracked in a separate list or structure. In the present embodiment, local constraints are coded in the association between two intervals. Alternately, certain interval relations may carry externally specified offset relations between pairs of endpoints of the interval. The constraints, as described above, can be viewed as ordinal (e.g., relative or qualitative) or cardinal (specific numeric). The constraints are created as a result of the specified interval relations and are recalculated when the relations change during any phase of the broadcast process, such as scheduling, editing or operating a broadcast schedule of programs, commercials and the like. The merger of ordinal and cardinal constraints enables immediate cross-checking for violation of ordering constraints when time values are changed or for violation of temporal constraints when the play structure (e.g., playlist) is changed. The process is so efficient that it can be used, for example, during a live broadcast to automatically eliminate schedule alternatives that will not be able to execute in the time remaining. In such a system, the system would evaluate the constraints after predetermined time period lapses, or on a continuous basis, to identify when certain constraints are violated and eliminate schedule alternatives based upon the constraints violated. For example, the operator is provided two programs, Pgm1 which is of one length and Pgm2 which is longer, that can follow a live broadcast of unknown duration. The interval relations specify constraints that the selected program must terminate before a specified hard time. The system accessible to the operator making the selection at run time evaluates the constraints, relative to the current duration of the live program, and eliminates the program(s), e.g., Pgm2, that is too long to play to completion following the live broadcast and completing before the specified hard time. As is readily apparent from the description herein, the system can be configured for different portions of the broadcast process (e.g., scheduling, operating, relevant billing and accounting of commercials) by specifying a broadcast in terms of interval relations. Furthermore, the complexity and sophistication of the system can vary according to application. For example, the system can be configured to provide simple information and feedback to a user, such as a scheduler of broadcasts. Alternately, for example, the system can be configured as the basis of an automated broadcast generation system that accesses the necessary media for generating a broadcast specified in accordance with the teachings of the present invention. In addition, the information can be used to account for commercial broadcasts and insure the broadcast constraints (e.g., temporal or content) specified by the sponsor are met. Below is a simple example combining the application of interval algebra, temporal constraints, event variables, and variable typing. The situation is the scheduling of a VTR to play a clip. VTRs require a pre-roll time (PRT) before playing a specified clip, but the scheduling of the clip must happen in relation to other scheduled clips.
Interval relation: ; define a schedulable VTR entity
VP:VTR started-by PRT
VP:VTR finished-by ?V:VTR
; define VTR as pre-roll plus clip
PRT meets ?V:VTR
; create an external interface (?C)
?C equals ?V
; set the pre-roll time
Duration (PRT) = 5
; add a real event binding
DNC-ad equals ?C
This set of rules defines a resource (VP) which is a VTR, can play some clip ?C and includes a pre-roll time (PRT) of 5 seconds. The externally visible interface is the clip variable ?C. ?C is bound to a commercial event named DNC-ad. Therefore, a restriction is specified on the start of VP that it begin 5 seconds before the DNC-ad. When this composite event is combined into a schedule, processing the interval relations results in a specification of the order and time to start the VTR. The power of system and method of the present invention is sufficient to detect a resource conflict (VTR not available at that time) or an inability to fix a precise offset to some preceding event, meaning that manual intervention is necessary to start the VTR. The present invention is a powerful tool for the specification, sequencing and execution of events in a broadcast environment. These events include, for example, playing clips, triggering actions, cutting, switching, transmitting conditional access data, tuning receivers, choosing a story alternate, and any other machine-controllable or displayable action which must occur in the operation of a modern broadcast plant. It is independent of the specific applications such as spot insertion, news, or editing, and can be used to integrate the complete sequencing of all these applications. The ability to define patterns of event relations and include event variables within the interval algebra framework can be utilized to create more powerful scheduling, traffic and automation systems allowing these applications to retain the higher level semantic intent of the programmer or director. The system is further enhanced by the utilization of the data structure described herein and the processes implemented to access the structure to perform a wide variety of broadcast functions. The basic component of the structure is a constituent. A constituent is the umbrella category for the types of data structures in the model. All constituents share certain attributes and behaviors. Although a specific set of attributes are described herein, it is contemplated that varying sets of attributes, including some of what is described here as well as additional attributes not described herein, can be used. In the present embodiment, the attributes include: ID: A globally unique identifier which may be used to reference or retrieve the constituent from anywhere in the system. Name: A human-readable designation for the constituent. Temporal Specification: A specification of the starting, ending and duration values. Track Set: A set of tracks (e.g., audio, video or control) defined by the constituent. The track set can have any number of tracks of any number of types. There are three specializations (types) of the basic constituent: element, material and media. Each specialization models the behavior of one aspect of broadcast systems. For purposes of composition, each type can be composed from other members of that type. This provides recursion in the structure and additional flexibility in defining broadcasts and its constituents. Elements are a type of constituent used for scheduling actions such as play a clip, trigger a GPI (General Purpose Interrupt), start an effect, or choose among alternative programming to broadcast. The element type of constituent can be representative of the element constituent as well as any of the higher order constituents such as a pod, program, etc. The element type is different from the other types in that a constituent of the type element is not bound to specified times but may be organized in a time-independent manner. This is particularly important for implementing the specification and handling of live events, default actions and alternative choices. Elements can be represented by material. Material is a kind of constituent that represents the actual signals played or specific action that should occur during a broadcast. For example, material may be the signal in a turnaround feed at a broadcast center or the content of a syndicated program. In a digital sense, it is "the bits" without regard to where the bits are stored or where they come from. A material may also be a specification for a trigger action ("Take") although the specific GPI line, for example, is not specified. Material may also be non-audio/video data, such as conditional access data typically used in cable and satellite transmissions to control access to transmissions or scripts to be output on a teleprompter. Material may be represented by media. Media is the constituent which represents the physical manifestation of a material. Media is the constituent that transports the material. It can be magnetic tape, files on a hard disk, an incoming signal feed, a GPI trigger line or any other physical representation of material. Using the above three constituents, the identification broadcast elements, how the elements are embodied and how the element is physically manifested are described. As will be described below, this basic structure enables systems, for example, to identify what is broadcasted and the availablility of broadcast resources. The capabilities of systems that operate in accordance with the teachings of the present invention are enhanced through the use of constituent variables. A constituent variable can be viewed as a place holder constituent which can be bound to a real constituent at a later time. There are three types of constituent variables corresponding to the three types of constituents. Constituent variables can only be bound to constituents of the same type. Constituent variables have neither compositions nor representations. Constituent variables are bound to constituents via association relationships where the relationship is defined as equality. The naming convention used herein for constituent variables is ?name:type, where name is a user-chosen label and type is one of the element, material or media constituents described. Constituent variables, in addition to substituting for a single constituent, can take the place of a list of constituents. This can be indicated, for example, by concatenating the `[]` symbol to the end of the variable's type indicator. Such a constituent variable can take the place of any number of constituents of that type. As will be described below, constituent variables can be used to construct common, re-usable patterns of broadcast elements. Constituents may relate to each other in several different ways. A constituent relation is a three-way relationship among constituents such that the relation is defined between two constituents and that relation is valid in the context of a third (which may be one of the pair). The presence of the context constituent may introduce dynamism into the information structure in that, for some situations, the meaning of the structure at any point can be completely determined only by knowing how that point was reached in traversing the structure. In the present embodiment, there are three different ways in which constituents may relate to each other, composition, representation and association. The composition relation defines the set of subcomponents of a constituent and may provide additional information such as the time base or track mapping. Composition defines a child-parent, part-to-whole relationship. A composition is defined in the context of the parent constituent. The representation relation defines how a constituent may be manifested by another constituent. The second constituent contains sufficient information and data to completely manifest the intent of the first constituent. Representation defines an equality relationship in time and track sets. A representation is defined in the context of the constituent whose representation is being defined or a compositional ancestor of that constituent. In particular, in the present embodiment, a scheduling element can be represented by a material and/or media. The association relation defines arbitrary ordering relationships among constituents of the same type. These ordering relationships are based upon the context of the association wherein the context of the association is the composition. In the present embodiment, a composition object is created for the constituents which are composed of other constituents. The composition object identifies the parent as the contact and specifies any associations between siblings. Usually, associations are created between sibling constituents but are not restricted to siblings. The context of associations is usually the parent of the siblings involved. It may be an ancestor constituent of the siblings in the data structure. Associations control the relative ordering of elements in the context of the composition. For example, associations are used to define a temporal relationship between scheduling elements, e.g., that element 1 is to be played before element 2. Constituent relations define two distinct mappings between constituents. One relationship, a time base transform, defines how to transform between the time specifications of each constituent. For example, different constituents describe time in different ways. For example, an element may be defined to start at 12 noon and the material that represents the element is identified to being at 1 hour, 6 minutes and 30 seconds. The time base transform in the example would generate an offset value (e.g., 1 hour, 6 minutes, 30 seconds) to identify where to go in the material for the element. The time base transform may also adapt a shorter segment in the media constituent to the longer material constituent, or a longer media constituent into a shorter material constituent. A second relationship, a track set transform, defines how to select and merge tracks of one constituent into the tracks of the containing constituent. Signals, audio video or control, come across tracks. The types of tracks are defined as to the kind of input it can handle. For example, audio tracks convey audio information. A track transform may be a simple 1:1 correspondence wherein the track material corresponds to a specific track media. The transform also can function to combine two tracks into one. This is particularly useful when the media contains more track than the material. A predetermined mix function, e.g., 50/50 mix, a mix function that is a function of time, or other mixing function can be used. The time base transform is reversible (i.e, it can go in either direction). The track set transform may not be reversible. This would be the case when, for example, the track set transform specifies a mix-down of four audio tracks into one. Containment refers to the ability of one constituent to use another as a portion of its content. This is seen in the present embodiment wherein one scheduling element can be composed of one or more scheduling elements of the same or lower order. The data model enforces a containment restriction that constituents can contain only constituents of the same general type. Scheduling constituents are arranged in a hierarchy defining what kind of constituent can contain another. One embodiment of a scheduling hierarchy is illustrated in FIG. 8. The hierarchy is defined by current broadcast conventions of terminology and usage and is not a fixed requirement of the data model. The implication of the hierarchy is that a scheduling constituent at some level can contain constituents of that type and ones lower in the hierarchy. Thus, a channel broadcast can contain a schedule which can contain a program which can contain a pod which can contain an element. However, for example, a pod can contain elements but not programs. An element is the lowest level scheduling component. For example, a clip or spot may be considered as an element. An element can contain only other elements. This situation occurs, for example, in the specification of an editing sequence of clips, a particular grouping of related spots such as a donut, or a choice of alternative clips or news stories, one of which is subsequently selected to play at time of broadcast. A pod is a grouping of short-form elements (spots) into a commercial break. In the news program context, new stories correspond to the pod structure. A program is a grouping of long-form elements and pods into a broadcast segment (typically an hour or half hour). Programs are also known as rundowns in news operations. A schedule is a grouping of programs, pods and other elements into a meaningful broadcast unit such as "Prime Time" (e.g., from 8:00 PM to 11:00 PM) or "Morning News". One illustration of the recursive nature of a schedule are daytime schedules which are often built from other schedules reflective of schedules of portions of the day. A channel is a grouping of schedules and other subordinate elements into a continuous stream of input or output. Channels may be grouped into other channels. For example, in a multi-channel facility, a transmission channel may be formed from several different facility channels during a broadcast day. As noted above, the scheduling constituents can be represented by other constituents such as material and media. When one constituent represents another, the one constituent has the property of being able to manifest the intent of the represented constituent. For example, a tape cassette is a media which can represent a material constituent, e.g., the 6 o'clock news. In the present illustration, a constituent may be represented by one other constituent not of the same constituent type. For example, a material can represent an element and media can represent material. Note that the representation relation is transitive. That is, an element can be represented by a material which is in turn represented by a media. It is correct to assume that the media can represent the elements. The representation of one constituent by another has the property of that duration of the representing constituent, as mapped to the presented constituent, equals the as presented duration of the represented constituent. It follows that, since the context of a representation may be anywhere in the represented constituent's composition hierarchy, the as-represented duration is dependent on the traversal history of the information structure. The context of a representation is the "represented" constituent or one of the represented constituent's ancestors in the composition or scheduling hierarchy. Unlike composition, the context is dependent on the dynamic behavior of the structure-traversing algorithm. Thus, the context of the representation depends on how the represented constituent was reached in traversing the structure. As should be readily apparent, different structure traversing algorithms may dictate different context. Preferably, in order to track the context, and as will be illustrated below, the traversing algorithm should implement a form of push-down automaton (a finite state machine combined with a stack). For example, when walking or traversing the tree structure, as each node is reached, the constituent is pushed on a stack (the stacks indicating the path traversed). Furthermore, each association encountered is in the context of its composition. Therefore, when a composition, i.e., composition object, is encountered, the object is pushed on the stack and subsequently is traversed and executed when the child nodes are traversed. An association is an ordering relationship between a pair of constituents, usually used in the context of a composition to provide relative ordering information between sibling constituents in the composition. For example, three elements are composed together where the first two are audio-video (a/v) elements and the third is a GPI trigger. While the composition defines that all three make up the parent constituent, e.g., a pod, associations are used to specify that the GPI element and the first a/v element start together (the GPI triggers the first element) and the second element starts sometime after the first (e.g., an offset voice-over). Note that associations can be independent from any time base specifications. The context of an association is the parent or ancestor constituent of one of the associated constituents. It is recommended practice that associations be used between constituents of the same type in order to avoid significant computational complexity. Any constituent can have at most one composition and one representation. However, it is also necessary to accommodate the possibility that some constituents might have alternative compositions or multiple representations. Alternative compositions occur in elements when there are alternative possible actions in a playlist, for example, following a live event. Multiple representations occur in elements where there may be several materials which represent the element or in material where there may be multiple tapes ("clones") that each store a copy of the desired program. The model handles this situation by providing the concept of a proxy relation. A proxy relation is a container for a set of possible relations of the same type. The proxy appears to any user to be a constituent relation in terms of methods and attributes it exports. However, in addition to the contained relation set, the proxy also carries a set of selection rules that are executed at runtime to determine which relation, if any, is appropriate. A proxy relation is owned by the creating constituent. It is this constituent which specifies the selection rules for the proxy. Proxies which have not calculated a constituent relation to return are said to be "unresolved". Proxies may be resolved dynamically during active traversal of the structure or at any other appropriate time. The context of an unresolved proxy is the creating or owning constituent. The context of a resolved proxy relation is the same as that of the selected relation. The selection rules may be implemented as a computer function, a set of logic rules, a pattern for a pattern-matching algorithm, or any other computable entity. It is possible to have proxies for compositions and representations. A proxy is created by an uncomposed constituent, the parent constituent of a composition, or a represented constituent. FIG. 9 is a simplified illustration of the representations used to track programming and associated media used to generate broadcasts. Referring to FIG. 9, a broadcast is said to be composed of elements 310 that are represented (R) by material 315 which are represented by media 320. The elements, materials and media are recursive in nature enabling, for example, element 310 to be composed of (C) element 325. As is readily apparent and as will be illustrated in subsequent examples, a higher level element, material or media can be composed of one or more lower level elements, materials or media, respectively, thereby generating a hierarchy of materials and/or media. As one moves to the right of the illustration, through element 310, material 315 and media 320, more details of the manifestation of the constituent in time and physical attributes are provided. As one moves down the hierarchy illustrated, more details of the formation of the constituent (element 310) are provided. Associations (A) enforce ordering constraints on constituents of the same type. Association may cross multiple levels of composition to connect arbitrary constituents. Associations may also relate constituents that are not in the same compositional structure, enabling multiple structures, perhaps originating from a variety of interconnected computing resources, to generate structures that can be used in conjunction with each other to provide the broadcast functionality. In the preferred embodiment, the system is implemented using object oriented technology. FIG. 10a illustrates the object classes which implement the semantics of the structure described herein and further shows the relationships among instances of those classes. Referring to FIG. 10a, a constituent class is composed of an element class, material class and media class. Constituents relate to one another through mapped relations, associations and proxy relations. Mapped relation includes composition and representation. The composition relation defines the set of subcomponents of a constituent and may provide additional information such as time base of clips and track mappings. For example a commercial break constituent is composed of a determined number of spots or commercials. The representation relation defines how a constituent may be manifest by another constituent. The representation relation is used to manifest the intent of the constituent and can be said to define an equality relationship between the time and track sets. FIG. 10b illustrates the relationships of instantiated objects. In particular, FIG. 10b shows that an element can be represented by material which is represented by media. The representation can be specifically identified at the time the schedule is created or specified using proxy representation wherein a variable object is bounded to a real object at a later point in time. This would be used when alternative programming is provided for selection during the actual broadcast. Each instance of an element object can be composed of other elements, either by proxy composition, wherein the constituent the element is composed of is selected at a later time, or by direct composition. Furthermore, each element object can be associated with other elements, usually in the context of a parent object of which the element is a constituent. FIGS. 10c-10g illustrate a representative object oriented description of the data model used herein. The illustrations are simplified for purposes of explanation and are not meant to convey all of the attributes nor all the methods which might exist in a complete implementation. Such additional attributes and methods would be apparent to one skilled in the art and may be particular to specific implementations. FIG. 10c sets forth the conventions used to depict the various object types in the information structure. These include the following: class name--the name of the object class superclasses--a list of the class from which this class inherits. Multiple inheritance is use, for example, in the definition of proxy relations where a proxy mixing class is defined. In the present illustration the remaining classes singly inherit. methods--this abbreviated definition gives only the name of the methods for an object class without the parameter signature. The intent is that the method name is a key to its functioning, and a strict signature definition will be provided by the implementation. attributes--attributes are listed which are relevant to illustrating the function of the object class. FIGS. 10d, 10e, 10f and 10g illustrate the various objects that may be used in one implementation include a constituent object 1007, element object 1009, material object 1011, media object 1013, constituent relation object 1015, time base map object 1017, track set map object 1019, constituent map object 1021, composition object 1023, representation object 1025, association object 1027, proxy object 1029, proxy composition 1031 and proxy representation 1033. Thus by using the above described structure to specify broadcasts on one or more channels, broadcast operations can be simplified. For example, using the structure, an innovative method for setting up commercials can be used. In particular, a particular commercial break structure, for example a doughnut, can be easily specified numerous times for different spots by reusing the same structure and identifying different material for the different spots to use. Programming in different time zones is also simplified. For example, in an earlier time zone, the broadcast may be live; therefore the media would be identified as a live feed from a particular resource. However, in the later time zone, a tape of the live broadcast is broadcasted. Thus, the element and the material that represents the element is the same as before, but the media is changed to identify a certain tape that is contains the recorded broadcast. A number of illustrations of the flexibility of the system and method of the present invention are described below. It will be apparent to one skilled in the art that these illustrations are exemplary and numerous other situations not described herein can be addressed utilizing the teachings of the present invention. Spots or commercials for upcoming news broadcasts are quite common. Typically, as shown in the compositional structure of FIG. 11a, these spots consist of three elements: a lead in, a teaser and a close. A news spot "News @ 11" is illustrated in FIG. 5a. Implicit in the illustration is the left to right play sequence of elements. FIG. 11b shows the representation of the News @ 11 spot. In this situation, the lead in is a standard news lead in, i.e., the lead in is represented by the material "Std News Lead In". The teaser is a short description of a story "Cat in tree" and the close is a standard news close "Std News Close". It should be noted that the element News @ 11 is not represented by any material; thus the search process for identifying the material which represents the spot must be identified by looking to the constituents which form the spot. Thus to play the News @ 11 spot, the system must assemble the constituent elements and their representation (i.e., materials) and play the individual representations in order. As noted earlier, each constituent has a time value associated with it. Thus the element News @ 11 has a duration identified that corresponds to a scheduled duration (typically used for planning purposes). For example, the News @11 may be planned as a 20 second spot. The as-composed and as-represented durations are derived from, and the values depend on, the respective durations of the component spots which form the spot News @ 11. For example, the as-composed duration is derived from the scheduled durations of the elements Lead-in, Teaser and Close. The as-represented duration is derived from the durations of the representing materials. Typically the values of all three durations are close but are not required to be exactly the same. This example can further be used to show how an element an be repetitively used. A spot, such as News @ 11 will typically be used in a number of schedules. For example, it would be run every day and several times a day. Scheduling this spot consists of composing it into higher order element structures. FIGS. 11c and 11d provide simplified examples. FIG. 11c shows for the schedule "today", at least part of the composition is News @ 11 spots. Similarly, FIG. 5d shows that for the schedule "tomorrow", at least part of the composition is News @ 11 spots. Each mapping of a News @ 11 into another element provides a unique context for the News @ 11 element. There are two direct implications from this mapping requirement. First any structural change made to the News @ 11 element will be directly and immediately reflected in all uses of the element. Second, each unique context can be the context for a representation of News @ 11 for its subordinate elements, implying that in every place it is scheduled to execute, it may have a different representation based on which schedule (e.g., day) it is composed into. Proxy representations can also be used to provide different representations of one or more of the constituents which compose the News @ 11 spot. Unlike specifying representations in the context of a composition, proxy representations enable the specifications of multiple representations in the context of the element it represents. A proxy representation allows an element to maintain multiple possible representations with the choice made at runtime. One or more of the elements, for example, those which compose a higher order element, can have proxy representations. The use of a proxy representation will be illustrated below. The structure enables the change of the composition of the spot to be easily performed. Continuing with the illustration shown in FIG. 5, suppose a news team arrives with footage of a breaking story regarding a flood with more audience value than the cat in the tree story currently scheduled in the News @ 11 spot (see FIG. 11b). An editor calls up the stored footage, or accesses the scheduled structure for the News @ 11 spot (e.g., it is stored on a RAID server) and finds that the spot is composed of three elements. The lead in and close does not change; therefore, the editor needs only to prepare new material to represent the teaser element. From the footage just arrived, a teaser is constructed that is of about the same duration as the scheduled duration of the teaser and stores the new footage on a media and adds to the schedule an additional representation by using a proxy representation. The new scheduling hierarchy is illustrated in FIG. 11e. Thus when an automated broadcast system, which accesses the scheduling hierarchy to determine what to play next, reaches the News @ 11 spot, it sees three representations in the context of the Today playlist available, one for the lead in constituent, one for the teaser constituent and one for the close constituent. For the teaser constituent, however, there is a proxy. Thus the teaser constituent asks the proxy representation execute the selection rules defined to select an appropriate representation, i.e., the flood segment or the cat in the tree segment. The proxy representation contains a selection rule to pick the most recently created representation and therefore returns the new flood clip just created by the editor. It can be seen that because of the knowledge built into the information structure, only one person, the editor, had to take an action to change the execution of the spot. This edit did not involve changing the definition of the structure of News @ 11 so the scheduling department of the broadcast facility was not involved. An as there was no change to the running playlist as the scheduled spot remained the same, the operator did not have to take any action. This example illustrates the benefit to a system that utilizes such a unified structure as it brings together broadcast automation, commercial insertion and editing into a unified operation. The advantages are further illustrated by the repetitive utilization of commonly used commercial pod structures. One commonly used structure is a donut. A donut is a commercial insertion specification between two spots, one of which is composed of two segments. The donut plays the first segment followed by the second (unrelated) spot followed by the second segment. The following illustrates how the same donut structure can be reused using different second spots. To provide this feature, the schedule is set up with the second spot identified as a variable element. FIG. 12a illustrates the compositional pattern of donut1. The pattern specifies that Seg 1 plays followed by one other element unspecified, ?Middle, followed by Seg 2. As noted above, an element is bound to a variable element at runtime. The runtime representation of the variable element is determined by the choice of representation made by the element to which the variable is bound at runtime. The Donut1 element (e.g., pod) can be composed into any schedule just as any other spot. However, the binding of ?Middle is created by an association between ?Middle and the schedule element the broadcast programmer (or traffic system) wishes to place between the two segments Seg 1 and Seg 2. The association specifies that the relationship is equality, therefore the new element can be substituted for ?Middle. The context of the association, which determines the active binding, is the composition of the element the donut is composed into. So this same pattern could be reused in any number of other places with a different binding context in each. FIG. 12b illustrates the association binding. By this pattern, during the execution of Break22, which is composed (C) of Donut1, Donut1 is played, wherein Spot31 plays between Seg 1 and Seg 2 of the donut. Preferably Spot31 is not specifically composed into Break22 which may affect the binding for any representations or element variables contained within Spot31. FIG. 13 illustrates another pattern referred to as a bookend. This example illustrates a reusable pattern which has temporal specification. A bookend is similar to a donut except that the specification states that the first segment plays at the beginning of a commercial break and the second segment terminates the commercial break. Thus there can be any number of spots in between the two segments. The compositional structure shown in FIG. 7 restricts the type of constituent to a pod, as opposed to identifying it as a general element as in the donut case, restricting the use of this pattern to compositions (i.e., schedules) where pods can be contained. This is desirable as the intent is specify the beginning spot and the ending spot of a commercial break, the entity a pod typically represents. It should be noted that there is an association relation defined between the child elements of the pod restricting the possible ordering relations. The structure defines specific interval relations which must exist among the various elements. The "Starts" relation between Seg 1 and the parent Bookend5 means that no other element can come between the beginning of the pod and the beginning of Seg 1. Similarly, the "Finishes" relation between Seg 2 and the parent pod means that nothing can come between the end of Seg 2 and the end of the pod. These restrictions are implied by not enforced in a donut pattern which is typically identified as a pod. In addition, the use of the precedes relation between Seg 1 and Seg 2 means that there is an indeterminate gap between the two which is filled by other spots. Use of the interval relations makes explicit and visible the semantics of the bookend pattern. As in the case of a donut, this allows an automatic broadcast system playing a schedule to retain the semantics and aid the transmission operator when changes are made to the schedule. Although the pod would be organized in the proper sequence, the conventional playlist, for example, would not explicitly indicate that Seg 1 must start the pod and that Seg 2 must end the pod. Thus subsequent changes in the playlist may unknowingly change the sequence and therefore the original structure. This can be particularly problematic with respect to commercial spots as advertisers only play when the play requirements are met. The next example illustrates the scheduling and playing of a live event. The live event is the classic, hard to handle situation in a broadcast. Most current broadcast systems depend on the specification of time. Live events, however, do not have exact time specifications (particularly duration) making it difficult for broadcast operators to manage. An operator will typically find that they must continually update an estimate of an end time of a live event in order to keep the broadcast automation system from improperly cutting off the event or running long and not switching properly. The broadcast automation system that generates the broadcast must have a complex special case set of "ripple rules" for understanding how to propagate time changes to programs following a live event. The present invention enables the handling of live events when there is an incomplete time specification and further enables the handling of downstream events without resorting to special case ripple rules. The present example further shows how to relate the derived start and end relations to SOM/EOM specifications. This example further shows how manual triggers can be treated as just another kind of element within the model and not handled as special attributes of events. FIG. 14 illustrates the composition a schedule, Wed Sched, which includes a live event, ball game, followed by a filler of indeterminate duration followed by an event, ER, with a hard start time of 20:00. The "meets" relation used in this example defines that whatever the time values are, filler starts exactly when the ball game ends (as ballgame meets filler) and that the filler ends exactly when ER starts. Only the ER start time is defined in the schedule. The "finishes" association between the Ball game and the Take element (The take element indicates a manual operation or trigger to begin a broadcast of an something) defines that these two elements complete at the same time. Since we know that the Take element is manifested by a GPI, without showing it here we can correctly infer that the Ball game event is ended by the Take element. From this structure, the broadcast automation system knows that it must maintain Filler in a ready state so that it is ready to be broadcast after the end of the Ball game which occurs when the operator presses the Take button. The system can also trivially update the expected start time of Filler since it is always the current time. The system also knows the ending time of Filler (20:00:00) and can therefore continually update the expected duration of Filler. However, if the Ball game is still running at 20:00:00, the system will switch to ER as it has a defined start time. Although this appears to be a simple illustration, extending this slightly creates situations that prior art time based broadcast systems find difficult. If the Ball game were followed by a series of events each defined to automatically follow the previous one, a conventional broadcast automation system would require an approximate specification of the end time of the Ball game so that it could calculate the start times of each succeeding event. Having calculated these times, as the Ball game goes on, the automation system must continually ripple the time changes through the playlist according to it built in set of ripple rules. These rules must, among other conditions, recognize and differentiate between calculated start times (soft starts) and specified start time (hard starts). The one time calculation of end point sequences using interval relations produces a simple dependency chain which renders ripple rules unnecessary. As is well known in broadcasting, broadcast playlists or news room rundowns often contain alternates which the operator can choose at different points during operation. Often the alternates have different durations. One example where alternates may be used is following a live event. Continuing with the example of Wed Sched composed of the Ball game followed by Filler and ER, Filler is defined to be composed of alternates Alt 1 and Alt 2. This is illustrated in FIG. 15. The alternates are specified by use of proxy compositions. This mechanism enables the resulting end point relations to be used to eliminate possible alternates. In particular, during the course of the live event, alternate playlists are evaluated in view of the time remaining and whether there is enough time to play the alternate before the next event is broadcast at the fixed time. Referring to FIG. 15, the first layer of composition operates exactly as it did in the previous example. However, the definition of the Filler element now provides a different behavior. By using the proxy composition (PC) to define Filler, we have explicitly stated that there are two entirely different ways, Alt 1 and Alt 2, to define Filler and that the choice will be made at runtime. In this situation, the Filler element does not specify a scheduled duration; its as-composed duration will be taken from the duration of the chosen element. The set of interval relations within Filler defines how the two alternatives relate to each other and to their parent Filler. Here Filler and Alt 1 are defined to start at the same moment as specified by "starts". But Alt 1 and Alt 2 have a relation between them even though only one will pay. This disjunctive relation specifies that both will start at the same time, but when one ends bears no relation to when the other would end. This specifies that the two elements may have very different durations. The runtime choice of Alt 1 or Alt 2 creates an equality relationship between Filler and the chosen alternative, defining at runtime that Filler ends whenever that alternative ends. Furthermore, since the end time of Filler is known (20:00:00) by deduction from the hard start of ER), the alternative ends at that time also. The broadcast automation system can therefore be programmed to automatically prune or eliminate alternatives that can no longer be completed in the time left by monitoring the progress of the end of the live event (e.g., Ball game) and updating the expected duration of Filler. By comparing the expected duration of Filler to the known durations of the alternatives, those alternatives that are too long can automatically be eliminated from the available alternatives from which the operator selects from at the end of the Ball game. This concept can also be applied to news program rundowns. News stories in a rundown often have alternate versions which are chosen at runtime. It can readily be seen that alternate versions, typically having different durations can be chosen based on available time left before a commercial break or the hard time specified for a segment, such as weather, or before the end of the program. Not only does the system and method of the present invention enable the scheduling and the broadcasting of schedules programmed, in one embodiment, the system generates "as run" logs reflective of the actual material executed (i.e., broadcast). To generate an as run log, new material and media representations are generated that define exactly what was broadcast. To generate an as run log, the play start and stop functions used to perform broadcast operations of constituents are modified. The play function takes a new argument: a parent representation. The parent representation identifies the material, and therefore through the representation, the media that functions as the parent material for the sibling materials form the as run log material. When an element or material is played a new representation (labeled as a type "as-run") is created, and attached by representation to itself, adding a proxy representation if at least one other representation exists prior. The proxy includes selection rules that ignore the as run representation unless the proxy is executed in one context of generating an as run log. The new representation representative of what is aired is composed into the parent representation (e.g., material media). The start and stop functions are also modified to generate an as run log. When the start function detects the as run constituent, it records the start time (current time or SOM depending upon the media type) and other needed information. For example, the start function of a media for a tape will record which VTR it was played on and the SOM of the tape. The stop function will record the end time (current time or EOM). Just prior to completion of play, the play command records the last recorded end time into the end time field of the parent representation. Thus, creation of an as run log is performed by creating compositions of material and media as a result of executing the specified schedule composition. FIG. 16a shows a simple schedule composition, Sched, composed of Elt 1 and Elt 2, each having basic material and media representations. The as run log creation process starts with the execution of the parent element Sched. The broadcast automation system first creates a new material representation "as-run" for Sched. The new material will by definition be the combined as executed signal. The system also creates a new media representation for the new material. This new media might be a router crosspoint or switcher where the material signal is manifest as a whole unit. This is illustrated in FIG. 16b. At the start of play Elt 1, see FIG. 16c, Elt 1 receives the parent representation As-Run and generates a new material mat 1 and media tape Z. The broadcast automation system creates new compositions (e.g., composition objects) on the As-Run material and the MS 5 media. These compositions map Mat 1 and Tape 1 to their respective parent constituents, As-Run and MS 5. The time base mappings of the compositions reflect the exact SOM/EOM of the child material and media. At the start of Elt 2, see FIG. 10d, the automation system completes the material and media composition structures. As each element finishes, the systems performs any additional accounting processes such as recording the actual EOM values in their respective compositions. FIG. 16d illustrates a completed as run log. To produce a convention as run log from the present structure, the system, e.g., the broadcast automation system, would walk the tree rooted as the as run material constituent. The tree walk can be composed of any search algorithm that visits the desired nodes in the correct order. For example, to list only the individual low level constituents and the media which manifest them, the search visits all the leaf nodes of the material tree and lists the corresponding media nodes. To list the complete structure, the walk would be an in order traversal of all nodes in the tree. As is illustrated in the above examples, the flexibility achieved and the advantages enjoyed using this structure are great. As noted earlier, the structure may be utilized by a variety of systems in the broadcast field to program schedules, edit schedules, edit elements of schedules, execute schedules and generate as run logs of executed schedules. The following discussion provides further discussion of the processes executed by the broadcast systems in accordance with the teachings of the present invention. An exemplary process for generating a schedule is illustrated in FIG. 17. As is shown in the structure of FIG. 8, a channel can be composed of one or more schedules, e.g., a schedule for Monday, a schedule for Tuesday, etc. Thus, at step 1110, the broadcast programmer will specify, preferably with a user friendly graphical user interface or some other input means, that the schedule for a particular day, e.g., Thursday, Octpber 20, is to be created. The programmer then proceeds, at step 1115, to specify the constituent that compose the schedule. See for example, FIG. 18a. The interval relationships, via associations, are specified, step 1120, in accordance with the relative order the programmer wishes the elements to execute to subsequently generate a broadcast. See FIG. 18b in which the associations (A) are simply specified to be that the end of program 1 precedes the beginning pod1, the end of pod 1 precedes the start point of program 2 and the end point of program 2 precedes the start point of program 3. The composition of each constituent, e.g., program 1, pod 1, program 2, program 3, is further specified as needed. The programmer can also specify and addition relations and compositions, including proxy relations and compositions, step 1125, setting the proxy rules for execution when the proxy is subsequently executed during run time. See FIG. 18c which shows that the pod 1 is composed of proxy composition which will select the latest compiled components which have a duration less than or equal to the duration of pod 1. The programmer also can specify where the program elements can be found so the broadcast automation system, when executing a schedule, can identify where the broadcast material should originate from. Thus the programmer specifies the material and media for a particular constituent. See FIG. 18d which shows that program 1 is represented by material "XVII" which is represented by videotape "1200d". The programmer can use the variety of programming techniques described herein to generate a schedule. The system can then perform a check of consistency to determine whether there are any problems with the schedule and the resources needed for the schedule. For example, the system can automatically identify conflicts in the schedule. Furthermore, in view of the media specified, available resources can be determined. A simplified process for generating a schedule is illustrated in FIG. 19. It is noted that FIG. 19 sets forth a simplified version of an exemplary process. It should be readily apparent to one skilled in the art, that a variety of scheduling processes can be used, having varying degrees of complexity and sophistication. Referring to FIG. 19, at step 1205, the highest order or root node for the schedule to be programmed is established. It should be recognized that this process can be used to generate smaller portions of the schedule that are later inserted into a larger entity, such as the schedule for a particular day. Furthermore, this process can be used to generate structures that can be reused in the same or different schedules at different times. If generating a component for insertion into a schedule, the root node can be any type of element, such as those set forth in FIG. 8. However, when generating a schedule for a channel, for example, the root node would typically be identified as the channel which can be composed of one or more schedules, such as a Monday morning schedule. To establish a root node, the parameters for the node are entered into the same, including the name of the element, temporal information and track information. In addition, representations that manifest the node may optionally be identified at this time. Once the root node is established, at step 1310, lower order nodes which compose the root node are entered into the system. As with the root node, the parameters of each node are entered and any corresponding representations can also be entered. Thus, for example, if the root node is a schedule element, a first level of lower order nodes may consist of a plurality of programs. In order that it is clear in what order the nodes which are the composition of a higher order node play, associations are established reflective of the sequence the nodes play, step 1315. Although any temporal or relative technique can be used to indicate the sequence, it is preferred that interval algebra as discussed herein is used. At step 1320 it is determined whether the lower order nodes are further composed of lower order elements. If so, for each lower order node, the elements are identified, parameters provided and associations established, step 1325. This process, steps 1320, 1325 is repetitively performed until the lowest order constituent is identified. At the time the nodes are established, or at a later point, representations are provided identifying material and media which represent the scheduling constituent. FIGS. 20a-20d illustrate the building of a program. Referring to FIG. 20a, the element B5-episode 14 is established and is composed of a number of lower order constituents, such as a prelude 1402, clips 1404, 1406, 1408, Act 11410, clips 1412, 1414, 1416 Act 21418, etc. FIG. 14b shows a more structured representation wherein the clips 1404, 1406, 1408, 1412, 1414, 1416 are representative of commercial spots and are organized into pods P11422 and P21424. Both embodiments shown in FIGS. 20a and 120b are representative of a particular broadcast of the program. The program B5-episode 14 can be structured for repetitive use. In particular, it is desirable to change the clips, e.g., constituents 1404, 1406, 1408, 1412, 1414, 1416, to be selected depending upon when the program is broadcast. FIG. 20c illustrates the structure of program B5-episode 14 for repetitive use. In this example, the pods P1 and P2 are established as a variable pod ?P1 and ?P2. As the pods ?P1 and ?P2 are variable, the program can be instantiated or broadcast multiple times with differing commercial spots. For each broadcast an association is established with the pods ?P1 and ?P2 are respectively to be bound to. The association is one of equality wherein the context, e.g., the day or time zone the program is broadcast, specifies the particular pods the variable constituents are to be bound to. This is illustrated in FIG. 20d which shows two schedules, one for the east coast, EC_Sched and one for the west coast WC_Sched. Both schedules are to broadcast the program B5 episode 14. Depending upon the context which in the present example would be the schedules EC_Sched or WC_Sched, the broadcast automation equipment will select one of pods PA, PB or PC to be bound to variable pod ?P1 and one of pods PA, PB or PC to be bound to variable pod ?P2 and therefore broadcasted. The issue with different time zones and use of a common schedule is further complicated when dealing with a live broadcast in one time zone and a recording of the live broadcast in the later time zone. Proxy representations enable a programmer to schedule the same constituent multiple times with different representations playing at different times. This situation typically occurs in a broadcast when a similar program airs on different days. For example, a program The Evening News is a 1/2 hour news show broadcast every evening at 6 PM. It is very convenient to set up the same element to be scheduled every day. However, each day will provide a different actual show. To address this issue, each day's show is represented by a different material representation of the element coordinated by a proxy representation. A constituent, e.g., a program is created and titled "The Evening News". A proxy representation is generated for The Evening News. the selection criteria programmed into the proxy representation is "pick the latest version". As each day's program is edited, a new material and media is created for the show (representing the show for that day). This new material is attached to the proxy representation by updating the proxy representation object to include the representation in its list of possible representations. When the evening schedule is played and it is determined that The Evening News is to be played, a play command is sent from the broadcast process to the constituent The Evening News. The Evening News, finding that it has a representation, tells the representation to play. The proxy representation in response to the play message, searches its list of material representations and find the latest one created (in accordance with its search criteria). This material representation identified is sent a play message by the proxy representation. The material representation, finding that it has a representation, passes the play message to its media that it is represented by. The media broadcasted is the most current version of The Evening News. This process can be used to update components of a higher order constituent, such as elements, pods, programs and schedules and can be used to affect a subset of components which compose a higher order constituent. Proxy representations can also be used to manage the play out of alternative compositions or playlists. A common situation in broadcast or news programming is the run-time selection of one constituent or event to air from a set of alternative events. For example, in news programming, stories will often have multiple versions of the same news piece, each one edited to a different length. The news piece of the desired length is then chosen during the news broadcast. Furtherm | ||||||