HK1128341B - User interface for searching and displaying legal case histories - Google Patents

Description

User interface for searching and displaying legal case history

Copyright notice

One or more portions of this patent document contain material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or the patent disclosure, as it appears in the patent and trademark office patent file or records, but otherwise reserves all copyright rights whatsoever. The following statements apply to this document: copyright 2005, Thomson Legal and Regulation, Inc.

RELATED APPLICATIONS

This application claims priority to U.S. patent application No.11/370,194 filed on 6/3/2006, which is a continuation of U.S. patent application No.11/182,028 filed on 13/7/2005. This application is related to U.S. patent application No.09/746,557 filed on 12/22/2000, which is a continuation of U.S. patent application No.09/120,170 filed on 21/7/1998. All of these applications are incorporated herein by reference.

Technical Field

Various embodiments of the present invention relate to systems, methods, and software for enabling a user to interact with documents (interfaces), such as judicial opinion or case (case) documents, having complex temporal, semantic, and/or legal relationships.

Background

The legal system of the united states, as well as some other legal systems around the world, relies heavily on written judicial opinions (i.e., written statements by judges) to clarify or explain the laws governing resolution of disputes. Each judicial opinion is not only important to resolving a particular legal dispute, but also to resolving or avoiding similar disputes or cases in the future. For this reason, judges and lawyers within the U.S. legal system are continually researching entities that have expanded past opinions or case law to find those that are most relevant to resolving new disputes.

To facilitate the search, companies such as Thomson legacy regulation, Inc, by st. paul, minnesota (as a Thomson West practice) collect and publish judicial opinions of court in the united states wide electronically and make them available to online researchers through their Westlaw research service. Many of these opinions are published with literature citations or hyperlinks to historically relevant opinions, referred to as previous cases, from other courts who have previously sanctioned all or part of the same dispute.

The references and hyperlinks enable researchers to easily access relevant opinions electronically within the Westlaw research system. For example, opinions in a patent case from the highest court of the United states (i.e., the highest court in the United states) will typically cite not only opinions from the federal go-round court (the next highest court in the patent case), but will also cite opinions from the local federal court that began the patent case, thereby providing a documentation of the history or progress of the case through the U.S. federal judicial system.

Disputes are particularly important through the history or progression of the court architecture, as subsequent cases in the chain may override one or more portions of previous cases. Therefore, judges and lawyers need to understand their history before relying on a decision in any given case. To this end, the Westlaw research system includes a KeyCite feature that not only provides a list of sequential cases that constitute a history of virtually any given case, but also provides a short description of the relationships between cases. For example, a history list with two cases might list the motions (movements) that one case warrants in the other case. Such a list is a powerful tool.

However, the inventors of the present invention have recognized that some case histories are quite complex and difficult to follow in an ordered list. This is particularly true for listing one case more than once, presenting a listing with gaps between cases, or listing a history listing of cases that affect a given previous case.

Thus, the inventors of the present invention have recognized a particular need for better ways to enable users to interact with historically related legal cases.

Brief Description of Drawings

FIG. 1 is a block diagram of a computer system that may be used to automatically generate case law reference data in a legal text processing system according to the present invention;

FIGS. 2A-2D are screen shots illustrating an example of a user interface displayed to a user of a case law referencing system according to the present invention;

FIG. 3 is a diagram illustrating the flow of data in a caselaw citation system according to the present invention;

FIG. 4 is a diagram showing more detail of the data flow in the case law citation system of FIG. 3;

FIG. 5 is a diagram illustrating a quotation identification system according to the present invention as part of the case law citation system;

FIG. 6 is a flow chart illustrating a method for a quote identification and verification process in accordance with the present invention;

FIG. 7 is a flow chart illustrating a method for withdrawal verification according to the present invention;

FIG. 8 is a diagram illustrating a negative history determination process in accordance with the present invention;

FIG. 9 is a flow chart illustrating a method for determining a processing depth for legal cases in accordance with the present invention; and

fig. 10 is a diagram illustrating a method for assigning a subject matter classification according to the present invention.

Description of the exemplary embodiments

The specification describes and describes one or more exemplary embodiments of the present invention, with reference to and in conjunction with the accompanying drawings, and in conjunction with the appended claims. These embodiments, offered not to limit but only to exemplify and teach, are shown and described in sufficient detail to enable those skilled in the art to make and use the invention. Accordingly, the description may omit certain information known to those skilled in the relevant art, where appropriate, to avoid obscuring the present invention.

Some embodiments of the invention are particularly applicable to computer-implemented legal document processing systems and methods for semi-automatically identifying features (such as citations and quotations) within a legal document and identifying relationships between the legal document and other legal documents stored in a database. The legal document may be a legal case, statute, legal review article, ALR article, or legal paper. The exemplary embodiments are described in the context of legal cases. However, it should be recognized that the system and method according to the present invention has broader utility and may be used for different legal documents, such as statutes, legal histories, and administrative procedures and patents. Some embodiments apply the teachings herein to non-legal documents, such as scientific documents.

Before describing a preferred embodiment of the present invention, a brief description of the terms used to describe the present invention will be provided. Any reporting decision of a legal case is assumed to be an authoritative statement of the law as it was written. Later events may then affect the authority of the legal case's decisions. These later events may include later programs or written decisions during the same litigation (e.g., direct history), decisions from later legal cases that solve the same problem in a different way or using different reasoning and that overrule different litigation of earlier cases, or decisions from later legal cases that solve the same problem differently but do not explicitly overrule different litigation of the cases. The direct history of legal cases may include records of connections between legal cases that are part of the same litigation. The direct history may have different degrees of relevance and may include a positive history (i.e., the authority that maintains or supports the legal case) or a negative history (e.g., the legal case may no longer have all of its authority once). An indirect history of a case is a record of the connection between a legal case and other legal cases that are not part of the same litigation. The indirect history of legal cases may be positive or negative. The importance of a particular case can often be represented by the number of discussions (i.e., the number of texts) that the case uses later in discussing the decision of another legal case (which may follow, override, or explain the case). This is referred to as the processing depth for this case, as described below. One or more of the embodiments described herein may also be implemented on a system such as that described in co-pending U.S. patent application 10/751,269, filed 12/30/2003, which is hereby incorporated by reference.

FIG. 1 is a block diagram of a computer system 30 in which the present invention may be embodied. The system can semi-automatically identify features (such as citations and quotations) within legal case documents and then generate information about the legal case in the context of other legal cases. The computer system may include a computer 32, a server 34, and a plurality of client computers 36. The computer 32 may also include a Central Processing Unit (CPU)38, memory 40, and one or more processes 42, which may be software applications stored in the memory 40. The CPU controls the operation of the computer and executes software applications stored in the memory. In operation, a plurality of items of electronic data corresponding to the text of the published decisions of the legal case are fed into the computer and temporarily stored in the memory 40. In the discussion that follows, the written opinions of the legal case are referred to as legal cases. Each electronic data (i.e., each written opinion of a legal case) may be automatically processed by the CPU using the processes contained within the software applications contained in the memory to generate information about the legal case, as described below. For example, the CPU may perform the following operations: parsing the text of the legal case to identify candidate (i.e., unverified) citations to other legal cases and marking those citations for later processing; identifying candidate (i.e., unverified) quotations in the text of the legal case and labeling the text accordingly; verifying a source of a quotation in the text of the legal case; determining the processing depth of the cited legal case (i.e., determining the importance of the cited legal case based on some predetermined criteria); determining a negative treatment of the legal case; and assigning subject matter text (such as headnotes) to the citations in the legal case according to a predetermined classification system. Each of the processes may be performed by a software application in memory 40 that is executed by CPU 38. The details of each of the processes will be described below.

Once processing is complete by process 42, computer 32 outputs a data record 44 corresponding to the particular legal case, which contains: information about the history of the legal case; information about the processing depth of each citation of the legal case; information about various quotations of the legal case; and information about the subject matter text (i.e., endorsements) assigned to each citation in the legal case. The data records generated by the computer 32 for each legal case may be stored in a database 33 within the server 32. Subsequently, the user of one of the plurality of client computers 36 requests information about one legal case, the server 34 generates a user interface containing various information about the requested legal case based on the data records stored in the database 33, and presents various information about the legal case to the user who refers to the legal case.

One example of a user interface provided to a user of each client computer is described below with reference to fig. 2A-2D. In this way, a user of the client computer may request data regarding a particular case and the system according to the invention provides the data to the user.

When the computer 32 receives electronic data corresponding to the text of each written opinion of a new legal case, the legal case is processed as described above, and the results of the processing are stored as data records 44 in the database 33 within the server 34. The user of each client computer may then retrieve data from the server 34 regarding the particular legal case. Thus, while server 34 provides data regarding legal cases to one or more users of client computers, computer 32 may simultaneously process additional new legal cases and add information corresponding to the new legal cases to database 33 within server 34. One example of a preferred user interface and information provided to the client computer is described in more detail below.

2A-2D are screen shots illustrating examples of preferred user interfaces and information provided to and displayed by a client computer in accordance with the present invention. FIG. 2A shows a computer screen 50 on a client computer that is displaying legal cases that are reviewed by a user of that particular client computer, where the user interface has a Windows format, a toolbar, a drop-down menu, and the like. In this example, the display is text of a legal case called Pleasant v.celli, which was awarded by the law of upper prosecution in california. As described above, any application to a legal case has a well-defined format, thereby facilitating the identification of such citations within the text of the written opinions of the legal case. To access more information about the displayed legal case, the user of the client computer may select a citation service, referred to as KeyCite, from a service menu 51 by clicking on a "KC" button 52 or clicking on a symbol 54^TM。KeyCite^TMIs a trademark of the assignee of the present invention (citor). The symbol may be a colored symbol (e.g., flag) that gives the quickness of the legal case. As described below, a red flag may alert that at least some portion of the legal case being reviewed may not be a good law, a yellow flag may indicate that the legal case has some negative history, or another color symbol (such as blue H) may indicate that the legal case has some non-negative history. However, the present invention is not limited to any particular type of symbol or color. Once the user of the client computer has selected the referrer system in some manner, the screen shown in FIG. 2B may be displayed.

FIG. 2B is a screen shot showing one example of a computer screen 50 that may be employed with the present invention, having a control interface portion 58 and a display portion 60. The control interface portion of the display allows the user to customize the information displayed. For example, if the first radio button 62 is selected, the full history of the legal case may be displayed, including negative or positive direct history, negative indirect history, and any relevant references. If the second radio button 64 is selected, only negative direct and indirect histories may be displayed. If the third radio button 66 is selected, only the direct history of the legal case may be displayed, leaving no real secondary direct history (including references), distant direct history (such as complaints after a review) and somewhat negative indirect history. The displayed control section 58 may also indicate the number of cases that are considered to be a history of the legal case. The control section 58 may also include a fourth radio button 68 and an indication of the number of citations for the legal case displayed. When the fourth radio button is selected, a list of other documents is displayed.

In the example shown in fig. 2B, the complete history of the Pleasant v.celli case is shown. As shown, various types of histories, such as a direct history and a negative indirect history, which are separated from each other by a title, are displayed in the display section 60. For each history, a short description or label of the history may indicate a relationship between the listed case and the source case, such as "opinion is invalidated by …", "not agreed by …", or "divergent from …". In this example, the earlier decisions of the same court were invalidated by the Pleasant case. (the appendix shows other exemplary interfaces that may be used in conjunction with the embodiments described by fig. 2B and/or described elsewhere in this specification.) reference to such legal case will now be described with reference to fig. 2C.

Fig. 2C is an example of a screen shot showing a computer screen 50 having a control interface portion 58 and a display portion 60. The screen displays legal cases that reference the legal case currently being referred to (i.e., Pleasant v.celli in this example). In this screen shot, the fourth radio button 68 is selected. Thus, the control section may also have a button 70 that allows the user of the system to limit the type of references displayed, as described below with reference to FIG. 2D. The display portion 60 may also display quotation marks symbols 72 and treatment depth symbols 74 associated with citations to the legal case that are cited to the legal case of interest. Quotation marks 72 indicate that the cited legal case is directly abstracted from the case of interest (i.e., in this case Lubner v.City of Los Angeles contains an abstraction from Pleasant v.Celli). A method according to the invention for identifying a withdrawal and verifying the origin of said withdrawal will be described below. The treatment depth symbol 74 may be, for example, one or more stars, where the number of stars indicates the degree of treatment of the written opinions of the legal case, such as the amount of text in the cited case opinions for the case of interest. The details of the process depth assignment process will be described in more detail below. A screen that allows the user to restrict references displayed in the display section will now be described with reference to fig. 2D.

Fig. 2D is an example of a screen shot showing a computer screen 50 having a control portion 58 and a display portion 60. In this screen shot, it is assumed that the user of the system selects the restrict application button 70 shown in FIG. 2C. As shown, the user of the system can restrict the displayed application based on headnotes or topics, and the system will evaluate all citations with reference to the selected headnotes or topics, so that only legal cases containing the selected headnotes or topics are displayed in the screen shot shown in fig. 2C. An annotation can be a few sentences/paragraphs at the beginning of a legal case that indicate a summary of the law for a particular part of the legal case. The user interface of the system allows researchers to quickly and efficiently perform validation and collocation functions on legal cases. Details of a system for generating information about the legal case and providing the verification and collocation function according to the present invention will now be described.

FIG. 3 is a diagram illustrating a method 100 according to the present invention that may be implemented on the computer system of FIG. 1 for processing legal cases to generate information about the legal cases that may be used for verification and collocation functions. To aid in understanding the process, the actions of a single legal case will be described below. However, it should be understood that multiple legal cases may be processed simultaneously, as each legal case may be at a different point in the process. An electronic version 102 of the text of the legal case (referred to herein as "WLLOAD") is fed into a citation identification process 104 (acid) that identifies candidate citations to other legal cases and other legal materials within the text of the legal case and labels the text (i.e., adds feature labels to the text) so that the citations can be easily identified later. An example of a marker symbol may be the combination of symbols "% v" placed at the beginning and end of the reference. Which identifies the application for later processing.

Briefly, the citation identification process identifies candidate citations by identifying specific text patterns in legal documents and comparing these patterns to a predetermined set of reference patterns. In particular, the numbers in the text may be recognized first. Next, the text is scanned for abbreviations adjacent to the numbers, which correspond to known bookmarker abbreviations, such as "cal. Once a piece of text in a particular format and punctuation is identified for a candidate reference, the case control database 124 is queried to determine whether the identified candidate reference corresponds to a valid reference in the case control database. If the identified candidate citation matches a citation in the case control database, a second pass is performed. If a match is not located, the identified candidate reference may be flagged for later manual review. As described above, each item reference has a predetermined format. The format may be < case name > < volume number > < bookmarker name abbreviation > < serial number (if more than one) > < page number in volume >. For example, in "18 cal.app.4th 841", the "cal.app.4th" refers to the "california complaint" bookkeeper 4th sequence; "18" refers to volume 18; "841" refers to page 841, which is the case decision starting in cal.app.4th volume 18.

An example of a reference to a legal case is "Pleasant v.Celli, 18 Cal.App.4th841, 22 Cal.Rptr2d 663 (1993)", where the first name section (i.e., Pleasant v.Celli) represents the parties to the legal case; the second bookmarker section (i.e., 18 Cal app.4th841 and cal.rptr2d 663) identifies the bookmarker itself as having a particular characteristic format as described above.

Once the text corresponding to the bookmarker name is located, the text adjacent to the bookmarker name is analyzed to identify the volume number, serial number, and page number of the reference and the year of the opinion published. Once this information is found, the candidate references are identified and marked as described above to identify them as references. The citation identification process may use a two-pass process in which full format citations such as "Pleasant v.Celli, 18 Cal.App.4th841, 22 Cal.Rptr2d 663 (1993)" are first identified, matched to the case control database, and placed in a table. In a second pass of the legal case, a citation abbreviation such as Pleasant may be identified based on the full application text contained in the table. It should be noted that these reference abbreviations cannot be automatically identified without first identifying each full reference. For in-doubt citation abbreviations that do not match the table, tentative identification may be performed.

Citation identification process 104 in fig. 3 outputs file 106 containing text in which any cited legal case is marked. The file 106 may then be fed to a quotation identification process 108 (IQUET) in which the text of the legal case is parsed and various quotations in the text of the legal case are identified and marked, in addition to identifying possible sources of the quotations. At this point, the marked withdrawal has not been verified. The quotes are only candidate quotes that must be further processed for validation. Details of the quote identification process will be described below with reference to fig. 4-6. The quotation identification process can output a file 110 containing text of legal cases in which citations and quotations are marked. At this point, the text of the legal case with the quote and quote tags is stored in a database for later use and may also be fed into several processes. These processes may include a quote validation process 112, a deep processing process 114, and a negative processing process 116. As shown, these processes may be performed in parallel on the same document because the information generated by each process about the legal case is separate and independent from the information generated by the other processes. Each of these processes will be described in more detail below with reference to fig. 7, 9 and 8, respectively.

Generally, the quotation verification process 112 verifies that the candidate quotations identified by the quotation identification process 108 do come from the source (i.e., the cited cases) by comparing the candidate quotations in the cited cases with the quotations in the cited cases. The process then generates a data record 118 containing information about the verified quote. The deep processing process 114 uses the information generated by the system, including the verified quotation, to generate deep processed information such as the number of occurrences of the reference and the characteristics of the reference based on its location (e.g., whether it is independent, at the very front of the string or inside the string). The process then generates a data record 120 containing information about the processing depth information applied to each reference in the case of interest. The negative handling process 116 generates information about any negative handling of the case of interest received by any of the cited cases and manually updates the database 124 containing information about each legal case handled in step 122 to reflect the negative handling. As described below, the data records 118, 120 from the quote verification process and the deep processing process, respectively, may be combined with an annotation assignment data record 128(HNRESULT) by a grouper process 126 to generate a single data record containing deep processing information, quote information, and annotation assignment information about the processed legal case. This single data record may then be used to generate information that is displayed to the user on a computer screen, as shown in fig. 2A-2D.

The data records 118 containing information about the validated quotations in the legal case can also be fed into a citation location identification process 130, which attempts to identify supporting text around the quotations and citations in the legal case, in order to generate a citation location data record 132. The quote location data record may then be input into a topic assignment and thresholding process 134 that matches the words or phrases in the quote to one or more headnotes or topics, and then determines which headnotes to select based on a threshold, as described below with reference to FIG. 10. The subject matter assignment and thresholding process 134 outputs a data record 128(HNRESULT) containing selected subject matter text (e.g., annotations) that is fed into the grouper 126, as described above. Thus, the system according to the invention automatically generates information about legal cases and then provides this information to the user using the system using a graphical user interface when requested. The user can quickly and efficiently locate various information about the legal case from a single source, such as citation information, processing depth information, negative processing information, and subject matter text (e.g., endorsements). Further details regarding the system will be described below with reference to fig. 4.

Fig. 4 is a diagram showing more details of the quote identification process 108, the quote verification process 112, the deep processing process 114, the negative processing process 116, the reference location identification process 130, and the topic assignment and thresholding process 134 of fig. 3. As shown, the output from each of the processes is fed into a system information database 33, as described above.

The quotation identification process 108 uses the file containing the text of the legal case in which the citation is marked to identify and mark quotations as described above. The text of the legal case contains unverified quotations and a file 144 containing verified quotations is stored in the database 33. The output of the quote identification process is a plurality of data records, where each data record has an identified quote and a possible source of the quote. The output of the quotation identification process can be combined with the file containing the text of the legal case and the marked-up citations to produce a file with marked-up citations and quotations 110 which is used as input to the deep processing process 114, the negative processing process 116, the location identification process 130 and the subject matter assignment and thresholding process 134. During the quote identification process 108, several processes are performed, as described in more detail below with reference to fig. 5 and 6. First, candidate quotations in the text file are identified by scanning the text to identify symbols, such as quotation marks, which indicate the beginning or end of a quotation. Next, the beginning and end of the identified reference are marked with a reference identifier symbol ("% q"). Finally, possible sources of the quotations are heuristically identified, such as legal cases or other legal materials from which the quotations originate. The source of the quote is then verified during a quote verification process 112, as described below. The output of the extraction identification process 108 may include a Qdata file 140 containing information about each extraction that is subsequently verified with reference to a possible source of the extraction, and a Qtxt file 142 containing the actual text of the extraction.

The Qdata file 140 and the Qtxt file 142 are then fed into a quote validation process 112 that uses an electronic database of legal cases already available to find and validate the likely source of each quote found by the quote identification process. For each quote, the possible sources of the quote are retrieved. Next, the quotations identified by the quotation identification process are matched to the text of the possible sources to locate the text in the sources corresponding to the quotations. This verifies the source as the origin of the quote. For each quote having a verified origin, a data record 144 containing the verified quote for a certain legal case is stored in the database 33. Subsequently, when a legal case containing a verified quotation is displayed as a reference to a legal case, the reference will contain a quotation symbol indicating that the legal case has a verified quotation, as described above. The deep processing procedure will now be described.

The depth processing process 114 may receive the document 110 containing legal case text in which citations and quotations are marked and in step 146 the depth processing process performs several processes to determine the importance of the citations according to a set of predetermined criteria related in some way to importance. These criteria may be the number of times the citation appears in the legal case, the type of citation, and the association of the verified quotation with the citation. First, the deep processing process reads the entire file 110 and identifies references that have been previously tagged by the reference identification process. For each identified reference, the type of the reference is determined to be a normal reference, the middle of a string reference, or the top of a string reference. Common citations are typical citations that are usually found within legal cases and do not have other citations next to them. A mid-string (internal) reference is a reference that appears in the middle or at the end of a string reference where a series of legal documents are referenced together in a sentence or paragraph. Internal references are generally perceived by the user as contributing less to the depth at which the referenced case is discussed. The string top reference is the reference that appears at the beginning of the string application, which is perceived as contributing more to the depth, since the most relevant reference is usually placed at the top of the string reference. The deep processing process may also identify the legal case page number for all available pages on which the reference appears, so that deep records are written as many times as page breaks appear in the legal case.

Information about each citation in a legal case, such as the total number of times the citation appears in the legal case document, the type of each citation, and the page number corresponding to each citation appears, is output in a file 148, which is stored in database 33. The above information can be used to generate the "quote to case" section and process depth symbols described above, in addition to any verified quote associated with any occurrence of the quote. Techniques for generating the processed depth symbols are described in more detail below.

The negative handling process 116 may include an automatic handling step 150 and a manual validation step 152 that generate a list of negative histories (i.e., other written opinions from other legal cases that are divergent from or otherwise invalid for the current legal case) corresponding to the legal case. During the automatic processing step 150, the document containing the legal text in which the references and quotations are marked is scanned in order to identify stems of certain words, such as "invalid", "retrieved", "disapprove" or "distingguish", which may indicate negative processing. By way of illustration, a process is described for identifying the root word of the word "invalid" in the text of legal cases. When an instance of the root word "invalid" is identified, as described below, a set of heuristic rules is applied to determine whether the sentence containing the identified root word is actually invalid, as described below with reference to FIG. 8. Subsequently, during the manual verification process 152, the operator of the system verifies the results of the automatic process and adds the actual verified invalidity to the case control database 124. The operator may also identify other negative histories with respect to legal cases that cannot be easily identified automatically, as described below. The negative handling process helps the operator to quickly identify the inefficiency. These inefficiencies are a negative history of authority affecting reasoning of the legal case.

The location identification process 130 uses the file containing legal case text in which citations and quotations are marked and the file containing verified quotations 144, which identifies any marked citations and applies a set of heuristic rules to identify and select a portion of text from the vicinity of each citation that may indicate the text supported by the citation, as described below. If a reference appears multiple times in a legal case, the surrounding text corresponding to each occurrence of the reference is combined. Furthermore, if the quote verification process has verified any quotes associated with the application as described above, the text of the verified quote is also combined with other text surrounding the quote. All of the recognized text surrounding each reference may then be used to determine one or more headnotes or subject matter headings that may be applicable to the reference. The subject matter title classifies the citation based on a predetermined number of subject matter areas, such as intellectual property or patents. The process 154(Headqf) reads all of the identified text adjacent to the given reference and generates a natural language search query to search an existing database for matches to the identified text, as described below. The natural language query process is generally described in U.S. patent nos. 5,265,065 and 5,418,948, which are assigned to the same assignee as the present application and are incorporated herein by reference. The Headqf process 154 generates a file 156 containing the natural language query. Using the natural language query, the subject matter distribution process step 158 performs the natural language query against an annotation database to identify subject matter titles (such as annotations) that may match the text surrounding the reference. For each matching subject matter title, the query also generates a confidence score value indicating how well the subject matter title matches the text. A predetermined number of the most closely related subject matter titles and their confidence scores are provided to the thresholding process step 160.

The thresholding step uses the identified subject matter titles and performs various calculations that take into account the ordering of the subject matter titles, the confidence scores of the subject matter titles, and the number of references referencing the subject matter titles. After the calculation is performed, a predetermined number of top annotation hits and a flag corresponding to each annotation indicating whether the annotation passed the thresholding are stored in the database 33 together with a link to the reference. These subject matter titles allow for the classification and searching of citations in legal cases using these subject matter titles, as described above with reference to FIG. 2D. The quote identification process will now be described in more detail.

Fig. 5 and 6 are diagrams illustrating more details regarding the quote identification process 108 according to the present invention. The quotation identification process 108 can include a vocabulary scanner process 170, a paragraph buffer 172, and a main loop process 174 for receiving text of legal cases and automatically generating a file containing each quotation identified and the possible sources corresponding to each quotation. The vocabulary scanner 170 breaks down the document into logical segments (called glyphs) and then uses these glyphs to identify the quotations by the main loop process 174. The glyphs recognized by the vocabulary scanner may include capitalized words, punctuation marks that may end a sentence, white spaces such as one or more spaces, case names, footnote references, the beginning of quotation marks, and the end of quotation marks. The vocabulary scanner process used may be based on any of a number of commercially available software applications, such as the application known as FLEX, available from Sun Microsystems Inc., mountain View, CA. The vocabulary scanner accepts a pattern specifying a grammar and recognizes an action when a particular pattern is located. In particular, the lexical scanner according to the present invention may divide legal cases into paragraphs of a particular type based on a predetermined set of criteria (such as a rule set): 1) paragraphs that may contain a quote; 2) a paragraph extracted as a predetermined block; 3) paragraphs that contain important information about the document (such as the beginning of the document, the serial number of the document, or the end of the document); and 4) paragraphs that are not meaningful to the quote identification process, such as headnotes, headings, and the like. A variety of different criteria and rules may be used to identify these paragraphs.

An example of a rule set that may be used by the present invention will now be described. The rule set may include the following rules: a rule that identifies and stores in the paragraph buffer a paragraph that does not contain any excerpts, where the paragraph is overwritten by a next paragraph; and rules corresponding to paragraphs that have possible excerpts, wherein the vocabulary scanner returns a label to the main loop indicating that the paragraph is a plain text paragraph, a predetermined block excerpt paragraph, or indicating that the excerpt text appears in footnotes. Once the type of the passage is determined, the vocabulary scanner processes the text within the passage in the same manner to identify any glyphs in each passage.

Within each paragraph, the vocabulary scanner may identify the following glyphs: capitalized words, non-capitalized words, numeric strings, abbreviations, proper names (i.e., "mr. smith"), case citations, chapter references (i.e., "Section 150"), case names (i.e., "Roe v. wade"), embedded references, any period end punctuation, any other punctuation characters, a colon, semicolon, or comma followed by a space, a single or multiple blank space characters, the beginning of a quote, the end of a quote, a footnote, left and right parentheses, left and right curly braces, a mark corresponding to a quote, and a mark corresponding to an embedded reference. More details on the operation and modification of the FLEX software application are available from Sun Microsystems Inc. reference Manual, Programmer's Overview Utilities and Libraries, Chapter9, pp, 203-.

Paragraph buffer 172 is where the glyph for the paragraph most recently scanned by the vocabulary scanner is stored prior to processing by main loop 174, which is then written out to the output file if a disjunct is identified in the paragraph. The main loop 174 may decide what action to take for each glyph returned by the vocabulary scanner, manage the paragraph buffer, and decide when to discard data corresponding to a previous paragraph from the paragraph buffer, link several physical paragraphs together into one virtual paragraph for a abstraction that exists over several physical paragraphs, determine where within a paragraph breaks between sentences occur, and decide when to process a virtual paragraph through a set of heuristic rules, as will be described below.

Fig. 6 is a flow diagram of a quote identification process 108 in accordance with the present invention. In step 180, the legal case text is scanned paragraph by paragraph and, for each paragraph, the sentences and glyphs in the paragraph are identified. In step 182, a set of heuristic rules is applied to each glyph in the paragraph to determine whether a quotation is identified. One of the most important functions of the vocabulary scanner and the quotation recognition procedure is to recognize the beginning and end of a quotation. This is difficult because each writer may use slightly different formats for the beginning and end of the quote. Therefore, several rules are required to identify the beginning and end of a reference. One example of a heuristic rule set that may be applied to achieve such identification is described below. These rules may use the vocabulary scanner to identify traditional beginning punctuation marks (such as "or"), to identify traditional ending delimiters (such as "or"), or to identify ending/beginning symbols in longer strings. For example, a rule may attempt to identify strings in which a traditional quote tail symbol is embedded in a sentence. For each of these rules, the characters surrounding the glyph may be checked to ensure that the glyph is actually the beginning or end of a quote.

Once the rules have been applied to each glyph in a paragraph, the quotation identification process determines in step 184 whether there is another paragraph and loops to step 180 to process the new paragraph. Once all paragraphs have been analyzed, the process outputs a data record containing the identified quotations and possible sources of those quotations in step 186. The quote verification process will now be described.

Fig. 7 is a flow chart illustrating a method 112 for verifying a quote in accordance with the present invention. In step 200, the quotation verification process reads the text string identified as a quotation by the quotation identification process 108 and identifies separators (when present) from a predetermined set of separators in the text string. The separators may include ellipses, bracketed expressions, and stop phrases. The stop phrases include various legal phrases that do not help in identifying the source of the quotation, as well as other phrases such as "position(s) updated", "sic", "ephelis provided", and the like. When the separator is present, the separator is used to parse the text string into segments, where each segment includes content that appears between a pair of separators. In step 202, the text string is parsed to determine its length, since the minimum verifiable abstraction length may be, for example, 6 non-stop words, where a stop word is a word without content, such as articles and prepositions. The text string is also parsed to compress any words containing ellipses or other punctuation marks, such as "T ] hen". The parsed quoted text strings fall within one of two different categories: (1) a text string having a single segment; or (2) a text string having a plurality of segments. Thus, in step 204, the system determines whether the text string has a single segment. If the text string has a single segment, then in step 205, a set normalized Inverse Document Frequency (IDF) for each term (word) in the single segment of the text string is determined. The document frequency value represents the frequency of a particular item in a typical document collection, while the IDF is equal to the reciprocal of the document frequency (i.e., 1/document frequency), or in other words, the rarity of an item in the document collection. In a preferred embodiment, if the number of occurrences of a word is greater than 0, the set normalized Inverse Document Frequency (IDF) can be calculated as follows:

wherein Doc _ Occurents is the number of documents in which the given item resides, and Collection _ Docs is the total number of files in the Collection. The IDF is used for the purpose of determining good items for matching, since a rare word may often be unique and provide a good indication that the quotation is from a candidate source.

Once the IDF is computed for each term, a selected number (e.g., 6) of the terms having the highest IDF value below a selected threshold may be sorted by IDF value (step 206) and placed into a "template" (i.e., storage array) (step 207) indicating the position of each term in the text string. Any terms with an abnormally high IDF value (e.g., greater than 0.80) are not used because such infrequently occurring terms are often misspelled words. If several terms have the same IDF value, the alphanumeric order of the terms may be used as a secondary key to ordering the terms for the template. If equivalent terms (e.g., terms with the same IDF value and alphanumeric spelling) still exist, the position of the terms in the text string can be used as a third key to rank the terms in the template. The template may then be compared to the excerpts from the candidate source documents to determine if an exact match has occurred based on the locations of the high IDF entries in step 208. If an exact match occurs, the validated quote is output in step 210 and fed into the database, as described above. If an exact match does not occur, a verify match failure message is generated in step 212 and the quote is not stored in the database.

In step 204, if The text string has multiple segments (i.e., contains one or more separator items in The text string, such as "The roof all in. A selected number (e.g., 4) of entries within each segment having the highest IDF value below the threshold are then sorted by IDF (step 215) and placed into a template (step 216) to determine the position of the entries in the segment for matching purposes (step 208). For text strings with more than 4 segments, the first two and last two segments may be used to match against the candidate source documents (step 217-218). In this way, the quotations identified by the automatic quotation identification process are automatically verified and any verified quotation is identified by quotation marks, as described above. The negative treatment process according to the present invention will now be described.

FIG. 8 is a diagram illustrating a negative process for determining legal cases according to the present invention. The document 110 containing the text of the legal case in which the quotation and citation is marked is input to an automatic negative handling process 150. The automatic negative handling procedure may perform the following operations: 1) identifying the occurrence of the stem "invalid" in a legal case; 2) determining a proximity of the stem to the reference; and 3) excluding any bad legal case. Before identifying the stem "invalid", the case control database 124 may be checked and the automatic process stops if there has been any history corresponding to the legal case. To identify the occurrence of the stem "overrule", the text of the legal case is scanned and the verb tense of any occurrence of the stem is determined. The verb tense of the stem indicates whether the current case is invalid for the previous case or some other type of invalidation. A heuristic rule set may look for a particular verb tense, and then take an action based on the verb tense.

One example of a rule set used will now be described, but the invention is not limited to any particular rule set. For example, one rule may find "overrule" or "overrules" in a sentence and scan back for up to 4 words. If "not" or "nover" is found, the sentence is discarded because it does not refer to any actual invalidation. If "we" is found, the sentence is added to a list of possible invalidities, which is examined by a human. If no one of the phrases is found during the backward scan, the sentence is also added to the list.

Another rule may find "overrule" and then scan back up to 5 words in an attempt to find non-case words that would indicate that something other than the legal case is invalidated, so that the sentence is not added to the list. Several examples of these non-case words include "request", "motion", "object", "close", and "verdict". If the rule finds "point" or "points," the sentence may be scanned forward to the end of the sentence, and if "case," "cases," or "supra" is found, the state of the sentence is unknown and passed to a human inspector.

Another rule may find "overrule" and scan backward or forward and reject or accept possible invalidity based on other words immediately adjacent to the word "overrule" because these additional words will provide context in which to use the word "overrule". For example, once "over rule" is found, the 4 words preceding the word may be scanned, and the following actions taken when the following words are found: 1) if "we" is found and the word before "we" is "that," we is ignored (discussed only with respect to invalidity), but if the word "that" is not found, the sentence is potentially invalid: 2) if the verb is modified by a word representing uncertainty, such as "rateher", "might", etc., the sentence is rejected because the court may simply indicate that it may invalidate the case; 3) if any word indicates that the discussion is invalid, the sentence is rejected; 4) if a word indicates that another person has performed the invalidation, the sentence is rejected; and 5) if "will" or "should" is found, the process looks back at 5 words to find the front word in order to accept the sentence. There may also be a similar set of rules for the verb "overrules", the indefinite form of the verb, and the passive morphism of the verb.

Another rule set may look for words that indicate whether the discussion is about to be invalidated, whether the court has an invalid authority, or whether the past is invalid, which sentences are rejected as not containing the actual invalidity. Another rule set may reject sentences that indicate that others are implementing an invalid (i.e., another court in the past). Another rule set may look for "overruling" and then reject or accept the sentence based on the sentence around the word, as described above.

There may also be other rules that look for specific characteristics of a sentence independent of the verb "overlap". For example, if the phrase "cure: "at the beginning of a sentence, it indicates a direct quote from a judge, and the sentence is acceptable. If the word "Congress" is at the beginning of a sentence, it may indicate that the Congress statute is invalid or that the Congress itself is invalidating an instance, and the sentence may be rejected. If the word "circular" is found in the sentence near the word "oval," the sentence may be accepted to capture unusual languages, such as "bega determination of overlapping words an intent of this country, and it wa circular to all active judges," which cannot be automatically recognized in some other way. Another rule may look for "overrule" within the extracted string and reject the sentence because it is usually an invalidity made by another court to the case extracted by the current court.

In addition to the stem "over rule", other synonyms may be searched and identified. For example, the rules may also detect the stem of words "abrogat" for the california case, which uses the term "abrogated", and the rules may also detect the phrase "retrieved from" for the florida case, as these terms are used to indicate invalidity within each corresponding state. These verb tense rules may be applied in any order, and the present invention is not limited to any particular set of rules or any particular order of execution of the rules.

The output of the verb tense rule set from the automatic negative handling process is a list of possible invalidations. Proximity rules are then applied for each possible invalidation to determine whether the invalidation applies to a particular legal case. For example, the proximity rule may remove possible invalidity if: the sentence containing the stem contains no reference, the preceding or following sentence contains no reference, or the sentence with the stem "override" contains no words such as "case", "opinion", "holding", "precedent", its plural form or "progeny" or "v.", "ex rel", "ex part" or "supra" which are used to refer to the case. Any sentence that contains the stem "overrul" and satisfies the proximity rule is added to the implied invalid list 222 of the legal case. These implied invalidations are then reviewed and checked in a manual review process step 152 by a human. During the manual review process, one also determines cases that are invalidated and the determination data is entered into case control database 124, which tracks legal cases within the legal case database.

According to another aspect of this negative handling procedure, the automatic procedure may also identify relationships other than those that are invalidated, such as "distinguishing" or "about" in legal cases, by extending the method to languages that characterize the other relationships. In short, the negative processing procedure helps the reviewer determine possible invalidity in the legal case by automatically determining possible locations of invalidity in the legal case, thereby greatly reducing the amount of text that must be actually reviewed by the human. Thus, the negative handling process increases the speed at which invalidities in legal cases can be identified and added to the negative history of the legal cases. The depth processing will now be described in more detail.

FIG. 9 is a flow chart of a depth processing procedure 114 according to the present invention in which a depth processing symbol is assigned to each citation in a legal case so that a person using the system can quickly determine the amount of text that specifically discusses a particular citation. This information can be used as a representation of the relevance of the quote, since courts will use more text and discussion for highly relevant quotes.

In step 230, the document with the text of the legal case and the marked-out citations and quotations is received by the deep processing process, as described above. In step 223, the deep processing process identifies the citation in the legal case and then determines the citation type in step 234. Each citation in this legal case may be: 1) a reference at the very front of a string reference; 2) no other references that accompany the reference; 3) references that are internal to the string reference; or 4) formal history citation (i.e., citation in the context of this document solely as an auxiliary history reference to one of the cases cited in its own right). The importance of each of these reference types is different. For example, references that are either singly or at the forefront of a string reference tend to be more important than references that are in the middle of the string.

The depth processing next determines whether there are any additional citations in the legal case in step 236 and loops back to step 232 to process the next citation in the legal case. Once all of the citations in the legal case have been identified and classified into one of the types described above, they are fed to a grouper process 126 as shown in FIG. 3. After the grouper process, the deep processing process determines the total number of each citation type in the legal case for each different citation in step 238. For example, for the citation of Pleasant v.celli, there may be a total of 5 citations in the legal case, three at the very front of the string citation and two inside the string citation. This information about each citation in the legal case and any data about the verified quotation associated with a particular citation is used in step 240 to determine the depth symbol to be assigned to that particular citation. Once the depth symbols are assigned to each reference, the depth processing is complete.

One example of a technique for assigning depth symbols for a particular application will now be described, but the invention is not limited to any particular technique for assigning the depth symbols. Furthermore, the present invention is not limited to any particular type of depth symbol. In this example, references in the legal case that appear 1 to 3 times in any reference type (i.e., independent reference, reference at the forefront of string reference, or reference in the middle of string application) in the legal case are assigned two stars (e.g.,) and references that appear 4 to 8 times in any reference type in the legal case are assigned three stars (e.g.,) and references that appear 9 or more times in any reference type are assigned four stars (e.g.,). To further refine these assignments, a reference that appears 3 times in any reference type and has a verified quote associated with it is assigned three stars (e.g., x), while if the reference only has an internal string reference type, one star is subtracted from the reference. Thus, depth symbols corresponding to a particular application in the legal case are automatically assigned by the system according to the invention. The depth symbols help users of the system to more quickly determine which references are likely to be more relevant. The subject matter text assignment process according to the present invention will now be described.

FIG. 10 is a flow chart illustrating a method 250 for assigning a piece of text from a cited case to the citation in the legal case in accordance with the present invention. In the examples described below, the text of the annotations in the cited cases is assigned to the citations, but the text assignment process according to the invention can be used with different text passages in the cited cases. In step 252, a reference location (i.e., a text region likely to correspond to text supported by the reference) corresponding to each reference is assigned according to a rule set that will be discussed below.

To identify the reference location, several text parsing rules may be used, some of which are stronger than others, but which as a whole will likely identify the text. To allow for different validity of the different rules, the extracted text may be divided into three groups, "high", "medium", and "low", based on the likelihood that the extracted text is part of the correct text position. The rules include:

categories	Text
		Height of	1. Any non-referenced material contained in a sentence that includes a base reference. 2. All first non-referenced sentences within the same paragraph that precede the base reference. 3. If there is no type 2 sentence, it is all the first non-referenced sentences within the same paragraph that follow the reference. 4. All text from the quotes of the quoted case can be identified.
In	5. All sentences that appear between the next preceding reference of the base reference (not contiguous with the base reference) and the class 2 sentence. 6. The next following reference appearing in the base reference (not contiguous with the base reference) is a type 3 sentenceAll sentences in between. 7. If there is no type 2 or type 3 sentence and the paragraph containing the base reference ends with a colon or comma, it is the whole of the next succeeding paragraph of the paragraph containing the base reference. 8. If there is no type 2 or type 3 sentence and no type 6 paragraph, it is the whole of the next preceding paragraph of the paragraph containing the base reference. 9. If any of the text regions identified by any of the rules includes an annotation reference, then the annotation and its key line are included.
		Is low in	10. The whole of this paragraph containing the base references.
In general	11. If the reference appears in a footnote, it is treated as appearing at the footnote location and at the location of the footnote mark.
		Is not recorded in	12. A reference only appears in the subsequent history of another reference.

Then, in step 254, the terms at the reference location are weighted with high, medium or low matches corresponding to weights of 2.0, 1.0 and 0.5, respectively, according to the rule used to identify the reference location. Different rule sets may be required to determine the weights for different types of documents, such as legal cases or legal review articles. Once the pieces of text have been identified and confidence values have been assigned, the identified pieces of text are matched with pieces of text that may be within the cited documents in step 255. In one example, the pieces of text within the cited documents may be headnotes, but the invention is not limited to any particular type of text with which the identified pieces of text match. The matching may be accomplished using natural language queries as described in previously referenced U.S. patent nos. 5,265,065 and 5,418,948, which are owned by the assignee of the present application and incorporated herein by reference. The result of the search is a list of possible text segments (such as endorsements) from the cited case, which may be assigned to the citations in the legal case, and a confidence score corresponding to each possible text segment may be assigned.

Next, in step 258, the one or more pieces of text to be assigned to the reference are selected through a thresholding process. The thresholding process orders the sections of text for each reference based on the confidence score. Each text segment may be entered into the database whenever the following amount equals or exceeds 0.5:

the β values corresponding to this equation are as follows:

literature reference	Sorting	β0	β1	β2	β3
						non-ALR	1	4.0451	3.1975	0.8477	.9033
non-ALR	2	0.5573	9.0220	1.0348	0.6743
						non-ALR	3	-2.0421	11.2619	0.8949	0.2954
ALR	1	-1.4256	50.4929	0.3488	0.0000
						ALR	2	-2.8199	65.6148	0.6207	0.0000
ALR	3	-2.3701	40.8479	1.2445	0.0000
						ALR	4	-3.3474	60.8075	1.7349	0.0000
ALR	5	-3.0805	55.6003	1.3188	0.0000

Where each column labeled "ALR" contains a variable corresponding to an ALR article, as described above, which has a higher confidence score than non-ALR documents. Each column labeled "non-ALR" contains variables corresponding to the non-ALR literature.

In the equation, Freq is the total citation frequency corresponding to the citation pair, and lag2 is the confidence score of the second following candidate when the candidates are ranked by confidence score in descending order (or 0.4 if there is no such candidate). Once the thresholding is complete and the one or more pieces of text are assigned to each reference in the legal case, the one or more pieces of text are stored in the database as described above in step 260 so that the text can be retrieved for the user when requested.

Briefly, the subject matter distribution process generates one or more pieces of text for references in the legal case based on the pieces of text (e.g., annotations) in the referenced case. The process first automatically identifies supporting text in the legal case and assigns confidence values to the supporting text that match all of the pieces of text with the pieces of text in the cited case, and then automatically assigns a piece of text from the cited case (such as an annotation) to the particular citation. These subject matter assignments allow for sorting or selecting references to the legal case through the subject matter, which may be helpful in the ranking process.

Thus, the machine-implemented system according to the present invention automatically processes documents, such as legal cases, and generates information about the documents that can provide useful information about the content of the documents to a user of the system. On the other hand, in conventional systems, such information about the document would be generated by a human by reading the document and taking notes about the document, which is a very slow, expensive, error-prone process. For a legal case, the system may automatically generate information about the negative history of the legal case, information about the deep processing of citations made by the legal case, information about the citations in the legal case that are verified as originating from a particular source, and information about one or more endorsements assigned to particular citations in the legal case. Thus, the operator of the system can quickly generate this information about the legal case, and the user of the system can quickly find this information, since it can be easily accessed from a graphical user interface.

Appendix

Slide 1

Slide 2

Slide 3

Slide 4

Slide 5

Slide 6

Slide 7

Slide 8

Slide 9

Slide 10

Slide 11

Slide 12

Slide 13

Slide 14

Slide 15

Slide 16

Slide 17

Slide 18

Slide 19

Slide 20

Slide 21

Slide 22

Slide 23

Slide 24

Slide 25

Slide 26

Slide 27

Slide 28

Slide 29

Slide 30

Slide 31

Slide 32

Slide 33

Slide 34

Slide 35

Slide 36

Slide 37

Slide 38

Slide 39

Slide 40

Slide 41

Slide 42

Slide 43

Slide 44

Slide 45

Slide 46

Slide 47

Slide 48

Slide 49

Slide 50

Slide 51

Slide 52

Slide 53

Slide 54

Slide 55

Slide 56

Slide 57

Slide 58

Slide 59

Slide 60

Slide 61

Slide 62

Slide 63

Slide 64

Slide 65

Slide 66

Slide 67

Slide 68

Slide 69

Slide 70

Slide 71

Slide 72

Slide 73

Slide 74

Slide 75

Slide 76

Slide 77

Slide 78

Slide 79

Slide 80

Slide 81

Slide 82

Slide 83

Slide 84

Slide 85

Slide 86

Slide 87

Slide 88

Slide 89

Slide 90

Slide 91

Slide 92

Slide 93

Slide 94

Slide 95

Slide 96

Slide 97

Slide 98

Slide 99

Slide 100

Slide 101

Slide 102

Slide 103

Slide 104

Graphic KeyCite for case

The

The program-based, in-process, a/will-be-subject layout a/will be in-process, a/will be in-process

The one

The

Case frame layout

IconShort captionQuoting court line date file line or UPO …
	History processingReference toHistory processingReference to…

Layout of program animation box

[ icon]History processingCase box for court line date [ KeyCite ]]

Situation(s)

36

Case (Miranda-86 sct 1602): a- > B: a is the parent A- > c of B: a is the parent A- > d of c: a is parent B- > e for d: b is the parent B- > f of e: b is the parent B- > g of f: b is the parent B- > h of g: b is parent B- > i of h: b is the parent B- > P of i: b is parent K- > l of P: parent K- > P where K is l:k is the parent J- > m of p: j is parent J- > P of m: j is parent O- > n for P: o is parent/homologous O- > P of n: o is parent P- > q of P: parents A and J with P being q B, c, d, e, f, g, h, K and O in court 3 i, l, m, n, P and q in court 1 in court 2

Rule: the parent is centered under its children. The non-backreview relationship arrow should enter from the bottom of the box. When there is a T-shaped intersection between parent and child, the bottom vertical line will be offset. Relationship segregation will occur at the forensic level of the parent. To map all of its parent relationships, the width of the child will be extended to the necessary point. The motives corresponding to the same parent within the same court are layered on top of each other. Stacked program campaigns are ordered with the earliest at the bottom and the latest at the top. When program trends have the same date, they are stacked in whatever order we receive them. When the children are sorted from left to right, the entire program will vote on the date on which the tier picked its earliest member. The order of the children will be the earliest on the left and the latest on the right.

37

Case (Festo-172 F.3d 1361): a- > E: a is the parent A- > J of E: a is the parent/great-parent B- > E of J: b is the parent B- > J of E: b is the parent/great-ancestor parent B- > C of J: b is the parent D- > E of C: d is the parent E- > F of E: e is the parent E- > G of F: e is a parent/ancestor of GThe generation E- > H: e is the parent/ancestor F- > G of H: f is the parent F- > H of G: f is parent F- > J of H: f is parent H- > I of J: h is the parent H- > J of I: h is parent/homologous J- > k of J: j is parent J- > L of k: j is parent L- > M of L: l is the parent M- > N of M: m is the parent N- > o of N: parent N- > p with N being o: parent F, k, L, p with N being p&o E, G, H, I, J, M in court 1&N in court 2C&D in court 3A&B in court 4A (12/00/1973) B (10/12/1982) C (04/12/1982) D (02/03/1994) E (12/14/1995) F (03/17/1997) G (06/09/1997) H (04/19/1999) I (08/20/1999) J (11/26/2000) k (06/18/2001) L (05/28/2002) M (09/20/2002) N (09/26/2002) o (04/19/2004) p (04/19/2004)

Rule: the parent is centered under its children. The non-backreview relationship arrow should enter from the bottom of the box. When there is a T-shaped intersection between parent and child, the bottom vertical line will be offset. Relationship segregation will occur at the forensic level of the parent. The case sent down will start from the right side of the case to the top of the case to which the case is sent down. When more than one case is sent down from a single parent, the lines from the right side of the parent will continue to accommodate additional cases. The parent that has reached the top of the case will have a line that emanates from its right and connects to its children. If the parent cannot be centered under the child, it will be offset under the child. The motives corresponding to the same parent within the same court are layered on top of each other. Stacked program campaigns are ordered with the earliest at the bottom and the latest at the top. When program trends have the same date, they are stacked in whatever order we receive them. When the children are sorted from left to right, the entire program will vote on the date on which the tier picked its earliest member. The order of the children will be the earliest on the left and the latest on the right.

41

Case (125 SCt 82): b- > A: b is the parent C- > A of A: b is the parent D- > A of A: b is the parent E- > A of A: b is the parent F- > A of A: b is the parent G- > A of A: b is the parent H- > A of A: b is the parent I- > A of A: b is the parent J- > A of A: b is parent K- > A of A: b is the parent L- > A of A: b is the parent M- > A of A: b is the parent N- > A of A: b is the parent O- > A of A: b is the parent P- > A of A: b is the parent Q- > A of A: b is the parent R- > A of A: b is the parent S- > A of A: b is the parent T- > A of A: b is the parent U- > A of A: b is the parent V- > A of A: b is the parent W- > A of A: b is the parent X- > A of A: b is the parent Y- > A of A: b is the parent Z- > A of A: b is the parent AA- > A of A: b is the parent BB- > A of A: BB is the parent A of A B, C, D, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V, W, X, Y, Z, AA on court level 1&BB on court level 2

Rule: the parent is centered under its children. The non-backreview relationship arrow should enter from the bottom of the box.If the parent cannot be centered under the child, it will be offset under the child. To map all of its parent relationships, the width of the child will be extended to the necessary point. The order of the children will be the earliest on the left and the latest on the right.

Rule summaries

The parent is centered under its children.

The non-backreview relationship arrow should enter from the bottom of the box.

When there is a T-shaped intersection between parent and child, the bottom vertical line will be offset.

Relationship segregation will occur at the forensic level of the parent.

The case sent down will start from the right side of the case to the top of the case to which the case is sent down.

When more than one case is sent down from a single parent, the lines from the right side of the parent will continue to accommodate additional cases.

The parent that has reached the top of the case will have a line that emanates from its right and connects to its children.

When a parent and a child are on the same court level and their relationships are on a horizontal litigation list, the parent will be drawn to the left of the child.

If the parent cannot be centered under the child, it will be offset under the child.

To map all of its parent relationships, the width of the child will be extended to the necessary point.

When a parent will need to bypass one child to reach another child, the parent will be biased.

The motives corresponding to the same parent on the same court level are layered on top of each other.

Stacked program campaigns are ordered with the earliest at the bottom and the latest at the top.

When program trends have the same date, they are stacked in whatever order we receive them.

When the children are sorted from left to right, the entire program will vote on the date on which the tier picked its earliest member.

The order of the children will be the earliest on the left and the latest on the right.

When lines need to intersect, there will be a protrusion on the horizontal line at the point where the lines intersect.

If a program box has different history processing to be displayed, it should be promoted to a substantive box.

A case with backreviewed children and transversely litigated children should have two lines emanating from the right side of their box.

Only program campaigns with a common single parent are stacked.

Emphasis is given to

Out of reference checking: graphical KeyCite Pictures a Screen outlining the history of program cases

One picture outperforms a myriad of languages. The value of the powerful new Graphical KeyCite may be even higher for legal researchers, since it exactly illustrates the procedural history of case law.

Thomson West is a company under the flags Thomson Corporation (NYSE: TOC; TSX: TOC), whose Graphical KeyCite now introduced is KeyCiteAnd said KeyCiteServices, when first introduced, have triggered a revolution in the area of reference checking.

In the current year, legal librarians of the American national library of Law council agreed upon the intuitive KeyCite flag, which lets legal researchers know immediately whether a judicial opinion is still a good law, and the treatment depth stars and symbols indicating the degree of dependence on the case among other opinions. KeyCite is also the first citation checking service that allows researchers to explore the history of the case without difficulty.

The Graphical KeyCite brought these innovations to a new height by exactly mapping the immediate history of the case. As the case ascends to higher court, the feature links the citation to later proceedings, debates, and low-level court decisions. This exclusive KeyCite feature helps researchers see immediately that the case if moving through the court system over time and understand quickly the impact at each level.

KeyCite's product development director Jon Medin says: "the history of court cases was illustrated for the first time, so that researchers can be helped to understand their effects more quickly".

Medin supplement, KeyCite combines analysis from the legal editors of Thomson West with techniques to describe problems, such as how many cited cases are discussed by referring to the cases. Medin mentions: "the same attorney who created the massive collection of the West corporation's endorsements to authoritative case law — the editors also assigned KeyCite flags and symbols that attorneys and judges rely on to see if citation is still a good law".

The literature on Westlaw includes more links to related sources than any other legal research service. Medin mentions that while forensic documents move within a judicial system, grapanickeycite utilizes these links and uses proprietary technology to illustrate the connections between the forensic documents. In addition, a researcher can simply click on a chart to open a full text document on Westlaw.

Westlaw marketing department of Thomson West advanced director Mike Bernstein: "in our tests, researchers using the Graphical KeyCite can understand the direct history of the case faster and more accurately. The graphpacalkeycite picture outperforms the vernacular absolutely for anyone who performed the citation study.

Although the foregoing description refers to particular embodiments of the present invention, it will be appreciated by those skilled in the art that changes may be made in this embodiment without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims.

Claims (12)

1. A system for automatically processing legal case documents having legal authority, comprising:

means for identifying candidate citations in the legal case documents referring to other legal case documents;

means for marking the candidate references to produce marked references;

means for identifying candidate quotations in the legal case documents, the candidate quotations having corresponding sources;

means for marking the candidate quotations;

means for verifying that the candidate quotation originates from the corresponding source to produce a verified quotation;

means for generating a processed depth value for each marker citation, the processed depth value indicating a degree of importance that each marker citation appears in the legal case document, wherein the processed depth value is based on at least one of: the number of times each markup citation appears in the legal case document, the type of each markup citation, and the association of each markup citation with one or more validated quotations;

means for generating a list of written opinions of other legal case documents that negatively impact the legal authority of the legal case document; and

means for providing the processed depth value for each markup reference, and the list of written opinions from other legal case documents to a user interface.

2. The system of claim 1, further comprising means for scanning the other legal case documents for words indicating negative treatment of the legal case documents when generating the list of written opinions.

3. The system of claim 1, further comprising:

means for identifying in the legal case document a portion of text supported by each markup citation;

means for classifying each tagged reference in one or more subject matter titles based on the identified portion of text to produce a classified reference; and

means for providing the sorted references to a user interface in response to a search for the one or more subject content titles.

4. A method for automatically processing legal case documents having legal authority, comprising:

identifying candidate citations in the legal case documents referring to other legal case documents;

tagging the candidate references to produce tagged references;

identifying candidate quotations in the legal case documents, the candidate quotations having a corresponding source;

marking the candidate quotations;

verifying that the candidate quotation originates from the corresponding source to produce a verified quotation;

generating a processed depth value for each tag reference, the processed depth value indicating a degree of importance based on at least one of: the number of times each markup citation appears in the legal case document, the type of each markup citation, and the association of each markup citation with one or more validated quotations;

generating a list of written opinions of other legal case documents that negatively impact the legal authority of the legal case document; and

providing the processed depth value for each markup reference, and the list of written opinions from other legal case documents to a user interface.

5. The method of claim 4, further comprising:

displaying an interactive hierarchical representation of case history for additional legal cases corresponding to at least one of the marker citations,

wherein the hierarchical representation is organized and displayed according to a set of rules and includes a plurality of horizontally oriented rectangular regions, wherein each rectangular region corresponds to a level of a court system.

6. The method of claim 5, wherein the hierarchical representation of the case history is horizontally oriented.

7. The method of claim 5, wherein the hierarchical representation of the case history includes a plurality of text boxes, wherein each text box is displayed in association with at least one line segment connecting the box to another box, and wherein each box contains a case name.

8. The method of claim 7, wherein each box further comprises a hyperlink to an electronic judicial opinion corresponding to the case name.

9. The method of claim 7, wherein each box further comprises a symbol indicating a legal authority status of the case associated with the case name.

10. The method of claim 5, wherein displaying the hierarchical representation of case history includes referencing a data structure of a legal case, wherein the data structure associates a court area indicator with the legal case, and the court area indicator is used to determine a display location of information about the legal case.

11. The method of claim 5, wherein the rule set includes rules for which a parent case is displayed centrally under its associated child case.

12. The method of claim 11, wherein the rule set includes rules that sub-cases with a common mother are displayed from left to right in order of date.