Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1999-09-23
2004-08-03
Corrielus, Jean M. (Department: 2172)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C704S009000, C704S001000, C706S012000, C706S016000, C706S045000, C715S252000, C715S252000, C715S252000
Reexamination Certificate
active
06772149
ABSTRACT:
COPYRIGHT NOTICE: A portion of the disclosure (including all Appendices) of this patent document contains material that 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 U.S. Patent and Trademark Office patent file or records, but the copyright owner reserves all other copyright rights whatsoever.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to computer-assisted legal research (CALR). More specifically, the invention relates to systems and methods that identify and distinguish facts and legal discussion in the text of court opinions.
2. Related Art
Few patents and a very limited body of research literature are devoted to analysis and indexing of court decisions and case law. One reason for this phenomenon may be that the complexity of the current body of legal data overwhelms computing applications. Some applications, including artificial intelligence applications, were too ambitious and failed to follow the scientific approach of “divide and conquer”: decompose a large problem into smaller ones and tackle the smaller and easier problems one at a time.
The present invention is directed to a computing method to address one of these smaller problems: identifying and distinguishing the facts and the legal discussion in a court's legal opinion. This invention is fundamental to the improvement of the future CALR.
Factual analysis is the very first task in legal research process. A thorough analysis of facts leads to a solid formulation of legal issues to be researched. Facts are the dynamic side of law, in contrast to the relatively stable authoritative legal doctrines.
Most legal research and controversy concerns facts, not law—cases are most often distinguished on the facts. The rules stated by courts are tied to specific fact situations and cannot be considered independently of the facts. The rules must be examined in relation to the facts. In this sense, the facts of a legal issue control the direction of CALR.
Applicants are aware of no patent related to distinguishing fact from legal discussion in case law documents. Most of the patents that are at all related to legal documents are in the field of information retrieval, and these patents generally do not include differentiation of facts from legal discussions in their indexing of the documents (see U.S. Pat. Nos. 5,544,352; 5,771,378; 5,832,494). Some of the patents emphasize the usage of legal concepts, not facts, in the form of headnotes, classification codes, and legal lexicons (see U.S. Pat. Nos. 5,265,065; 5,418,948; 5,488,725).
In research literature, the FLAIR project (Smith 93 and 97) attempted to separate legal content from fact, but focused heavily on legal concepts. In FLAIR, a legal lexicon is carefully constructed manually by legal experts. In this lexicon, legal concept terms are hierarchically organized by their conceptual relationships, and synonyms and alternative word forms for each lexicon term are included for completeness. FLAIR defines facts as follows: “Fact words are every words in the database other than concepts, statute citations, cited cases, and noise words. Fact phrases are fact words that appear next to each other with or without noise words in between.” In other words, there is no specific process that specializes in identifying the facts themselves—facts are merely derivatives of look-ups from the concept lexicon. Also, FLAIR's notion of fact includes only words and phrases, and does not provide for entire passages in a court decision.
A few other research projects share the lexicon approach adopted in FLAIR, now referred to as “conceptual legal information retrieval” (Hafner 87; Bing 87; Dick 87). These research techniques are generally domain-specific, small-scale applications.
Some research techniques that do process facts apply case-based reasoning (CBR) technologies to legal data (Rissland 87, 93, 95; Daniels 97; Ashley 90). CBR represents a branch of artificial intelligence research and emphasizes knowledge acquisition and knowledge representation using a device known as a “case frame”, or “frame”. To populate their “frames”, the CBR researchers analyze sample case law documents to extract, condense, and categorize facts and other relevant information into pre-defined frames. The quality of the extraction of facts, then, is limited to the quality of the design of the frames themselves; a fact that is important in one CBR frame is not necessarily important in another. This manual extraction and processing is neither repeatable nor scalable—a CBR project usually employs only a few dozen to a couple of hundred case law documents on a very narrow legal subject, like personal bankruptcy or contributory negligence.
A broader approach than CBR is the application of artificial intelligence (AI) to legal reasoning. In any of these computerized AI applications, facts, as in the CBR applications, play a crucial role in automatic inference. In the earlier research, the assumption is that facts are already available to help legal reasoning (Meldman 77; Tyree 81). The same assumption is made in the theoretical works (Levi 42; Gardner 87; Alexy 89; Rissland 90). How these facts are obtained was not the concern in these works. After about 1980, some researchers started creating small fact data banks for their experiments in order to build empirical evidence of effectiveness of their proposed models (Nitta 95; Pannu 95). But their approach to gathering facts from court decisions was ad hoc, and has no real potential for processing millions of decisions found in modern commercial legal databases.
A relevant research work is the SALOMON project in Belgium (Moens 97). SALOMON performs detailed analysis on criminal case decisions to programmatically identify the semantic text segments and summarize the contents within each segment. A Belgian criminal case is typically made up of nine logic segments: the superscription with the name of the court and date, identification of the victim, identification of the accused, alleged offences, transition formulation, opinion of the court, legal foundations, verdict, and conclusion. SALOMON focuses on identifying three of these nine segments: alleged offences, opinion, and legal foundations. The locating of alleged offences in a Belgian criminal case is roughly equivalent to the locating of the facts, a focus of the present invention.
SALOMON's identification of these three segments in a decision relies on “word patterns” and the sequence of the segments. For example, the legal foundation segment follows an opinion segment, and might be introduced with the word pattern “On these grounds.” It is unclear in the reported study, how many of the word patterns are employed in analysis and how the patterns are generated. It seems that the patterns are created manually, specific to the Belgian criminal cases. This approach is not too dissimilar from the lexicon approach used in FLAIR.
In addition, SALOMON assumes that only the text units, such as paragraphs, that appear in an alleged offense segment are related to the facts in the case. In reality, facts can appear in any part of a court decision. Even when there is a section devoted to facts, as in many U.S. and U.K. criminal cases, the facts are also embedded in the reasoning, arguments, and ruling, throughout the opinion. SALOMON makes no attempt to recognize these scattered “applied” facts. In fact, it eliminates them during its summarization process after the structure of a court decision is determined through the word pattern analysis.
The process of summarization in SALOMON consists of consolidating important content texts in each of the three determined segments. It is realized through a clustering analysis of the paragraphs in one segment, and extracting the important keywords from a few large clusters because they represent important topics in the case, based on the assumption of repetitive human usage of words. The condensed texts an
Ahmed Salahuddin
Humphrey Timothy Lee
Lu Xin Allan
Morelock John T.
Wiltshire, Jr. James S.
Corrielus Jean M.
Lexis-Nexis Group
Ly Anh
LandOfFree
System and method for identifying facts and legal discussion... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for identifying facts and legal discussion..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for identifying facts and legal discussion... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3300292