System, method, and software for identifying historically related legal opinions
The American legal system, judges and lawyers are continually researching an ever-expanding body of past judicial opinions, or case law, for the ones most relevant to resolution of new disputes. To facilitate these searches, some companies collect and publish the judicial opinions of courts across the United States in both paper and electronic forms, with some of the cases containing references to prior cases from other courts that have previously ruled on all or part of the same dispute. Identifying the prior cases is problematic, because, for example, conventional computer text-matching not only suggests too many non-prior cases, but also misses too many actual prior cases. Accordingly, the present inventors devised systems, methods, and software that generally facilitate identification of one or more documents that are related to a given document, and particularly facilitate identification of prior cases for a given case. One specific embodiment retrieves prior-case candidates based on information extracted from an input case, and then uses a support vector machine to determine which of the prior-case candidates are most probably prior cases for the input case.
1. A method implemented using at least one processor and a memory coupled thereto, the method comprising:
receiving an electronic text of a legal case;
extracting party names from the electronic text;
searching a database, based on the extracted party names for a set of candidate legal cases, each candidate legal case having an associated electronic text;
comparing party names from each of the set of candidate legal cases to the extracted party names from the electronic text;
defining a multi-dimensional feature vector for each candidate legal case, with a set of features including a similarity feature indicating similarity of at least a portion of the candidate legal case to a portion of the legal case;
scoring each of the candidate legal cases based on the multi-dimensional feature vectors using support-vector processing;
identifying one or more of the candidate legal cases based on scores for the candidate legal cases; and
selecting at least one of the candidate legal cases for association with the legal case.
2. The method of claim 1 , wherein each feature vectors further includes:
a history-language feature indicating whether the legal case includes history-language;
a prior-probability feature indicating a probability that the legal case has a prior case.
3. A machine-readable storage medium storing a set of program instructions for:
extracting party names from the electronic text;
searching a database, based on the extracted party names for a set of candidate legal cases, each candidate legal case having an associated electronic text;
comparing party names from each of the set of candidate legal cases to the extracted party names from the electronic text;
defining a multi-dimensional feature vector for each candidate legal case, with a set of features including a similarity feature indicating similarity of at least a portion of the candidate legal case to a portion of the legal case;
scoring each of the candidate legal cases based on the multi-dimensional feature vectors using support-vector processing;
identifying one or more of the candidate legal cases based on scores for the candidate legal cases; and
selecting at least one of the candidate legal cases for association with the legal case.
4. A computerized system for identifying historically related legal cases, the system comprising:
One or more processors and memory;
means for receiving an electronic text of a given legal case;
extraction means for extracting party names, court names, docket numbers, and history language from the electronic text;
means for searching a database, based on the extracted party names, court names, docket numbers, and history language, for a set of candidate legal cases, each candidate legal case having an associated electronic text;
means for comparing party names from each of the set of candidate legal cases to the extracted party names, court names, docket numbers, and history language from the portion of the electronic text;
means for defining a multi-dimensional feature vector for each candidate legal case, with a set of features including:
a title-similarity feature indicating similarity of a title of the candidate legal case to a title associated with the electronic text;
a history-language feature indicating whether the electronic text includes history-language;
a prior-probability feature indicating a probability that the given legal case has a prior case; and
a title-weight feature estimating significance of the title of the given legal case for document discrimination;
support-vector-processing means for scoring each of the candidate legal cases based on the multi-dimensional feature vectors;
decision-making means for identifying one or more of the candidate legal cases based on scores for the candidate legal cases; and
user-operable means for selecting at least one of the candidate legal cases for association with the electronic text.
5. The computerized system of claim 4 , wherein the user-operable means comprises a graphical user interface.
6. The computerized system of claim 5 , wherein one or more of the recited means is implemented as computer-executable instructions carried on an electronic, optical, or magnetic medium.
7. The computerized system of claim 5 , wherein the graphical user interface includes:
a first region for displaying data regarding one or more of the candidate legal cases;
a second region for displaying text from at least one of the candidate legal cases;
a third region for displaying text from the given legal case; and
a fourth region having a command input for causing association of the given legal case with a selected one of the candidate legal cases displayed in the first region.
8. The computerized system of claim 4 , wherein the set of features further includes:
a docket match feature indicating whether or not the electronic text and the candidate legal case has been assigned the same docket;
a check appeal feature estimating the probability of the prior court for the respective candidate case given an identified court;
a cited case feature indicating whether or not the candidate legal case is cited in the electronic text; and
an AP1 search feature indicating whether or not the candidate legal case was retrieved through a query.