Back to results
US 7,003,511 B1Granted
Mining and characterization of data
AssigneeInfotame Corporation
InventorAlexander Antonov
Life of this patent
Estimating the term…
Abstract
Method and system for identifying and characterizing publications containing a specified search-phrase, using number of, and relative frequency of, occurrences of the search-phrase (and, optionally, of synonym-phrases) in each publication to develop a measure of significance that represents the relevance of each identified publication (in which the specified search-phrase occurs) to the original search. The measure of significance can also be based on minimum separation distance between search-phrase occurrences and/or upon occurrences of search-phrase synonyms and time intervals within which the publication release dates fall.
Claims2
A method for searching for one or more relevant publications in a collection of publications, the method comprising:
- receiving a search-phrase whose occurrence is to be searched for in a selected collection of publications:
- determining if at least one publication in the collection, referred to as a “found publication,” contains the search-phrase;
- when at least one found publication is identified:
- associating a date of publication with each found publication;
- identifying at least M m1-word phrases that each occurs at least m2 times in the at least one found publication, referred to as a “found phrase,” where M, m1 and m2 are selected positive integers;
- and
- determining a numerical value for a secondary measure of significance for each found phrase in the at least one found publication, and displaying the found phrases with the M highest secondary measures;
- providing said secondary measure of significance for at least one of said found phrases as a sum over all of said found publications of a primary measure of significance computed for said at least one found phrase;
- computing said primary measure for said at least one found phrase by a procedure including:
- determining a number W of words in said found publication and a number N of times said at least one found phrase appears in said found publication;
- providing a frequency of occurrence N/W of said at least one found phrase in said found publication and an estimate of an average frequency of occurrence f(avg) of said at least one found phrase in a selected group of publications;
- and
- determining a first measure of significance of said at least one found phrase that depends upon at least one of the numbers N, N/W and N/W−f(avg);
- where said at least one found phrase appears at least twice in said found publication, determining a smallest separation distance d(min;N;W) between two occurrences of said at least one found phrase in said found publication;
- determining a distance measure of significance based upon the distance d(min;N;W);
- providing at least one synonym-phrase for said at least one found phrase in said found publication;
- determining a number N(syn) of occurrences of the synonym-phrase in said found publication;
- determining a synonym measure of significance based upon the number N(syn);
- and
- providing said primary measure as a weighted sum of said first measure, the distance measure and the synonym measure for said at least one found phrase.
17 more elements
A system for searching for one or more relevant publications in a collection of publications, the system comprising a computer programmed:
- to receive a search-phrase whose occurrence is to be searched for in a selected collection of publications;
- to determine if at least one publication in the collection, referred to as a “found publication,” contains the search-phrase;
- when at least one found publication is identified:
- to associate a date of publication with each found publication;
- to identify at least M m1-word phrases that each occurs at least m2 times in the at least one found publication, referred to as a “found phrase,” where M, m1 and m2 are selected positive integers;
- to determine a numerical value of a secondary measure of significance for each found phrase in the at least one found publication, and displaying the found phrases with the M highest secondary measures;
- to provide said secondary measure of significance for at least one of said found phrases as a sum over all of said found publications of a primary measure of significance computed for said at least one found phrase;
- to compute said primary measure for said at least one found phrase by a procedure including:
- determining a number W of words in said found publication and a number N of times said at least one found phrase appears in said found publication;
- providing a frequency of occurrence N/W of said at least one found phrase in said found publication and an estimate of an average frequency of occurrence f(avg) of said at least one found phrase in a selected group of publications;
- and
- determining a first measure of significance of said at least one found phrase that depends upon at least one of the numbers N, N/W and N/W−f(avg);
- where said at least one found phrase appears at least twice in said found publication, to determine a smallest separation distance d(min;N;W) between two occurrences of said at least one found phrase in said found publication;
- to determine a distance measure of significance based upon the distance d(min;N;W);
- to provide at least one synonym-phrase for said at least one found phrase in said found publication;
- to determine a number N(syn) of occurrences of the synonym-phrase in said found publication;
- to determine a synonym measure of significance based upon the number N(syn);
- and
- to provide said primary measure as a weighted sum of said first measure, the distance measure and the synonym measure for said at least one found phrase.
16 more elements
Dependencies are inferred from the claim text.