Overlapping subdocuments in a vector space search process
AssigneeClaritech Corporation
InventorDavid A. Evans
Life of this patent
Estimating the term…
Abstract
The present invention is a method and apparatus for retrieving information from a database. Initially, the documents within the database are divided into mutually exclusive subdocuments that generally correspond to paragraphs of text. The present invention further creates a second set of subdocuments that overlap adjacent paragraphs of text. In particular, the location of the overlapping subdocuments depends on the size of the initial paragraphs. This second set of overlapping subdocuments are scored just as the mutually exclusive subdocuments are scored. The scores from both the mutually exclusive and overlapping subdocuments are used in ranking the relevance of documents to a query. The use of both sets of subdocument scores improves the effectiveness of the scoring algorithm.
Claims15
In a database containing a plurality of segments of data identified as a corresponding plurality of documents, a method for displaying documents from said database, comprising:
- generating a plurality of paragraph subdocuments comprising at least a portion of one of said segments of data for said plurality of documents in said database;
- generating a plurality of overlapping subdocuments from adjacent paragraphs for said plurality of documents in said database;
- generating a score for each of the paragraph subdocuments and overlapping subdocuments wherein said score relates a subdocument to a query of said database;
- sorting the subdocuments into a rank order based on said score;
- and displaying data from one of said segments of data from which said highest ranked subdocument originates.
A computer readable medium bearing a sequence of instructions for searching a database containing a plurality of segments of data identified as a corresponding plurality of documents, said sequence of instructions comprising:
- generating a plurality of paragraph subdocuments comprising at least a portion of one of said segments of data for said plurality of documents in said database;
- generating a plurality of overlapping subdocuments for said plurality of documents in said database;
- generating a score for each of the paragraph and overlapping subdocuments wherein said score relates a subdocument to a query of said database;
- sorting the subdocuments into a rank order based on said score;
- and
- displaying data from one of said segments of data from which said highest ranked subdocument originates.
In a database containing a plurality of segments of data identified as a corresponding plurality of documents, a method of ranking documents from a database, comprising the steps of:
- retrieving from memory a segment of data having at least one document;
- generating at least two overlapping paragraph subdocuments comprising at least a portion of one of said plurality of segments of data from each document;
- inputting a query;
- calculating a score for each subdocument wherein each score relates said subdocument to said query;
- and
- ranking the subdocuments based on the score.
Dependencies are inferred from the claim text.