Sort system for text retrieval
AssigneeClaritech Corporation
InventorMichael L. Horowitz
Life of this patent
Estimating the term…
Abstract
The present invention is a method for operating a computer system to retrieve information from a computer database. This method decomposes documents from the database into subdocuments and then inverts the database. Also, a query for retrieving documents from a database is decomposed into terms. The subdocuments from the inverted database and the terms from the query are then used to compute a score that indicates a quantitative relation between terms in the query and the subdocuments. The resulting list of the subdocuments and their scores is then reorganized into a heap form. The highest ranking subdocument is then selected by the computer and the text associated with this subdocument is displayed. The remainder of the subdocument score list is concurrently reheapified.
Claims10
A method for retrieving documents from a database, comprising:
- creating a plurality of subdocuments from a plurality of documents stored in a database;
- scoring the subdocuments;
- sorting the subdocument scores to identify the highest ranking subdocument;
- and
- displaying text from a document corresponding to said highest ranking subdocumnet after identification of said highest ranking subdocument.
An apparatus for retrieving documents from a database, comprising:
- a computer coupled to a disk storage unit and to a display unit, said disk storage unit stores a database in at least one file;
- said computer retrieves data from said database file to create a plurality of subdocuments from a plurality of documents stored in a database file;
- said computer scores said subdocuments;
- said computer sorts said subdocument scores to identify the highest ranking subdocument;
- and
- said computer displays on said display unit text retieved from a document corresponding to said highest ranking subdocumnet after identification of said highest ranking subdocument.
Dependencies are inferred from the claim text.