Method for statistically projecting the ranking of information

• US 5,765,150 A
• Filed: 08/09/1996
• Issued: 06/09/1998
• Est. Priority Date: 08/09/1996
• Status: Expired due to Term

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Digital Equipment Corporation (HP Inc.)
• Inventors: Burrows, Michael
• Primary Examiner(s): AMSBURY, WAYNE P
• Application Number:
  US08/695,905
• Time in Patent Office: 669 Days
• Field of Search: 395/613, 395/610, 395/603, 395/615, 395/605, 707/5, 707/102
• US Class Current: 1/1
• CPC Class Codes: G06F 16/319 Inverted lists
• CPC Class Codes: G06F 16/38 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

A computer implemented method selectively searches an index of a database according to scores assigned to records of the database located during the searching. The records of the database are index by storing index entries in a memory. Each index entry includes a word entry representing a unique portion of information of the database and one or more location entries indicating where the unique portion of information represented by the word entry occurs in the records of the database. A weight is assigned to each index entry according to a relative frequency of occurrence of the portion of information in the database. The index is sequentially searched to locate records qualified by a query having terms and operators. The terms correspond to index entries. The located records are scored according to the number of times portions of information corresponding to the terms of the query occur in the records and their associated weights. The scores and identities of the located records are stored in entries of a ranking list having a predetermined number of entries. In response to searching a predetermined fraction of the index, a determination is made to see if any unlocated records of the database can receive a score higher than one of the records stored of the ranking list using index entries having a lowest weight. If not, the index is searched using only the index entries having weights higher than index entries having the lowest weigh.

Claims

1. A computer implemented method for ranking records of a database located during a search of an index to the database, comprising:

indexing the records of the database by storing index entries in a memory to create the index, each index entry including a word entry representing a unique portion of information of the database and one or more location entries indicating where the unique portion of information represented by the word entry occurs in the records of the database;

assigning a weight to each index entry according to a relative frequency of occurrence of the portion of information in the database;

parsing a query into terms and operators, each term associated with a corresponding index entry;

sequentially searching index entries to locate records of the database which are qualified by the terms and operators of the query;

scoring each located record according to the number of times portions of information corresponding to the terms of the query occur in each record and their associated weights;

storing the scores and identities of the located records in entries of a ranking list, the ranking list having a predetermined number of entries; and

in response to having searched a predetermined fraction of the index, determining if any unlocated records of the database can receive a score higher than one of the records stored of the ranking list based the index entries corresponding to the terms having a lowest weight, and if not, searching the index using only using the index entries having weights higher than the lowest weight.

2. The method of claim 1 further comprising:

assigning the weight to each index entry according to a difference between the number of records indexed and the number of records including the portion of information corresponding to the word entry of the index entry.

3. The method of claim 1 further comprising:

ordering the entries of the ranking list according to the scores.