Internet searching using semantic disambiguation and expansion

• US 7,895,221 B2
• Filed: 08/20/2004
• Issued: 02/22/2011
• Est. Priority Date: 08/21/2003
• Status: Expired due to Fees

• Current Assignee: Idilia, Inc.
• Original Assignee: Idilia, Inc.
• Inventors: Colledge, Matthew, Barnes, Jeremy
• Primary Examiner(s): Lie; Angela M
• Application Number:
  US10/921,820
• Publication Number:
  US 20050080776A1
• Time in Patent Office: 2,377 Days
• Field of Search: 707 1- 5, 707/713, 707/715, 707/736, 707/738, 707/758, 707/759, 707/765, 707/766, 707/767, 707/771, 707/999.003, 707/999.004, 707/999.005
• US Class Current: 707/758
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

The invention provides a system and a method of searching for information in a database using a query. In the method, it comprises the steps of: disambiguating the query to identify keyword senses associated with the query; disambiguating information in the database according to the keyword senses; indexing the information in the database according to the keyword senses; expanding the keyword senses to include relevant semantic synonyms for the keyword senses to create a list of expanded keyword senses; searching the database to find relevant information for the query using the expanded keyword senses; and providing search results of the included information containing the keyword senses and other semantically related words senses. The system comprises modules which disambiguate queries and information and indexes the information in a database of word senses.

Claims

1. A computer based method of searching for information in a database using a query, said method comprising the steps of:

a) disambiguating information in a store of information, said information comprising documents containing text, to identify information keyword meanings, said information keyword meanings comprising meanings of words contained in each respective document and in the context in which said words are used in the respective documents, wherein said disambiguation step is executed by a disambiguation module that comprises encoded linguistic algorithms for analyzing both the syntax and the semantics of an input text, and is associated with a knowledge base containing semantic relationships between word meanings, the disambiguation module being operable to identify contextual meanings of words in the input text; and

,

indexing said documents in said database according to said information keyword meanings;

and,

b) disambiguating said query using said disambiguation module to identify query keyword meanings, said query keyword meanings comprising meanings of words contained in said query and in the context in which said words are used in the query;

automatically expanding said query keyword meanings, without user intervention, using a query processing module operable to identify other word meanings that are semantically related to the query keyword meanings and to create a list of expanded query keyword meanings, wherein said list comprises the query keyword meanings and the identified semantically related word meanings;

searching said database to identify matches between the expanded query keyword meanings and the indexed information keyword meanings;

identifying documents associated with the matched information keyword meanings, said identified documents comprising information relevant to said query; and

providing search results comprising the information relevant to said query.

2. The method of claim 1 wherein steps (a) and (b) are conducted independently of each other.

3. The method of claim 2, wherein disambiguating said documents comprises assigning probabilities to said information keyword meanings.

4. The method of claim 2 wherein the step of searching said database further comprises using probabilities of said query keyword meanings when identifying matches between the expanded query keyword meanings and the information keyword meanings.

5. The method of claim 2 wherein said step of expanding the query keyword meanings further comprises paraphrasing said query by parsing the syntactic structure of said query and transforming it into additional semantically equivalent queries using the original query keyword meanings or the expanded query keyword meanings.

6. The method of claim 5, wherein said keyword meanings represent a coarse grouping of fine keyword meanings.

7. The method of claim 2, wherein said keyword meanings represent a coarse grouping of fine keyword meanings.

11. The method of claim 2, wherein disambiguating the query comprises assigning probabilities to said query keyword meanings.

12. The method of claim 2 wherein said steps of disambiguation include using a knowledge base of semantic relationships between word meanings.

13. The method of claim 2 wherein said results are weighted based on semantic relationships between the query keyword meanings and the matched information keyword meanings.

14. The method of claim 2 wherein said results are weighted based on probabilities assigned to the query keyword meanings and on probabilities assigned to those same meanings among the matched information keyword meanings.

15. The method of claim 3 wherein the step of searching said database further comprises using probabilities of said information keyword meanings when identifying matches between the query keyword meanings and the information keyword meanings.

8. A system for providing information from a database responsive to a query, said system comprising:

an input means operable to receive said query;

an output means operable to provide results responsive to said query;

a database containing information to be searched by said query, said information comprising documents containing text;

an indexing module to create a reference index for said information to be used by said query;

a query processing module to apply said query to said database;

a disambiguation module for disambiguating words contained in the documents and the query into information and query keyword meanings, respectively, the disambiguation module comprising encoded linguistic algorithms for analyzing both the syntax and the semantics of an input text, and being associated with a knowledge base containing semantic relationships between word meanings, the disambiguation module being operable to identify contextual meanings of words in the input text;

wherein,

said disambiguation module being operable to disambiguate said documents to identify information keyword meanings, said information keyword meanings comprising meanings of words contained in each respective document and in the context in which said words are used in the respective documents;

said indexing module being operable to index the documents in said database according to said information keyword meanings;

said disambiguation module being further operable to disambiguate said query to identify query keyword meanings, said query keyword meanings comprising meanings of words contained in the query and in the context in which said words are used in the query;

said query processing module being operable to;

automatically expand said query keyword meanings, without user intervention, by automatically identifying other word meanings that are semantically related to the query keyword meanings and to create a list of expanded query keyword meanings;

search said database to find relevant information for said query by matching the expanded query keyword meanings to the information keyword meanings indexed in said database, wherein said relevant information comprises documents in said database associated with the matched information keyword meanings;

and provide search results comprising said relevant information.

9. The system as claimed in claim 8, wherein said disambiguation module ranks said query keyword meanings based on semantic relationships between the query keyword meanings and the matched information keyword meanings.

10. The system as claimed in claim 9, wherein said keyword meanings represent a coarse grouping of fine keyword meanings.