SEMANTICALLY WEIGHTED SEARCHING IN A GOVERNED CORPUS OF TERMS

• US 20100161601A1
• Filed: 12/22/2008
• Published: 06/24/2010
• Est. Priority Date: 12/22/2008
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: GRUBER, JOCHEN
• Granted Patent: US 8,156,142 B2
• Time in Patent Office: Days
• US Class Current: 707/736
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/367 Ontology

Abstract

A method and system for conducting semantically weighted searches in a governed corpus of terms is provided. A search expression having a plurality of terms for performing a search in the governed corpus of terms is received. The governed corpus of terms comprises a plurality of corpus expressions each comprising a plurality of terms, each term within the governed corpus of terms being associated precisely with a single concept within a lexical database. At least one concept of the lexical database is assigned to each term in the search expression based on a syntactical analysis. A semantic similarity is calculated between pairs of concepts of the search expression and one of the corpus expressions. A total semantic similarity is calculated between the search expression and the one of the corpus expressions by aggregating the semantic similarities of the pairs of concepts based on an order of significance of the terms.

Claims

1. A computer implemented method for conducting semantically weighted searches in a governed corpus of terms, the method comprising:

receiving a search expression having a plurality of terms for performing a search in the governed corpus of terms, wherein the governed corpus of terms comprises a plurality of corpus expressions each comprising a plurality of terms, wherein each term within the governed corpus of terms is associated precisely with a single concept within a lexical database;

based on a syntactical analysis, assigning at least one concept of the lexical database to each term in the search expression;

calculating a semantic similarity between pairs of concepts of the search expression and one of the corpus expressions, the pairs of concepts comprising a concept assigned to one of the terms in the search expression and a concept assigned to a correspondingly positioned term in the one of the corpus expressions; and

calculating a total semantic similarity between the search expression and the one of the corpus expressions by aggregating the semantic similarities of the pairs of concepts based on an order of significance of the terms.

2. The method of claim 1 wherein aggregating the semantic similarities of the pairs of concepts based on the order of significance of the terms comprises:

assigning a weight to each term of the one of the corpus expressions such that a most significant term of the corpus expression is assigned the highest weight and the least significant term is assigned the lowest weight.

3. The method of claim 2 wherein the rightmost term of the search expression and the corresponding rightmost term of the one of the corpus expressions are the most significant terms.

4. The method of claim 2 wherein the leftmost term of the search expression and the corresponding leftmost term of the one of the corpus expressions are the least significant terms.

5. The method of claim 1 wherein calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions comprises determining a conceptual distance between the pairs of concepts in a conceptual space defined by the concepts of the lexical database.

6. The method of claim 5 wherein the conceptual space comprises one or more hierarchical arrangements expressing relationships between the concepts of the lexical database.

7. The method of claim 5 wherein determining the conceptual distance between the pairs of concepts in a conceptual space comprises determining minimum number of elementary relationship segments connecting the pairs of concepts.

8. The method of claim 7 wherein the elementary relationship segment comprises an elementary distance between two concepts without an intervening concept positioned between.

9. The method of claim 1 further comprising:

designating the one of the corpus expressions as related to the search expression upon the value of the total semantic similarity exceeding a threshold value.

10. The method of claim 1 wherein calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions comprises:

iteratively calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions for each pair of concepts associated with the correspondingly positioned terms of the search expression and the corpus expressions until there are no the correspondingly positioned terms.

11. The method of claim 1 further comprising calculating the total semantic similarity between the search expression and all of the corpus expressions within the governed corpus of terms.

12. The method of claim 11 further comprising:

determining whether the total semantic similarity between the search expression and each one of the corpus expressions within the governed corpus of terms is above a threshold value; and

displaying only the corpus expressions with the total semantic similarity above the threshold value in a descending order of the total semantic similarity values.

13. A system for conducting semantically weighted searches in a corpus of terms, the method comprising:

a lexical database storing a plurality of terms;

a concept store storing a hierarchy of concepts associated with terms in the lexical database;

a term to concept mapping table for storing mappings between the terms in the lexical database and the concepts assigned to the terms;

a governed corpus store for storing a governed corpus of terms comprising a plurality of corpus expressions, each governed corpus expression comprising a plurality of terms;

a governed corpus to concept mapping table for storing mappings between the terms of the corpus expressions and a unique concept assigned to each of the terms from the concept store;

a semantic search engine for receiving a search expression comprising a plurality of terms for performing a search in the governed corpus of terms to find at least one related corpus expression, the semantic search engine comprising;

a concept mapper operable for syntactically matching every term of the search expression to the terms in the lexical database and assigning one or more concepts from the concept store to the terms of the search expression;

a calculation module operable for calculating a semantic similarity between each pair of concepts of the search expression and the corpus expressions, the pair of concepts comprising a concept assigned to a term in the search expression and a concept assigned to a correspondingly positioned term in one of the corpus expressions; and

calculating a total semantic similarity between the search expression and the corpus expressions by combining the semantic similarities of the pairs of concepts of the individual terms based on the order of significance of the terms; and

a relation designator for designating the corpus expression as being related to the search expression when the value of the total semantic similarity is greater than a threshold value.

14. A machine-accessible medium that comprises instructions which, when executed by a machine, causes the machine to perform operations comprising:

receiving a search expression having a plurality of terms for performing a search in the governed corpus of terms, wherein the governed corpus of terms comprises a plurality of corpus expressions each comprising a plurality of terms, wherein each term within the governed corpus of terms is associated precisely with a single concept within a lexical database;

based on a syntactical analysis, assigning at least one concept of the lexical database to each term in the search expression;

calculating a semantic similarity between pairs of concepts of the search expression and one of the corpus expressions, the pairs of concepts comprising a concept assigned to one of the terms in the search expression and a concept assigned to a correspondingly positioned term in the one of the corpus expressions; and

calculating a total semantic similarity between the search expression and the one of the corpus expressions by aggregating the semantic similarities of the pairs of concepts based on an order of significance of the terms.

15. The machine-accessible medium of claim 14 wherein aggregating the semantic similarities of the pairs of concepts based on the order of significance of the terms comprises:

assigning a weight to each term of the one of the corpus expressions such that a most significant term of the corpus expression is assigned the highest weight and the least significant term is assigned the lowest weight.

16. The machine-accessible medium of claim 14 wherein calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions comprises determining a conceptual distance between the pairs of concepts in a conceptual space defined by the concepts of the lexical database.

17. The machine-accessible medium of claim 16 wherein determining the conceptual distance between the pairs of concepts in a conceptual space comprises determining minimum number of elementary relationship segments connecting the pairs of concepts.

18. The machine-accessible medium of claim 17 wherein the elementary relationship segment comprises an elementary distance between two concepts without an intervening concept positioned between.

19. The machine-accessible medium of claim 14 further providing instructions which when executed by the machine cause the machine to perform further operations comprising:

designating the one of the corpus expressions as related to the search expression upon the value of the total semantic similarity exceeding a threshold value.

20. The machine-accessible medium of claim 14 wherein calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions comprises:

iteratively calculating the semantic similarity between pairs of concepts of the search expression and the one of the corpus expressions for each pair of concepts associated with the correspondingly positioned terms of the search expression and the corpus expressions until there are no the correspondingly positioned terms.