Automatic data categorization with optimally spaced semantic seed terms

• US 8,396,824 B2
• Filed: 05/30/2007
• Issued: 03/12/2013
• Est. Priority Date: 05/28/1998
• Status: Expired due to Fees

• Current Assignee: DataCloud Technologies, LLC (IP Investments Group LLC)
• Original Assignee: QPS Tech LLC (Intellectual Ventures LLC)
• Inventors: Au, Lawrence
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Tran, Mai T
• Application Number:
  US11/806,260
• Publication Number:
  US 20070294200A1
• Time in Patent Office: 2,113 Days
• Field of Search: 706/55, 707/999.01, 704/9
• US Class Current: 706/55
• CPC Class Codes: G06F 16/338 Presentation of query results
• CPC Class Codes: G06F 16/358 Browsing; Visualisation therefor

Abstract

A method and system for automatic data categorization in response to a user query. A document set is retrieved in response to the user query. A semantic parser parses the document set and produces semantic term-groups by parsing a semantic network of nodes. A seed ranker produces a plurality of advantageously spaced semantic seeds based on the semantic term-groups. A category accumulator stores the advantageously spaced semantic seeds. The semantic network of nodes is augmented with the advantageously spaced semantic seeds.

Claims

1. A method, comprising:

retrieving, by a computing device including a processor, a document set in response to receiving a query;

parsing a semantic distance network according to a plurality of terms in the document set into a plurality of semantic term groups;

determining respective balanced desirability values for respective semantic term groups of the plurality of semantic term groups, wherein the respective balanced desirability values comprise respective measures of prevalence of the respective semantic term groups as a function of respective measures of closeness of respective terms in the respective semantic term groups; and

selecting a semantic term group of the respective semantic term groups as a category descriptor based on a balanced desirability value associated with the semantic term group.

2. The method of claim 1, further comprising

augmenting the semantic distance network with the semantic term group.

3. The method of claim 1, wherein the parsing the semantic distance network further comprises:

parsing the semantic distance network according to a semantic distance from a node of the semantic distance network to another node of the semantic distance network.

4. The method of claim 1, wherein the parsing the semantic distance network further comprises:

parsing the semantic distance network according to a semantic distance from a first phrase of a first sentence to a second phrase of at least one of the first sentence or a second sentence.

5. The method of claim 1, wherein the selecting the semantic term group as the category descriptor further comprises:

ranking the plurality of semantic term groups based upon the respective balanced desirability values; and

selecting the semantic term group based on an associated ranking of a balanced desirability value of the semantic term group.

6. The method of claim 1, further comprising determining the respective measures of prevalence of the respective semantic term groups including determining a number of distinct terms of the plurality of terms that are co-located with the respective semantic term groups.

7. The method of claim 1, further comprising determining the respective measures of closeness of the respective terms of the respective semantic term groups including determining a first number of distinct terms of the plurality of terms that are co-located with at least two of the respective terms of the respective semantic term groups.

8. The method of claim 7, further comprising:

determining a second number of other distinct terms that are co-located terms of the first number of distinct terms; and

adding the second number to the first number to determine the respective measures of closeness.

9. The method of claim 1, further comprising determining the respective measures of prevalence of the respective semantic term groups including determining a number of occurrences of the respective semantic term groups in the document set.

10. The method of claim 1, further comprising weighting at least one of the respective measures of prevalence or the respective measures of closeness.

11. A system, comprising:

means for retrieving a document set in response to a query;

means for parsing a semantic distance network according to a set of terms in the document set into a set of semantic term groups;

means for determining respective term values for respective semantic term groups of the set of semantic term groups, wherein the respective term values comprise respective measures of prevalence of the respective semantic term groups as a function of respective measures of closeness of respective terms in the respective semantic term groups;

means for selecting a semantic term group of the respective semantic term groups as a category descriptor based on a term value associated with the semantic term group; and

means for outputting the semantic term group in response to the semantic term group being selected.

12. The system of claim 11, further comprising:

means for augmenting the semantic distance network with the semantic term group.

13. The system of claim 11, further comprising means for determining the respective measures of prevalence of the respective semantic term groups by determining a number of distinct terms of the plurality of terms that are co-located with the respective semantic term groups.

14. The system of claim 11, further comprising means for determining the respective measures of closeness of the respective terms of the respective semantic term groups by determining a first number of distinct terms of the plurality of terms that are co-located with at least two of the respective terms of the respective semantic term groups.

15. The system of claim 14, further comprising:

means for determining a second number of other distinct terms that are co-located with terms of the first number of distinct terms; and

means for adding the second number to the first number to determine the respective measures of closeness.

16. The system of claim 11, further comprising means for determining the respective measures of prevalence of the respective semantic term groups by determining a number of occurrences of the respective semantic term groups in the document set.

17. A non-transitory computer-readable medium having stored thereon computer executable instructions that, in response to execution by at least one computing device, cause the at least one computing device to perform operations comprising:

retrieving a document set in response to receiving a request;

parsing a semantic distance network according to a plurality of terms in the document set into a plurality of semantic term groups;

determining respective optimization values for respective semantic term groups of the plurality of semantic term groups, wherein the respective optimization values comprise respective measures of dominance of the respective semantic term groups divided by respective measures of closeness of respective terms in the respective semantic term groups;

selecting a semantic term group as a category descriptor based on an optimization value associated with the semantic term group; and

responding to the request with the semantic term group.

18. The non-transitory computer-readable medium of claim 17, wherein the operations further:

augmenting the semantic distance network with the semantic term group.

19. The non-transitory computer-readable medium of claim 17, wherein the operations further comprise determining the respective measures of dominance of the respective semantic term groups including determining a number of distinct terms of the plurality of terms that are co-located with the respective semantic term groups.

20. The non-transitory computer-readable medium of claim 17, wherein the operations further comprise determining the respective measures of closeness of the respective terms of the respective semantic term groups including determining a first number of distinct terms of the plurality of terms that are co-located with at least two of the respective terms of the respective semantic term groups.

21. The non-transitory computer-readable medium of claim 20, wherein the operations further comprise:

determining a second number of other distinct terms that are co-located with terms of the first number of distinct terms; and

adding the second number to the first number to determine the respective measures of closeness.

22. An apparatus, comprising:

a processor; and

a memory communicatively coupled to processor, the memory having stored therein computer-executable instructions, comprising;

a semantic parser configured to parse a semantic distance network according to a plurality of terms in a document set into a plurality of semantic term groups; and

a seed ranker configured to;

determine respective balanced desirability values for respective semantic term groups of the plurality of semantic term groups, wherein the respective balanced desirability values comprise respective measures of prevalence of the respective semantic term groups divided by respective measures of closeness of respective terms in the respective semantic term groups, and

select a semantic term group as a category descriptor based on a balanced desirability value associated with the semantic term group.

23. The apparatus of claim 22, wherein the seed ranker is further configured to augment the semantic distance network with the semantic term group.

24. The apparatus of claim 22, wherein the seed ranker is further configured to determine the respective measures of prevalence of the respective semantic term groups by a determination of a number of distinct terms of the plurality of terms that are co-located with the respective semantic term groups.

25. The apparatus of claim 22, wherein the seed ranker is further configured to determine the respective measures of closeness of the respective terms of the respective semantic term groups by a determination of a first number of distinct terms of the plurality of terms that are co-located with at least two of the respective terms of the respective semantic term groups.

26. The apparatus of claim 25, wherein the seed ranker is further configured to:

determine a second number of other distinct terms that are co-located with terms of the first number of distinct terms, and

add the second number to the first number to determine the respective measures of closeness.