RANKING DOCUMENTS WITH SOCIAL TAGS

• US 20100169324A1
• Filed: 12/30/2008
• Published: 07/01/2010
• Est. Priority Date: 12/30/2008
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Meyerzon, Dmitriy, Shnitko, Yauhen
• Granted Patent: US 8,914,359 B2
• Time in Patent Office: Days
• US Class Current: 707/748
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• 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: G06F 16/951 Indexing; Web crawling techniques

Abstract

Technologies are described herein for ranking documents with social tags. A number ranking feature containing a number of times a document was tagged is received. A textual property ranking feature containing a union of each social tag associated with the document is also received. The number ranking feature is transformed into a static input value. Further, the textual property ranking feature is transformed into a dynamic input value. A document rank for the document is determined by inputting the static input value and/or the dynamic input value into a ranking function.

Claims

1. A computer-implemented method for ranking documents with social tags, the method comprising:

receiving a number ranking feature containing a number of times a document was tagged;

receiving a textual property ranking feature containing a union of each social tag associated with the document;

transforming the number ranking feature into a static input value;

transforming the textual property ranking feature into a dynamic input value; and

determining a document rank for the document by inputting at least one of the static input value or the dynamic input value into a ranking function.

2. The computer-implemented method of claim 1, further comprising:

receiving a query;

identifying a plurality of documents that satisfy the query; and

sorting the plurality of documents according to their corresponding document rank.

3. The computer-implemented method of claim 2, wherein identifying a plurality of documents that satisfy the query comprises identifying the plurality of documents where the corresponding textual property ranking feature for each document contains each term in the query.

4. The computer-implemented method of claim 1, wherein the number ranking feature contains the number of times the document was tagged by one or more users.

5. The computer-implemented method of claim 1, wherein the textual property ranking feature contains the social tags from one or more users.

6. The computer-implemented method of claim 1, wherein the ranking function comprises BM25F;

and the method further comprising inputting an output of BM25F into a two-layer neural network ranking function.

7. The computer-implemented method of claim 1, wherein transforming the textual property ranking feature into a dynamic input value comprises inputting textual property information associated with the textual property ranking feature into a BM25F ranking function to generate the dynamic input value.

8. The computer-implemented method of claim 1, wherein the textual property information comprises a number of times each term in a query appears in the textual property ranking feature, a length of the textual property ranking feature, and an average length of a plurality of textual property ranking features across a corpus.

9. The computer-implemented method of claim 1, wherein the social tags comprise one or more terms associated by users to the document.

10. A computer system comprising:

a processor;

a memory operatively coupled to the processor; and

a program module (i) which executes in the processor from the memory and (ii) which, when executed by the processor, causes the computer system to rank documents with social tags by;

crawling a profile database to obtain the social tags associated with a document;

determining a number ranking feature containing a number of times the document was tagged;

determining a textual property ranking feature containing a union of the social tags associated with the document;

transforming the number ranking feature into a static input value;

transforming the textual property ranking feature into a dynamic input value; and

determining a document rank for the document by inputting at least one of the static input value or the dynamic input value into a ranking function.

11. The computer system of claim 10, wherein the profile database comprises a plurality of user profiles, each user profile comprising the social tags associated to the document by a user;

and wherein crawling the profile database to obtain the social tags associated with a document comprises crawling the profile database to obtain social tags associated with the document from the plurality of user profiles.

12. The computer system of claim 11, wherein crawling a profile database to obtain the social tags associated with a document comprises crawling the user profiles marked as changed to indicate that the social tags have been added, modified, or deleted.

13. The computer system of claim 10, wherein the ranking function comprises BM25F;

and the program module which, when executed by the processor, further causes the computer system to rank documents with social tags by inputting an output of BM25F into a two-layer neural network ranking function.

14. The computer system of claim 10, wherein transforming the textual property ranking feature into a dynamic input value comprises inputting textual property information associated with the textual property ranking feature into a BM25F ranking function to generate the dynamic input value.

15. The computer system of claim 10, wherein the textual property information comprises a number of times each term in a query appears in the textual property ranking feature, a length of the textual property ranking feature, and an average length of a plurality of textual property ranking features across a corpus.

18. The computer-readable medium of claim 15 having further computer-executable instructions stored thereon which, when executed by a computer, cause the computer to:

receive a query;

identify a plurality of documents that satisfy the query where the corresponding textual property ranking feature for each document contains each term in the query; and

sort the plurality of documents according to their corresponding document rank.

19. The computer-readable medium of claim 15, wherein the ranking function comprises BM25F;

and the computer readable medium having further computer-executable instructions stored thereon which, when executed by the computer, cause the computer to input an output of BM25F into a two-layer neural network ranking function.

20. The computer-readable medium of claim 15, wherein to transform the textual property ranking feature into the dynamic input value, the computer-readable medium having further computer-executable instructions stored thereon which, when executed by a computer, cause the computer to input textual property information associated with the textual property ranking feature into a BM25F ranking function to generate the dynamic input value.

16. A computer-readable medium having computer-executable instructions stored thereon which, when executed by a computer, cause the computer to:

crawl documents to obtain document ranking features;

crawl a profile database to obtain social tags associated with a document, the profile database comprising a plurality of user profiles associated with a plurality of users, each user profile comprising the social tags associated to the document by one of the plurality of users;

determine a number ranking feature containing a number of times the document was tagged by the plurality of users;

determine a textual property ranking feature containing a union of the social tags associated with the document by the plurality of users;

transform the number ranking feature into a static input value that is query independent;

transform the textual property ranking feature into a dynamic input value that is query dependent; and

determine a document rank for the document by inputting at least one of the static input value or the dynamic input value into a ranking function, and further inputting the document ranking features into the ranking function.

17. The computer-readable medium of claim 16, wherein the document are privately accessible only by a subset of the plurality of users.