Query-based snippet clustering for search result grouping

• US 20060026152A1
• Filed: 07/13/2004
• Published: 02/02/2006
• Est. Priority Date: 07/13/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Ma, Wei-Ying, Chen, Zheng, Zhang, Benyu, Zeng, Hua-Jun, He, Qicai, Liu, Guimei
• Granted Patent: US 7,617,176 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: Y10S 707/99931 Database or file accessing

Abstract

A clustering architecture that dynamically groups the search result documents into clusters labeled by phrases extracted from the search result snippets. Documents related to the same topic usually share a common vocabulary. The words are first clustered based on their co-occurrences and each cluster forms a potentially interesting topic. Keywords are chosen and then clustered by counting co-occurrences of pairs of keywords. Documents are assigned to relevant topics based on the feature vectors of the clusters.

Claims

1. A system that facilitates the processing of search results, comprising:

an input component that receives search results;

an analysis component that extracts content of the search results; and

a grouping component that groups the search results according to a ranked saliency parameter.

2. The system of claim 1, the search results are a ranked list of results.

3. The system of claim 1, the search results are titles and snippets that are associated with documents.

4. The system of claim 1, the analysis component utilizes a frequent itemset algorithm to identify a content of the search results.

5. The system of claim 1, the analysis component applies a regression model that facilitates extraction and ranking of the content from the search results.

6. The system of claim 1, the content is analyzed according to properties that include at least one of term frequency/inverted document frequency, phrase length, intra-cluster similarity, cluster entropy, and phrase independence.

7. A computer readable medium having stored thereon computer executable instructions for carrying out the system of claim 1.

8. A computer that employs the system of claim 1.

9. A search engine according to the system of claim 1.

10. The system of claim 1, the saliency parameter is a salient score generated according to one or more properties calculated on the content.

11. A system that facilitates the processing of search results, comprising:

an input component that receives search results of a search engine in the form of at least one of titles and snippets;

an analysis component that extracts words from the at least one of titles and snippets, and identifies co-occurring words as keywords; and

a clustering component that groups the search results in a cluster according to a saliency score, and outputs the search results as a ranked list of associated documents.

12. The system of claim 11, the list of documents is presented to a user in response to selection of a cluster.

13. The system of claim 11, a document is assigned to the cluster based upon its similarity to relevant keywords of the cluster.

14. The system of claim 11, the cluster is named according to a keyword or phrase that includes multiple keywords.

15. A computer-readable medium having computer-executable instructions for performing a method for processing search results, the method comprising:

receiving the search results;

extracting phrases from the search results;

ranking the phrases to obtain salient phrases; and

dynamically grouping the search results as a cluster that is labeled by the salient phrases.

16. The method of claim 15, the search results are returned by a web search engine.

17. The method of claim 15, further comprising analyzing the search results using an HTML parser.

18. The method of claim 15, further comprising calculating properties for each of the search results during a parsing process.

19. The method of claim 15, further comprising generating a candidate cluster of documents that contain the phrases.

20. The method of claim 15, the search results contain at least one of titles and snippets, wherein the titles and snippets are weighted differently.

21. The method of claim 20, the titles are weighted greater than the snippets

22. The method of claim 15, further comprising applying a regression model to the properties calculated for each phrase to combine the properties into a single salient score.

23. The method of claim 15, the phrases are ranked in descending order according to a salient score.

24. The method of claim 15, the act of grouping includes associating the salient phrases with documents.

25. The method of claim 15, further comprising,

merging a first cluster and a second cluster into a third cluster, when overlap of the first and second clusters exceeds a predetermined threshold;

adjusting cluster names of the merged clusters to generate a new cluster name for the third cluster; and

presenting topmost clusters to a user.

26. The method of claim 15, further comprising automatically presenting an associated document list of a cluster when the user selects the cluster.

27. The method of claim 26, wherein at least one of query words and salient phrases are highlighted when presented to the user.

28. A method of processing search results, comprising:

receiving the search results as a list that includes at least one of titles and snippets of information;

parsing the at least one of titles and snippets;

extracting phrases from content of the at least one of titles and snippets;

calculating properties for each of the phrases;

applying a regression model to combine the properties of a phrase into a salient score;

ranking the phrases according to the salient score to obtain salient phrases;

assigning documents to the salient phrases to generate candidate clusters; and

generating final clusters by merging the candidates clusters.

29. The method of claim 28, further comprising generating short meaningful cluster names for the final clusters based on the salient phrases.

30. The method of claim 28, the properties include at least one of phrase frequency, document frequencies, and phrase length.

31. A system that facilitates the processing of search results, comprising:

means for receiving the search results that contain at least one of titles and snippets of information;

means for extracting phrases from the search results;

means for calculating one or more properties for each of the phrases;

means for processing the one or more properties into a salient score for each phrase;

means for ranking the phrases according to the salient scores to obtain salient phrases; and

means for dynamically grouping the search results as a cluster that is labeled by the salient phrases.

32. The system of claim 31, the means for processing includes at least one of a trained support vector regression model and a classification system.

33. The system of claim 31, further comprising means for presenting a list of relevant documents that are associated with the salient phrases, which list of relevant documents is presented in response to selection of the cluster.