Phrase-based detection of duplicate documents in an information retrieval system
An information retrieval system uses phrases to index, retrieve, organize and describe documents. Phrases are identified that predict the presence of other phrases in documents. Documents are the indexed according to their included phrases. Related phrases and phrase extensions are also identified. Phrases in a query are identified and used to retrieve and rank documents. Phrases are also used to cluster documents in the search results, create document descriptions, and eliminate duplicate documents from the search results, and from the index.
1. A method of detecting duplicate documents in search results, the method comprising:
receiving a query comprising at least one phrase;
retrieving a plurality of documents responsive to the query to form a search result, the retrieved documents being selected from a set of documents;
for each of the retrieved documents, generating, by operation of a processor within a computer system, a document description comprising selected sentences of the document, wherein the selected sentences are ordered in the document description as a function of a number of related phrases in each selected sentence, wherein a phrase g j is a related phrase of another phrase g k occurring in the set of documents when an information gain of g j with respect to g k exceeds a predetermined threshold;
responsive to the document description of at least two documents matching, discarding at least one of the two documents from the search result.
2. The method of claim 1 , wherein the information gain of g j with respect to g k is a function of both actual and expected co-occurrence rates of g j and g k in the set of documents.
3. The method of claim 1 , wherein the document description of a first document matches the document description of a second document when a hash value of the first document description equals a hash value of the second document description.
4. The method of claim 1 , further comprising:
concatenating the selected sentences of the document descriptions of each of the retrieved documents;
computing a hash value of the concatenated sentences of the document descriptions of each of the retrieved documents; and
comparing the hash values for the retrieved documents to determine if the document descriptions for two retrieved documents match.
5. The method of claim 1 , wherein a discarded document has a lower document significance measure than document whose document description matches the document description of the discarded document.
6. The method of claim 5 , wherein the document significance measure includes a page rank of the document.
7. The method of claim 1 , wherein the related phrases in the selected sentences of a document are related to other phrases of the document.
8. The method of claim 1 , wherein the related phrases in the selected sentences of a document are related to other phrases of the set of document.
9. The method of claim 1 , wherein the number (N) of selected sentences in the document description for the discarded document is identical to the number of selected sentences in the document description for a document whose document description matches the document description of the discarded document.
10. The method of claim 9 , wherein N is between 5 and 10.
11. A method of detecting duplicate documents in search results, the method comprising:
receiving a query comprising at least one phrase;
retrieving a plurality of documents responsive to the query to form a search result, the retrieved documents being selected from a set of documents;
for each of the retrieved documents, by operation of a processor within a computer system, retrieving a stored document description comprising selected sentences of the document, wherein the selected sentences are ordered in the document description as a function of a number of related phrases in each sentence, wherein a phrase g j is a related phrase of another phrase g k occurring in the set of documents when an information gain of g j with respect to g k exceeds a predetermined threshold;
responsive to the document description at least two documents matching, discarding at least one of the two documents from the search result.
12. The method of claim 11 , wherein the information gain of g j with respect to g k is a function of both actual and expected co-occurrence rates of g j and g k in the set of documents.
13. The method of claim 11 , wherein the document description of a first document matches the document description of a second document when a hash value of the first document description equals a hash value of the second document description.
14. The method of claim 11 , wherein a discarded document has a lower document significance measure than document whose document description matches the document description of the discarded document.
15. The method of claim 14 , wherein the document significance measure includes a page rank of the document.
16. The method of claim 11 , wherein the related phrases in the selected sentences of a document are related to other phrases of the document.
17. The method of claim 11 , wherein the related phrases in the selected sentences of a document are related to other phrases of the set of documents.
18. The method of claim 11 , wherein the number (N) of selected sentences in the document description for the discarded document is identical to the number of selected sentences in the document description for a document whose document description matches the document description of the discarded document.
19. The method of claim 18 , wherein N is between 5 and 10.
20. A tangible computer readable storage medium storing a computer program executable by a processor for detecting a duplicate document, the operations of the computer program comprising:
receiving a query comprising at least one phrase;
retrieving a plurality of documents responsive to the query to form a search result, the retrieved documents being selected from a set of documents;
for each of the retrieved documents, generating, by operation of a processor within a computer system, a document description comprising selected sentences of the document, wherein the selected sentences are ordered in the document description as a function of a number of related phrases in each selected sentence, wherein a phrase g j is a related phrase of another phrase g k occurring in the set of documents when an information gain of g j with respect to g k exceeds a predetermined threshold;
responsive to the document description of at least two documents matching, discarding at least one of the two documents from the search result.
21. The computer readable storage medium of claim 20 , wherein the information gain of g j with respect to g k is a function of both actual and expected co-occurrence rates of g j and g k in the set of documents.
22. The computer readable storage medium of claim 20 , wherein the document description of a first document matches the document description of a second document when a hash value of the first document description equals a hash value of the second document description.
23. The computer readable storage medium of claim 20 , wherein the operations of the computer program further comprise:
concatenating the selected sentences of the document descriptions of each of the retrieved documents;
computing a hash value of the concatenated sentences of the document descriptions of each of the retrieved documents; and
comparing the hash values for the retrieved documents to determine if the document descriptions for two retrieved documents match.
24. The computer readable storage medium of claim 20 , wherein a discarded document has a lower document significance measure than document whose document description matches the document description of the discarded document.
25. The computer readable storage medium of claim 24 , wherein the document significance measure includes a page rank of the document.
26. The computer readable storage medium of claim 20 , wherein the related phrases in the selected sentences of a document are related to other phrases of the document.
27. The computer readable storage medium of claim 20 , wherein the related phrases in the selected sentences of a document are related to other phrases of the set of document.
28. The computer readable storage medium of claim 20 , wherein the number (N) of selected sentences in the document description for the discarded document is identical to the number of selected sentences in the document description for a document whose document description matches the document description of the discarded document.
29. The computer readable storage medium of claim 28 , wherein N is between 5 and 10.
30. A tangible computer readable storage medium storing a computer program executable by a processor for detecting a duplicate document, the operations of the computer program comprising:
receiving a query comprising at least one phrase;
retrieving a plurality of documents responsive to the query to form a search result, the retrieved documents being selected from a set of documents;
for each of the retrieved documents, by operation of a processor within a computer system, retrieving a stored document description comprising selected sentences of the document, wherein the selected sentences are ordered in the document description as a function of a number of related phrases in each sentence, wherein a phrase g j is a related phrase of another phrase g k occurring in the set of documents when an information gain of g j with respect to g k exceeds a predetermined threshold;
responsive to the document description at least two documents matching, discarding at least one of the two documents from the search result.
31. The computer readable storage medium of claim 30 , wherein the information gain of g j with respect to g k is a function of both actual and expected co-occurrence rates of g j and g k in the set of documents.
32. The computer readable storage medium of claim 30 , wherein the document description of a first document matches the document description of a second document when a hash value of the first document description equals a hash value of the second document description.
33. The computer readable storage medium of claim 30 , wherein a discarded document has a lower document significance measure than document whose document description matches the document description of the discarded document.
34. The computer readable storage medium of claim 33 , wherein the document significance measure includes a page rank of the document.
35. The computer readable storage medium of claim 30 , wherein the related phrases in the selected sentences of a document are related to other phrases of the document.
36. The computer readable storage medium of claim 30 , wherein the related phrases in the selected sentences of a document are related to other phrases of the set of documents.
37. The computer readable storage medium of claim 30 , wherein the number (N) of selected sentences in the document description for the discarded document is identical to the number of selected sentences in the document description for a document whose document description matches the document description of the discarded document.
38. The computer readable storage medium of claim 37 , wherein N is between 5 and 10.