Reranking and increasing the relevance of the results of Internet searches

• US 7,574,436 B2
• Filed: 03/10/2005
• Issued: 08/11/2009
• Est. Priority Date: 03/10/2005
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Parikh, Jignashu, Kapur, Shyam
• Primary Examiner(s): Vy; Hung T
• Application Number:
  US11/078,685
• Publication Number:
  US 20060206476A1
• Time in Patent Office: 1,615 Days
• Field of Search: 707 3- 5, 707/1, 704/240, 704/251
• US Class Current: 1/1
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

Techniques are provided for reranking and increasing the relevance of the results Internet searches. A search query is initially parsed into individual units. Each unit corresponds to one or more words that represents a natural concept. A concept network is analyzed to locate concepts that are related to the units in the search query. Particular concepts are selected from the concept network. Independent Internet searches are performed for each of the selected concepts. The search results from these searches are then compared to units in the original search query and ranked according to their relevance to the original search query.

Claims

1. A method for increasing the relevance of search results retrieved from searches, the method comprising the machine-implemented steps of:

identifying one or more units in a search query, wherein each of the one or more units corresponds to one concept in the search query;

selecting one or more concepts in a concept network that are related to the one or more units in the search query, wherein the concept network comprises a plurality of concepts;

wherein selecting the one or more concepts includes determining how frequently a subset of the plurality of concepts appeared in previous search queries that were submitted by a plurality of users, wherein the one or more concepts appeared more frequently in the previous search queries relative to other concepts in the concept network;

using the search query to perform a search;

sorting search results retrieved from the search based on a relevance of the search results to the search query, wherein each of the search results includes a title and an abstract of the search result; and

reranking the search results based on occurrences of the one or more concepts in the title or abstract of the each of the search results.

2. The method according to claim 1 wherein selecting one or more of the related concepts from the concept network further comprises:

selecting a predetermined number of the related concepts from among the concepts in the concept network that are most related to the one or more units.

3. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 2.

4. The method according to claim 1 wherein selecting one or more of the related concepts from the concept network further comprises:

selecting all of the related concepts that are directly linked to the one or more units in the concept network.

5. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 4.

6. The method according to claim 1 wherein selecting one or more of the related concepts from the concept network further comprises:

selecting the related concepts in the concept network that are synonyms of the one or more units.

7. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 6.

8. The method according to claim 1 wherein sorting the search results retrieved from the search further comprises:

assigning a rerank score to each of the search results based on how frequently the one or more units in the search query appear in each of the search results.

9. The method according to claim 8 wherein the rerank scores each include a contribution from a rank score assigned by a search engine that performs the searches for each of the related concepts.

10. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 9.

11. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 8.

12. The method according to claim 1 wherein sorting the search results retrieved from the search further comprises:

assigning a rerank score to each of the search results based on how frequently the one or more units, associated units, and extension units in the search query appear in each of the search results.

13. The method according to claim 12 wherein sorting the search results retrieved from the search further comprises:

sorting the search results based on values of the rerank scores.

14. The method according to claim 13 wherein sorting the search results retrieved from the search further comprises ranking navigational queries more highly than search results retrieved from other types of queries.

15. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 14.

16. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 12.

17. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 13.

18. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 1.

19. A method for increasing the relevance of search results retrieved from searches, the method comprising the machine-implemented steps of:

identifying one or more units in a search query;

selecting one or more concepts in a concept network that are related to the one or more units in the search query, wherein the concept network comprises a plurality of concepts;

wherein selecting the one or more concepts includes determining how frequently a subset of the plurality of concepts appeared in previous search queries that were submitted by a plurality of users, wherein the one or more concepts appeared more frequently in the previous search queries relative to other concepts in the concept network;

using the search query to perform a first search to retrieve a first plurality of search results;

using at least one of the one or more concepts to perform a second search to retrieve a second plurality of search results; and

interleaving one or more of the second plurality of search results into the first plurality of search results.

20. The method of claim 19, wherein:

a result set is a set of search results that will be displayed to a user;

the result set comprises at least a subset of the first plurality of search results and the one or more of the second plurality of search results;

the one or more of the second plurality of search results are one or more particular search results;

interleaving the one or more of the second plurality of search results into the first plurality of search results includes positioning the one or more particular search results at one or more pre-determined positions in the result set;

at least one of the first plurality of search results, when displayed to the user, appears above at least one of the one or more particular search results; and

at least one of another of the first plurality of search results, when displayed to the user, appears below the at least one of the one or more particular search results.

21. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 20.

22. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 19.

23. A method for increasing the relevance of search results retrieved from searches, the method comprising the machine-implemented steps of:

identifying one or more units in a search query;

selecting one or more concepts in a concept network that are related to the one or more units in the search query, wherein the concept network comprises a plurality of concepts;

wherein selecting the one or more concepts includes determining how frequently a subset of the plurality of concepts appeared in previous search queries that were submitted by a plurality of users, wherein the one or more concepts appeared more frequently in the previous search queries relative to other concepts in the concept network;

using the search query to perform a search to retrieve a first plurality of search results;

using at least one of the one or more concepts to perform a second search to retrieve a second plurality of search results;

determining a click-through rate for each of the first plurality of search results; and

in response to determining that one or more first search results of the first plurality of search results are associated with a lower click-through rate relative to the other search results of the first plurality of search results;

excluding the first one or more search results from a search result set that will be displayed in a search results page, and

replacing, in the search result set, the first one or more search results with one or more search results of the second plurality of search results.

24. A volatile or non-volatile memory medium storing program code that, when executed by one or more processors, causes the one or more processors to perform the steps recited in claim 23.