Method and system for retrieving advertisement information

• US 20080172422A1
• Filed: 07/12/2007
• Published: 07/17/2008
• Est. Priority Date: 07/12/2006
• Status: Active Grant

• Current Assignee: Baidu Incorporated
• Original Assignee: Baidu Incorporated
• Inventors: Liu, JianGuo, Wang, Zhan, Yuan, Jie, Zhu, Hongbo, Guo, Dan, Wang, Chuang, Liu, ZiZheng, Yang, Wenkai, Zhou, Limin, Li, Yanhong
• Granted Patent: US 7,788,216 B2
• Time in Patent Office: Days
• US Class Current: 707/202
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Method and system for retrieving advertisement information. Information characterizing a user'"'"'s past search behavior is utilized in creating index data that associate individual users to advertisements. When a search request from a user is received, the index data are utilized to identify one or more advertisements with respect to the received user'"'"'s search request. Such identified advertisements match the user'"'"'s past search behavior.

Claims

1. A system for retrieving advertisement information, comprising:

a log system configured for processing a log recording users'"'"' visits;

an indexing system configured for obtaining index data based on the log or processing result thereof; and

a retrieval system configured for identifying one or more advertisements with respect to a search request from a user based on the index data, wherein the one or more advertisements are selected to match the user'"'"'s past search behavior characterized at least partially by the log.

2. The system of claim 1, wherein the index data include a first type of index and a second type of index.

3. The system of claim 2, wherein the first type of index associates a user with one or more keywords used in at least one past query initiated by the user.

4. The system of claim 2, wherein the second type of index associates a keyword with one or more advertisements.

5. The system of claim 2, wherein at least one of first type of index and the second type of index can be evaluated based on a measure associated with a keyword in connection with the index to be evaluated.

6. The system of claim 5, wherein the measure is a weight.

7. The system of claim 1, wherein the log system comprises:

a log storage configured for storing the log;

a log processing module configured for analyzing the log to generate processed log information, wherein the processed log information includes at least a user identifier and one or more queries associated therewith.

8. The system of claim 7, wherein the log system further comprises:

an index generation module configured for generating additional index data associating a user identifier with at least one query initiated by the user; and

an index storage configured for storing the additional index data.

9. The system of claim 8, wherein the log system further comprises:

a time increment generation module configured for generating a time increment of at least a portion of the index data based on the processed log information; and

a time increment storage for storing the time increment.

10. The system of claim 9, wherein the index generation module generates an increment of the portion of the index data based on the time increment.

11. The system of claim 1, wherein the indexing system comprises:

a first index generation module configured for generating an increment of a first type of index;

a first index storage for storing the increment of the first type of index;

a second index generation module configured for generating a second type of index; and

a second index storage for storing the second type of index.

12. The system of claim 1, wherein the retrieval system comprises:

a first index storage configured for storing history data of a first type of index; and

an advertisement matching module configured for selecting the one or more advertisements with respect to the user given the search request that match the user'"'"'s past search behavior based on information stored in the first index storage.

13. The system of claim 12, further comprising a history data generation module configured for generating the history data of the first type of index, wherein the history data generation module updates the history data of the first type of index by merging an increment of the first type of index with the history data of the first type of index.

14. The system of claim 13, wherein the advertisement matching module selects the one or more advertisements by further basing on the increment of the first type of index.

15. The system of claim 12, further comprising:

a second index storage configured for storing a second type of index, wherein the advertisement matching module selects the one or more advertisements based on information stored in the first index database and the second index database.

16. The system of claim 1, wherein at least one of the log system, the indexing system, and the retrieval system is distributed.

17. The system of claim 16, wherein each of the distributed system is configured geographically distributed or logically distributed.

18. The system of claim 16, wherein each of the distributed system performs an operation with respect to a search request distributed thereto in accordance with an identifier of the user contained in the search request.

19. A method for retrieving advertisement information, comprising the steps of:

obtaining a log recording users'"'"' visits;

obtaining index data generated based on the log or processing result thereof; and

identifying one or more advertisements with respect to a search request from a user based on the index data, wherein the one or more advertisements are selected to match the user'"'"'s past search behavior characterized at least partially by the log.

20. The method of claim 19, wherein the index data include a first type of index and a second type of index.

21. The method of claim 20, wherein the first type of index associates a user with one or more keywords that the user used in at least one past query.

22. The method of claim 21, wherein the second type of index associates a keyword with one or more advertisements.

23. The method of claim 22, wherein at least one of first type of index and the second type of index can be evaluated based on a measure associated with a keyword in connection with the index to be evaluated.

24. The method of claim 23, wherein

a measure evaluating a first type of index corresponding to a user is derived based on a weight associated with a keyword used in at least one past query from the user;

or

a measure evaluating a second type of index is derived based on a weight associated with a keyword deemed to characterize an advertisement.

25. The method of claim 22, wherein the step of identifying comprises:

searching the first type of index based on an identifier of a user associated with the search request to obtain a corresponding keyword list;

searching the second type of index based on the keyword list to obtain the one or more advertisements.

26. The method of claim 20, wherein the first type of index is generated by:

generating an increment of the first type of index based on the log or the processing result thereof;

merging the increment of the first type of index with history data of the first type of index to produce an updated history data of the first type of index.

27. The method of claim 26, wherein the step of searching the first type of index comprises;

locating one or more instances of the first type of index containing the identifier of the user;

obtaining a keyword list from each of the instances;

forming the corresponding keyword list based on the keyword list from each of the instances, wherein

the one or more instances are identified based on one of the updated history data and a combination of the history data and the increment of the first type of index.

28. The method of claim 26, wherein the merging comprises:

multiplying a weight of each keyword in the history data by an attenuation factor to produce an attenuated weight for each keyword in the history data;

for each first user identifier of each first type of index in the increment having a corresponding user identifier in the history data, merging each first keyword list of the first user identifier in the increment with each corresponding keyword list of the corresponding user identifier in the history data to form a non-overlap keyword list of the first user identifier in the updated history data;

for each second user identifier of each first type of index in the increment that has no corresponding user identifier in the history data, inserting the second user identifier and a second keyword list of the second user identifier to create a new instance of the first type of index in the updated history data.

29. The method of claim 28, wherein the step of merging each keyword list comprises:

generating a merged weight for a first keyword in the non-overlap keyword list, which comes from both the first keyword list and the corresponding keyword list, by adding a first weight of the first keyword in the first keyword list in the increment and a corresponding attenuated weight of the first keyword in the corresponding keyword list in the history data;

generating a merged weight for a second keyword in the non-overlap keyword list, which comes from one of the first keyword list and the corresponding keyword list, by inserting the second keyword and a weight associated with the second keyword.

30. The method of claim 29, wherein the step of inserting the second user identifier and a second keyword list further comprises inserting one or more weights corresponding to one or more keywords included in the second keyword list of the second user identifier.

31. The method of claim 30, further comprising removing a keyword from the non-overlap keyword list when the corresponding merged weight is less than a preset threshold.

32. The method of claim 31, further comprising removing an instance of the first type of index corresponding to a user identifier when a corresponding keyword list associated with the user identifier

has a fewer keyword than a first predetermined threshold;

or

has all keywords having respective weights less than a second predetermined threshold.

33. A system for retrieving advertisement information, comprising:

a log system configured for obtaining a log recording users'"'"' visits;

an indexing system configured for obtaining a first type of index associating individual users with advertisements based on index data created based on the log or processing result thereof; and

a retrieval system configured for identifying one or more advertisements with respect to a search request from a user based on the first type of index, wherein the one or more advertisements match with the user'"'"'s past search behavior.

34. The system of claim 33, wherein the index data comprises a second type of index and a third type of index.

35. The system of claim 34, wherein:

the second type of index associates a keyword with one or more users who used the keyword in at least one past query;

the third type of index associates an advertisement with one or more keywords that to characterize the advertisement.

36. The system of claim 34, wherein at least one of the second type of index and the third type of index can be evaluated based on a measure associated with a keyword in connection with the index to be evaluated.

37. The system of claim 36, wherein the measure is a weight.

38. The system of claim 34, wherein the log system is further capable of obtaining a fourth type of index associating a user with one or more queries initiated by the user based on the log or the processing result thereof.

39. The system of claim 34, wherein

the log system is further capable of obtaining an increment of the second type of index based on the log or processing result thereof;

and

the increment of the second type of index is merged with existing second type of index to produce an updated second type of index.

40. The system of claim 34, wherein the indexing system comprises:

a first index generation module configured for generating the first type of index;

a first index storage for storing the generated first type of index;

a second index generation module configured for generating the second type of index;

a second index storage for storing the generated second type of index;

a third index generation module configured for generating the third type of index; and

a third index storage for storing the generated third type of index.

41. The system of claim 33, wherein the retrieval system comprises:

an interface module configured for

receiving the search request from the user,

initiating a search for advertisements in connection with the search request, and

responding the search request by providing the one or more advertisements; and

an advertisement matching module configured for

searching the first type of index based on an identifier of the user,

determining the one or more advertisements based on the search result;

providing the one or more advertisements to the interface module.

42. The system of claim 33, wherein at least one of the log system, the indexing system, and the retrieval system is distributed.

43. The system of claim 42, wherein each distributed system performs an operation with respect to a search request distributed thereto in accordance with information associated with the search request.

44. A method for retrieving advertisement information, comprising:

receiving a search request from a user; and

identifying one or more advertisements with respect to the user based on a first type of index associating individual users with advertisements, wherein

the first type of index is created based on information characterizing the user'"'"'s past search behavior via a log recording users'"'"' visits, and

the one or more advertisements are selected to match the user'"'"'s past search behavior.

45. The method of claim 44, wherein the information characterizing the user'"'"'s past search behavior includes a second type of index associating each keyword with one or more users who utilized the keyword in at least one past query.

46. The method of claim 44, wherein the first type of index is created further based on information characterizing advertisements, which includes a third type of index associating each advertisement with one or more keywords characterizing the advertisement.

47. A system for retrieving advertisement information, comprising:

a retrieval system configured for identifying one or more advertisements with respect to a search request from a user based on a first type of index associating individual users with advertisements, wherein

the first type of index is created based on information characterizing the user'"'"'s past search behavior via a log recording users'"'"' visits, and

the one or more advertisements are selected to match the user'"'"'s past search behavior.

48. A system for distributing advertisement information, comprising:

an advertisement distributing system configured for identifying one or more users with respect to an advertisement based on a first type of index associating advertisements with individual users, wherein

the first type of index is created based on information characterizing users'"'"' past search behavior via a log recording users'"'"' visits, and

the one or more users are selected so that the advertisement matches the user'"'"'s past search behavior.

49. The system of claim 48, wherein

the information characterizing users'"'"' past search behavior includes a second type of index associating each user with one or more keywords that are utilized in at least one past query initiated by the user;

and

the second type of index is obtained based on the log recording users'"'"' visits.

50. The method of claim 49, wherein the first type of index is created further based on information characterizing advertisements, including a third type of index associating each keyword with one or more advertisements where the keyword is considered relevant to the one or more advertisements.

51. A method for distributing advertisement information, comprising the steps of:

identifying one or more users with respect to an advertisement based on a first type of index associating advertisements with individual users, wherein

the first type of index is created based on information characterizing users'"'"' past search behavior via a log recording users'"'"' visits, and

the one or more users are selected so that the advertisement matches the user'"'"'s past search behavior.

52. The method of claim 51, wherein

the information characterizing users'"'"' past search behavior includes a second type of index associating each user with one or more keywords used in at least one past query initiated by the user and obtained based on the log recording users'"'"' visits;

and

the first type of index is created further based on information characterizing advertisements, including a third type of index associating each keyword with one or more advertisements where the keyword is considered relevant to the one or more advertisements.

53. A machine readable medium having information recorded thereon, the information, when read by the machine, enables the machine to perform the following:

receiving a search request from a user; and

identifying one or more advertisements with respect to the user based on a first type of index associating individual users with advertisements, wherein

the first type of index is created based on information characterizing the user'"'"'s past search behavior via a log recording users'"'"' visits, and

the one or more advertisements are selected to match the user'"'"'s past search behavior.

54. A machine readable medium having information recorded thereon, the information, when read by the machine, enables the machine to perform the following:

obtaining index data based on a log recording users'"'"' visits or processing result thereof; and

identifying one or more advertisements with respect to a search request from a user based on the index data, wherein

the index data are derived based on the log and indicative of the user'"'"'s past search behavior, and

the one or more advertisements are identified to match the user'"'"'s past search behavior.

55. A machine readable medium having information recorded thereon, the information, when read by the machine, enables the machine to perform the following:

obtaining a log recording users'"'"' visits;

obtaining index data based on a log recording users'"'"' visits or processing result thereof; and

identifying one or more advertisements with respect to a search request from a user based on the index data, wherein

the index data are derived based on the log and indicative of the user'"'"'s past search behavior, and

the one or more advertisements are identified to match the user'"'"'s past search behavior.

56. A machine readable medium having information recorded thereon, the information, when read by the machine, enables the machine to perform the following:

identifying one or more users with respect to an advertisement based on a first type of index associating advertisements with individual users, wherein

the first type of index is created based on information characterizing users'"'"' past search behavior via a log recording users'"'"' visits, and

the one or more users are selected so that the advertisement matches the user'"'"'s past search behavior.