System and method for searching and recommending documents in a collection using share bookmarks

• US 6,493,702 B1
• Filed: 05/05/1999
• Issued: 12/10/2002
• Est. Priority Date: 05/05/1999
• Status: Expired due to Term

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Breuel, Thomas M., Adar, Eytan, Pitkow, James E., Cass, Todd A., Schuetze, Hinrich
• Primary Examiner(s): Breene, John
• Assistant Examiner(s): Black, Linh
• Application Number:
  US09/305,844
• Time in Patent Office: 1,315 Days
• Field of Search: 707/1, 707/2, 707/10, 707/104, 707/500, 707/3, 707/6, 707/100-104, 345/329, 345/330, 345/331, 345/332, 345/333-335, 345/352, 709/203, 709/217, 709/305
• US Class Current: 707/706
• CPC Class Codes: G06F 16/9562 Bookmark management
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

A search and recommendation system employs the preferences and profiles of individual users and groups within a community of users, as well as information derived from shared document bookmarks, to augment Internet searches, re-rank search results, and provide recommendations for documents based on a subject-matter query. The search and recommendation system operates in the context of a shared bookmark manager, which stores individual users'"'"' bookmarks (some of which may be published or shared for group use) on a centralized bookmark database connected to the Internet. The shared bookmark manager is implemented as a distributed program, portions of which operate on users'"'"' terminals and other portions of which operate on the centralized bookmark database.

Claims

1. A method of adding a content pointer to a collection of shared content pointers contributed by a plurality of users, wherein each shared content pointer is associated with a content source, the content pointers organized into a plurality of categories, comprising the steps of:

receiving a data item identifying a content source;

determining a category of the collection of shared content pointers in which to store the data item comprising the substeps of;

identifying a topic corresponding to the content source;

comparing the topic to a first list of topics in a user-specific hierarchy of categories;

if the topic is in the first list, identifying the category within the user-specific hierarchy corresponding to the topic;

if the topic is not in the first list,

comparing the topic to a second list of topics in a global hierarchy of categories;

if the topic is in the second list,

identifying the category within the global hierarchy corresponding to the topic; and

establishing a category within the user-specific hierarchy corresponding to the topic;

if the topic is not in the second list,

establishing a category within the global hierarchy corresponding to the topic; and

establishing a category within the user-specific hierarchy corresponding to the topic;

wherein the global hierarchy of categories is initially determined by the steps of;

for each content pointer in the collection of shared document content pointers, identifying a content vector corresponding to the content pointer;

clustering the content vectors into a plurality of clusters; and

recursively repeating the clustering step on each cluster in the plurality of clusters to derive a hierarchy of clusters; and

storing the hierarchy of clusters as the global hierarchy of categories; and

storing the data item as a content pointer in the determined category of the collection.

2. A system for maintaining a collection of shared document content pointers contributed by a plurality of users, wherein each shared document content pointer represents a document in a document repository, comprising:

a plurality of user terminals in communication with the repository;

a database in communication with the repository and the plurality of user terminals, wherein the database is adapted to store the collection of shared document content pointers; and

software adapted to maintain the collection of shared document content pointers.

3. The system of claim 2, wherein each user terminal includes a browser.

4. The system of claim 2, wherein a first portion of the software is executed by the database, and a second portion of the software is executed by each user terminal.

5. The system of claim 4, wherein the second portion of the software is downloaded on demand from the database to a user terminal.

6. The system of claim 5, wherein at least a portion of the collection of shared document content pointers is downloaded on demand from the database to a user terminal.

7. The system of claim 6, wherein the portion of the collection of shared document content pointers is uploaded to the database from the user terminal upon completion of a session of use.

8. The system of claim 6, wherein the portion of the collection of shared document content pointers is uploaded to the database from the user terminal on demand.

9. The system of claim 6, wherein a user command received by the second portion is transmitted from the user terminal to the database for processing.

10. The system of claim 9, wherein the portion of the collection of shared document content pointers is uploaded to the database from the user terminal together with the user command.

11. The system of claim 2, wherein the software adapted to maintain the collection of shared document content pointers is also adapted to maintain a plurality of user-specific document content pointer sets corresponding to the plurality of users.

12. The system of claim 11, wherein the software is adapted to publish a content pointer from a user-specific document content pointer set to the collection of shared document content pointers upon request by a user.

13. The system of claim 2, wherein the collection of shared document content pointers comprises a plurality of shared content pointers arranged in a global hierarchy of categories.

14. The system of claim 4, wherein the first portion is adapted to monitor the availability of each shared document content pointer in the collection of shared document content pointers.

15. The system of claim 14, wherein if a shared document content pointer is unavailable, the first portion is adapted to generate a recommendation for a similar document chosen from the collection of shared document content pointers.

16. A method for generating a recommendation from a collection of shared document content pointers contributed by a plurality of users, wherein each shared document content pointer represents a document, comprising the steps of:

identifying at least one unavailable content pointer;

searching at least a portion of the collection of shared document content pointers; and

retrieving a group of content pointers that match the unavailable content pointer.

17. A method for searching an information repository comprising the steps of:

establishing a collection of shared content pointers, the content pointers contributed by and shared among a plurality of users;

receiving at least one subject keyword;

searching at least a portion of the collection of shared content pointers; and

retrieving a group of content pointers that match the subject keyword;

identifying a context within the collection of shared content pointers;

ranking the group of content pointers based a computed match with the context; and

wherein the context comprises a profile for a group of users chosen from the plurality of users.

18. The method of claim 17, wherein the context comprises a profile for a single user chosen from the plurality of users.

19. The method of claim 18, wherein the profile comprises a content vector derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by the single user.

20. The method of claim 17, wherein the profile comprises a content vector derived from at least one document represented by at least one shared content pointer contributed by the group.

21. The method of claim 17, wherein the profile comprises a content vector derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by the group.

22. The method of claim 17, wherein the context comprises a profile for the plurality of users, said profile comprising a content vector derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by the plurality of users.

23. The method of claim 17, further comprising the step of ordering the group of retrieved content pointers in a ranking order, the order determined by a weighted set of user metrics.

24. The method of claim 23, wherein the set of user metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

25. The method of claim 24, wherein the step of identifying the popularity of each content pointer in the group of content pointers comprises determining what fraction of users in the plurality of users contributed the content pointer.

26. The method of claim 17, further comprising the steps of:

identifying a context within the collection of shared content pointers;

identifying a set of user metrics, the set of user metrics chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric;

combining a computed match with the context, and at least one of the set of user metrics, for each content pointer into a composite measure; and

ranking the group of content pointers based on the composite measure.

27. The method of claim 26, further comprising the steps of:

identifying a group of users comprising at least one user selected from the plurality of users; and

wherein, the set of user metrics are identified with respect to the group of users.

28. A method for searching an information repository comprising the steps of:

establishing a collection of content pointers, the content pointers contributed by a plurality of users;

receiving at least one query keyword;

augmenting the query with at least one additional keyword derived from the collection;

searching the information repository; and

retrieving at least one of a group of documents, each including content indicia, at least one content index matching the query keyword.

29. The method of claim 28, wherein the augmenting step comprises the steps of:

identifying a context within the collection; and

identifying at least one additional keyword within the context.

30. The method of claim 29, wherein the context comprises a profile, the profile selected from a group consisting of a user profile for a single user chosen from the plurality of users, a group profile for a group of users chosen from the plurality of users, and a profile for the plurality of users.

31. The method of claim 30, wherein the profile is derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by at least one user.

32. The method of claim 28, further comprising the step of ordering the retrieved documents in a ranking order, the order determined by a weighted set of user metrics.

33. The method of claim 32, wherein the set of user metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

34. The method of claim 33, wherein the step of identifying the popularity of each document in the group of documents comprises determining what fraction of users in the plurality of users contributed a content pointer representative of the document.

35. The method of claim 33, wherein the link structure comprises a computed inlink weight combined with a computed outlink weight.

36. The method of claim 29, further comprising the step of re-ranking the group of content pointers based on a spreading activation analysis of the group of content pointers.

37. The method of claim 28, further comprising the steps of:

identifying a context within the collection of shared content pointers;

identifying a set of user metrics, the set of user metrics chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric;

combining a computed match with the context, and at least one of the set of user metrics, for each content pointer into a composite measure; and

ranking the group of content pointers based on the composite measure.

38. The method of claim 37, further comprising the steps of:

identifying a group of users comprising at least one user selected from the plurality of users; and

wherein, the set of user metrics are identified with respect to the group of users.

39. A method for searching an information repository comprising the steps of:

establishing a collection of content pointers, the content pointers contributed by a plurality of users;

receiving at least one query keyword;

searching the information repository;

identifying a context within the collection;

retrieving at least one of a group of documents, each including content indicia, at least one content index matching the query keyword, wherein the retrieved at least one of a group of documents is ranked in accordance with a computed match with the context; and

wherein the context comprises a profile, the profile selected from a group consisting of a user profile for a single user chosen from the plurality of users, a group profile for a group of users chosen from the plurality of users,

and a profile for the plurality of users.

40. The method of claim 39, wherein the profile is derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by at least one user.

41. The method of claim 39, further comprising:

identifying at least one additional keyword within the context;

augmenting the query with the at least one additional keyword derived from the collection to define an augmented query keyword;

searching the information repository; and

retrieving at least one of a group of documents, each including content indicia, at least one content index matching the augmented query keyword.

42. The method of claim 41, wherein the augmented query keyword is derived from the collection context.

43. The method of claim 39, further comprising the step of ordering the retrieved documents in a ranking order, the order determined by a weighted set of user metrics.

44. The method of claim 43, wherein the set of user metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

45. The method of claim 44, wherein the step of identifying the popularity of each document in the group of documents comprises determining what fraction of users in the plurality of users contributed a content pointer representative of the document.

46. A method for searching an information repository comprising the steps of:

establishing a collection of content pointers, the content pointers contributed by a plurality of users;

receiving at least one query keyword;

searching the information repository;

retrieving at least one of a group of documents, each including content indicia, at least one content index matching the query keyword; and

ordering the retrieved documents in a ranking order, the order determined by a weighted set of user metrics.

47. The method of claim 46, wherein the set of user metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

48. The method of claim 47, wherein the step of identifying the popularity of each document in the group of documents comprises determining what fraction of users in the plurality of users contributed a content pointer representative of the document.

49. A method for generating ordered recommendations of content sources from an information repository comprising the steps of:

establishing a collection of content pointers, the content pointers contributed by a plurality of users, each content pointer associated with a content source;

receiving at least one query keyword;

searching the information repository;

retrieving a group of content sources that each includes a content profile matching the query keyword;

identifying a context within the collection; and

ranking the group of content sources based on a computed match between each content source'"'"'s content profile and the identified context.

50. The method of claim 49, wherein the retrieved group of content sources is ranked in accordance with a computed match with the context.

51. The method of claim 50, wherein the context comprises a profile, the profile selected from a group consisting of a user profile for a single user chosen from the plurality of users, a group profile for a group of users chosen from the plurality of users, and a profile for the plurality of users.

52. The method of claim 51, wherein the profile is derived from at least one document represented by at least one content pointer in at least one selected topical category of content pointers contributed by at least one user.

53. The method of claim 49, wherein the ranking step further comprises the step of ordering the group of retrieved content sources in accordance with a set of user recommendation priority metrics.

54. The method of claim 53, wherein the user recommendation priority metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

55. The method of claim 49, further comprising the steps of:

identifying at least one associate keyword from within the context;

augmenting the query with at least one associate keyword derived from the collection;

searching the information repository; and

retrieving at least one of a group of documents, each including a content profile matching the augmented query keyword.

56. The method of claim 55, wherein the augmented query keyword is derived from the collection context.

57. A method for generating ordered recommendations of content sources from an information repository comprising the steps of:

establishing a collection of content pointers, the content pointers contributed by a plurality of users, each content pointer associated with a content source;

receiving at least one query keyword;

searching the information repository;

retrieving a group of content sources that match the query keyword;

identifying a context within the collection;

identifying a set of user recommendation priority metrics;

combining a computed match with the context and the set of user recommendation priority metrics for each content source into a composite recommendation metric; and

ranking the group of content sources based on each content source'"'"'s composite recommendation metric.

58. The method according to claim 57 further comprising the steps of:

identifying a group of users comprising at least one user selected from the plurality of users;

identifying the set of user recommendation priority metrics with respect to the group; and

wherein the user recommendation priority metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

59. An information repository search engine comprising:

a collection of content pointers, the content pointers contributed by a plurality of users;

a user input engine, configured to receive at least a keyword query input;

a query augmentation engine, configured to derive an associate keyword from the collection; and

a search engine, configured to search the collection of shared content pointers.

60. The information repository search engine of claim 59, wherein the query augmentation engine combines the identified associate keyword with the keyword query input to define an enhanced keyword vector.

61. The information repository search engine of claim 60, further comprising a retrieval engine, the retrieval engine configured to retrieve at least one of a group of content sources, each identified by a content pointer and each including content indicia, at least one content index matching the enhanced keyword vector.

62. The information repository search engine of claim 59, further comprising a contextualization engine, the contextualization engine identifying a context within the collection, wherein the associate keyword is derived from the context.

63. The information repository search engine of claim 61, further comprising a recommendation engine, the recommendation engine ordering a set of retrieved content sources in accordance with a set of user recommendation metrics, the user recommendation metrics chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

64. In a search engine of the type configured to search an information repository on the basis of at least a query keyword input by a user, the search engine retrieving at least one of a group of content sources, each content source including content indicia, a recommendation engine comprising:

a collection of content pointers, the content pointers contributed by a plurality of users;

a metrics identification engine, the metrics identification engine identifying a set of user recommendation metrics; and

an ordering engine, the ordering engine ordering a set of retrieved content sources in accordance with the set of user recommendation metrics.

65. The recommendation engine of claim 64, wherein the user recommendation metrics are chosen from the group consisting of a popularity metric, a frequency of access metric, a recency of access metric, and a link structure metric.

66. The recommendation engine of claim 64, further comprising:

a contextualization engine, the contextualization engine identifying a context within the collection; and

wherein the ordering engine orders a set of retrieved content sources in accordance with a computed match between the context and each content source'"'"'s content indicia.

67. The recommendation engine of claim 66, further comprising:

combining means for computing a match between the context, and at least one of the set of user metrics, for each content pointer into a composite measure; and

wherein the ordering engine orders a set of retrieved content sources in accordance with the computed match between the composite measure and each content source'"'"'s content indicia.