Customer self service system for resource search and selection

• US 6,701,311 B2
• Filed: 02/07/2001
• Issued: 03/02/2004
• Est. Priority Date: 02/07/2001
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Jasura, Donn P., Keller, Neal M., Biebesheimer, Debra L., Oblinger, Daniel A., Podlaseck, Mark E., Rolando, Stephen J.
• Primary Examiner(s): Alam, Shahid Al
• Assistant Examiner(s): EHICHIOYA, IRETE FRED
• Application Number:
  US09/778,146
• Publication Number:
  US 20020107842A1
• Time in Patent Office: 1,119 Days
• Field of Search: 705/5, 705/6, 705/8, 705/9, 705/14, 705/1, 705/35, 705/27, 707/3, 707/10, 707/7, 707/201, 707/104, 707/100, 709/204, 709/206, 709/248, 709/214, 709/278
• US Class Current: 1/1
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99952 Coherency, e.g. same view to multiple users

Abstract

A customer self service system and method for performing resource search and selection. The method includes steps of providing an interface enabling entry of a query for a resource and specification of one or more user context elements, each element representing a context associated with the current user state and having context attributes and attribute values associated therewith; enabling user specification of relevant resource selection criteria for enabling expression of relevance of resource results in terms of user context; searching a resource database and generating a resource response set having resources that best match a user'"'"'s query, user context attributes and user defined relevant resource selection criteria; presenting said resource response set to the user in a manner whereby a relevance of each of the resources being expressed in terms of user context in a manner optimized to facilitate resource selection; and, enabling continued user selection and modification of context attribute values to enable increased specificity and accuracy of a user'"'"'s query to thereby result in improved selection logic and attainment of resource response sets best fitted to the query. More particularly, adaptive algorithms and supervised and unsupervised learning sub-processes are implemented to enable the self service resource search and selection system to learn from each and all users and make that learning operationally benefit all users over time.

Claims

1. A customer self service system for performing resource search and selection comprising:

a mechanism enabling entry of a query for a resource and, entry of one or more user context elements, each element representing a context associated with the current user state and having context attributes and attribute values associated therewith, said mechanism further enabling user specification of relevant resource selection criteria for enabling expression of relevance of resource results in terms of user context;

a mechanism for searching a resource database and generating a resource response set having resources that best match a user'"'"'s query, user context attributes and user defined relevant resource selection criteria, said resource response set being presented to said user in a manner whereby a relevance of each said resources being expressed in terms of user context in a manner optimized to facilitate resource selection; and

,

a mechanism for enabling continued user selection and modification of context attribute values to enable increased specificity and accuracy of a user'"'"'s query to thereby result in improved selection logic and attainment of resource response sets best fitted to said query.

2. The customer self service system as claimed in claim 1, further comprising a user interaction database comprising user interaction data relating to previous system interactions including data relating to past user queries entered into the system and associated user contexts for particular users, and user'"'"'s responses relating to those interactions.

3. The customer self service system as claimed in claim 2, wherein said search mechanism comprises:

a context attribute database comprising types of user contexts and one or more context attributes associated with each user context for processing by said system and, comprising functions for computing values for each context attribute; and

,

a user context classifier device for receiving a user query and a context vector comprising data associating an interaction state with said user, and processing said query and context vector against data included in said context attribute database for generating context parameters that predict a particular user context, wherein said classifier device populates said user context vector with context parameters specifying a user interaction state for use in a subsequent resource search.

4. The customer self service system as claimed in claim 3, wherein said user context classifier device includes processing mechanism for applying said functions to context for specifying said user interaction state, said mechanism further annotating the context vector with a set of context parameters for use in subsequent processing.

5. The customer self service system as claimed in claim 4, wherein said processing mechanism implements an inductive learning algorithm for predicting said user contexts.

6. The customer self service system as claimed in claim 4, further including updating mechanism for providing additions and modifications to a set of context attribute functions resulting in increasing ability to predict derived contexts as functions of the raw contexts, whereby the attribute functions database is enhanced.

7. The customer self service system as claimed in claim 6, wherein said updating mechanism for updating the attribute value functions database comprising

mechanism for analyzing historical user interaction data from the user interaction database and learning how context attribute values map to context attribute functions, wherein said data from the user records database serves as a training set for continuous improvement of said functions in said database.

8. The customer self service system as claimed in claim 7, wherein said previous system interaction data further includes prior transactions of a current user and prior transactions of other similar users, wherein common behaviors and acceptance criteria are determined for updating said functions.

9. The customer self service system as claimed in claim 3, wherein said search mechanism further comprises:

mechanism for receiving a current user query for requesting resources and said user context vector associated with said current user query;

mechanism for applying resource indexing functions to map each user query and associated context vector to a sub-set of resources from a resource library, and generating a response set including said sub-set of resources that are most relevant to said user'"'"'s query, said indexing functions including resource parameters for facilitating narrower searches.

10. The customer self service system as claimed in claim 9, further including:

adaptive indexing process for enhancing said resource indexing functions by increasing their relevance and specificity for mapping user queries to resources, said adaptive indexing function increasing the value of search results for a current user in their context.

11. The customer self service system as claimed in claim 10, wherein said database of user interaction records further includes actual resources selected by the users, said adaptive indexing process implementing a supervised learning algorithm for receiving user interaction data from among said database of user interaction records and resources from said resource library and, adapting resource indexing functions based on a history of user interactions with said system as provided in said database of user interaction records.

12. The customer self service system as claimed in claim 11, wherein said user interaction data comprises user interaction feedback including history of prior interaction with the resource search and selection system, said supervised learning algorithm optimizing a performance of said resource indexing functions as measured by an evaluation metric applied to the user interaction feedback.

13. The customer self service system as claimed in claim 9, wherein said search mechanism further comprises:

mechanism for receiving said resource response set of results obtained in response to a current user query, and receiving said user context vector associated with said current user query, and,

an ordering and annotation function for mapping the user context vector with the resource response set to generate an annotated response set having one or more annotations for controlling the presentation of the resources to the user, wherein the ordering and annotation function is executed interactively at the time of each user query.

14. The customer self service system as claimed in claim 13, wherein said annotations include elements for ordering resources results for presentation to said user via a graphic user interface.

15. The customer self service system as claimed in claim 13, wherein said user interaction records includes actual resources selected by the users and the annotation schemes used for presenting them, said ordering and annotation function implementing a supervised learning algorithm for receiving user interaction data and an annotation scoring metric representing a measure of performance in locating resource response results presented to said user, and, generating said ordering and annotation function, said annotation function being adaptable based on history of user interactions.

16. The customer self service system as claimed in claim 15, wherein said user interaction data comprises user interaction feedback, said supervised learning algorithm optimizing said annotation scoring metric as measured by said user interaction feedback.

17. The customer self service system as claimed in claim 13, further comprising context clustering mechanism for receiving said user interaction data and a distance metric for associating closeness of said user interaction data and, clustering said user interaction data according to said distance metric to determine new user contexts and associated attributes for use in subsequent resource searches initiated by users in said system, wherein improved query definition and resource lookup results from said new determined user context attributes.

18. The customer self service system as claimed in claim 17, wherein a system administrator updates said context attribute database with determined new user contexts and associated context attributes and, further, develops new context attribute functions for computing values for new user context attributes, and assigns new records in said user interaction records database with values for those attributes, said updating of context assignments serve as the training data for continuously improving said functions in said context attributes database.

19. The customer self service system as claimed in claim 17, wherein said distance metric includes determining closeness of parameters of said user interaction data, a closeness parameter including similarity of result sets of a user query.

20. The customer self service system as claimed in claim 17, wherein said system administrator develops new definitions and logic for mapping specific resources to specific context sets.

21. The customer self service system as claimed in claim 17, wherein said context clustering mechanism implements an unsupervised clustering algorithm for clustering said user interaction data records.

22. The customer self service system as claimed in claim 13, wherein said resource response set is presented to said user via a graphical user interface (GUI), said GUI comprising:

a first graphic element for displaying said response set according to a defined ranking, said one or more ranked resources from said first graphic element being user selectable; and

,

a second graphic element for displaying a multi-dimensional plot comprising two or more axes with each axis corresponding to a user specified results selection criterion and each axis including points representing each of said resources selected from said first graphic element along each dimension.

23. The customer self service system as claimed in claim 22, further including mechanism for enabling user selection of a single point of a desired resource from said multi-dimensional plot, and enabling visualization of the same resource represented as a data point on each of said axes of said multi-dimensional plot in response to said single resource selection.

24. The customer self service system as claimed in claim 23, wherein said visualization of the same resource upon each of said axes includes graphically connecting a point corresponding to the selected resource to all the other points for that resource in said plot.

25. The customer self service system as claimed in claim 23, wherein each axis enables visualization of a ranking of said resources according to each selection criterion at each dimension.

26. The customer self service system as claimed in claim 23, wherein each axis of said multi-dimensional plot is displayed according to a user-defined sequence.

27. The customer self service system as claimed in claim 23, wherein said second graphic interface comprises a third graphic element for displaying a detailed description of each of said selected resources of said response set.

28. The customer self service system as claimed in claim 22, wherein said second graphic interface includes a display indicating a weighting of each user selected criterion at each dimension.

29. A method for performing resource search and selection in a customer self service system, said method comprising:

a) providing an interface enabling entry of a query for a resource and specification of one or more user context elements, each element representing a context associated with the current user state and having context attributes and attribute values associated therewith;

b) enabling user specification of relevant resource selection criteria for enabling expression of relevance of resource results in terms of user context;

c) searching a resource database and generating a resource response set having resources that best match a user'"'"'s query, user context attributes and user defined relevant resource selection criteria;

d) presenting said resource response set to said user in a manner whereby a relevance of each said resources being expressed in terms of user context in a manner optimized to facilitate resource selection; and

,

e) enabling continued user selection and modification of context attribute values to enable increased specificity and accuracy of a user'"'"'s query to thereby result in improved selection logic and attainment of resource response sets best fitted to said query.

30. The method as claimed in claim 29, wherein said customer self service system comprises a user interaction database comprising user interaction data relating to previous system interactions including data relating to past user queries entered into the system and associated user contexts for particular users, and user'"'"'s responses relating to those interactions, said searching step c) comprises the steps of:

receiving a user query and a context vector comprising data associating an interaction state with said user; and

,

processing said query and context vector against data included in a context attribute database for generating context parameters that predict a particular user context, wherein said classifier device populates said user context vector with context parameters specifying a user interaction state for use in a subsequent resource search.

31. The method as claimed in claim 30, wherein said query processing step comprises the steps of:

applying said functions to context for specifying said user interaction state; and

annotating the context vector with a set of context parameters for use in subsequent processing.

32. The method as claimed in claim 31, further including the step of implementing an inductive learning algorithm for predicting said user contexts.

33. The method as claimed in claim 32, further including the step of providing additions and modifications to a set of context attribute functions resulting in increasing ability to predict derived contexts as functions of the raw contexts.

34. The method as claimed in claim 33, further including the step of updating an attribute value functions database by analyzing historical user interaction data from a user interaction database and learning how context attribute values map to context attribute functions, wherein said data from the user records database serves as a training set for continuous improvement of said functions in said database.

35. The method as claimed in claim 34, wherein said previous system interaction data further includes prior transactions of a current user and prior transactions of other similar users, said method including determining common behaviors and acceptance criteria for improving said functions.

36. The method as claimed in claim 30, wherein said searching step c) further comprises:

receiving a current user query for requesting resources and said user context vector associated with said current user query; and

applying resource indexing functions to map each user query and associated context vector to a sub-set of resources from a resource library; and

,

generating a response set including said sub-set of resources that are most relevant to said user'"'"'s query, said indexing functions including resource parameters for facilitating narrower searches.

37. The method as claimed in claim 36, further including the step of enhancing said resource indexing functions by increasing their relevance and specificity for mapping user queries to resources.

38. The method as claimed in claim 37, wherein said database of user interaction records further includes actual resources selected by the users, said enhancing step including the step of implementing a supervised learning algorithm for receiving user interaction data from among said database of user interaction records and resources from said resource library and, adapting resource indexing functions based on a history of user interactions with said system as provided in said database of user interaction records.

39. The method as claimed in claim 38, wherein said user interaction data comprises user interaction feedback including history of prior interaction with the resource search and selection system, said supervised learning algorithm optimizing a performance of said resource indexing functions as measured by an evaluation metric applied to the user interaction feedback.

40. The method as claimed in claim 36, wherein said searching step c) further comprises the steps of:

receiving said resource response set of results obtained in response to a current user query, and receiving said user context vector associated with said current user query; and

,

at the time of each user query, mapping the user context vector with the resource response set to generate an annotated response set having one or more annotations for controlling the presentation of the resources to the user.

41. The method as claimed in claim 40, wherein said annotations include elements for ordering resources results for presentation to said user via a graphic user interface.

42. The method as claimed in claim 41, wherein said presenting step d) further includes the step of presenting said resource response set to said user via a graphical user interface (GUI), said GUI comprising:

a first graphic element for displaying said response set according to a defined ranking, said one or more ranked resources from said first graphic element being user selectable; and

,

a second graphic element for displaying a multi-dimensional plot comprising two or more axes with each axis corresponding to a user specified results selection criterion and each axis including points representing each of said resources selected from said first graphic element along each dimension.

43. The method as claimed in claim 42, further including the steps of:

enabling user selection of a single point of a desired resource from said multi-dimensional plot; and

,

enabling visualization of the same resource represented as a data point on each of said axes of said multi-dimensional plot in response to said single resource selection.

44. The method as claimed in claim 43, further including the step of graphically connecting a point corresponding to the selected resource to all the other points for that resource in said plot.

45. The method as claimed in claim 40, wherein said user interaction records includes actual resources selected by the users and the annotation schemes used for presenting them, said method further including the steps of:

implementing a supervised learning algorithm for receiving user interaction data and an annotation scoring metric representing a measure of performance in locating resource response results presented to said user; and

,

generating an ordering and annotation function for performing said mapping, and adapting said annotation function based on history of user interactions.

46. The method as claimed in claim 45, wherein said user interaction data comprises user interaction feedback, said supervised learning algorithm optimizing said annotation scoring metric as measured by said user interaction feedback.

47. The method as claimed in claim 40, further comprising the steps of:

receiving said user interaction data and a distance metric for associating closeness of said user interaction data; and

,

clustering said user interaction data according to said distance metric to determine new user contexts and associated attributes for use in subsequent resource searches initiated by users in said system, wherein improved query definition and resource lookup results from said new determined user context attributes.

48. The method as claimed in claim 47, further including implementing services for:

updating said context attribute database with determined new user contexts and associated context attributes and;

developing new context attribute functions for computing values for new user context attributes; and

,

assigning new records in said user interaction records database with values for those attributes, said updating of context assignments serve as the training data for continuously improving said functions in said context attributes database.

49. The method as claimed in claim 48, wherein said distance metric includes determining closeness of parameters of said user interaction data, a closeness parameter including similarity of result sets of a user query.

50. The method as claimed in claim 48, further including the step of developing new definitions and logic for mapping specific resources to specific context sets.

51. The method as claimed in claim 50, further including the step of implementing an unsupervised clustering algorithm for clustering said user interaction data records.