Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1997-09-25
2001-02-27
Von Buhr, Maria N. (Department: 2771)
Data processing: database and file management or data structures
Database design
Data structure types
Reexamination Certificate
active
06195657
ABSTRACT:
FIELD OF THE INVENTION
The present invention is related to collaborative filtering, information filtering, and knowledge management, more specifically to automatically recommending to a user objects and other users of a computer system based on categories and objects identified by each user.
BACKGROUND OF THE INVENTION
In today's business environment it is becoming increasingly difficult for each of us to stay informed. Each separate task might involve both collecting a number of documents and seeking information from relevant colleagues. The number of potentially relevant documents is vast, encompassing both those internal to an organization, and those available over public computer networks such as the World-Wide Web. In addition, in many organizations, the number of potentially relevant colleagues within the organization can be so large that an employee is unable to locate the best sources of knowledge without assistance.
Filtering systems exist which attempt to keep users informed by delivering relevant documents (e.g., Tak W. Yan and Hector Garcia-Molina, “SIFT—A Tool for Wide-Area Information Dissemination”
Proc. of the
1995
USENIX Tech. Conf.,
pp 177-86, 1995). Unfortunately, these systems are based on preferences which need to be explicitly submitted by users. This is an onerous task. It is not always easy for users to clearly define their preferences nor to formulate them in a way that allows the computer system to make sense of them. Commonly used techniques require the user to specify a list of keywords denoting their interests, or to select from among a predetermined set of categories. Both of these requirements impose an additional workload upon the user.
Alternative systems exist which perform “collaborative filtering,” for instance systems described in U.S. Pat. Nos. 4,996,642 (issued Feb. 26, 1991) and 5,583,763 (issued Dec. 10, 1996). In these systems, the user is required to denote a single set of favorite objects exemplifying their interests, or to supply preference scores for a number of objects. Once again, these are onerous tasks outside of the normal workload of a user. Furthermore, in many settings a user will have several different contexts for which they might require entirely different sets of recommendations. For instance, a user might be working on a number of projects simultaneously. The collaborative filtering systems referenced represent the user as having a monolithic set of interests and do not make different recommendations for different contexts.
There are also many varieties of “push” systems which do not attempt to target individual users or the specific interests of each user, but broadcast the same information objects to large segments of the user population. With this non-personalized approach, these systems end up adding to the problem of information overload rather than alleviating it.
Database systems exist which hold records of employee experiences, interests, skills, etc. These systems can be used to locate colleagues relevant to a particular task or project. Unfortunately, maintenance of such a database is expensive and difficult, and its use is not integrated into the employee's regular flow of work. In addition, these systems do not provide a single source for both relevant documents and relevant colleagues.
SUMMARY OF THE INVENTION
The present invention provides an efficient means for presenting a user with recommendations relevant to their current tasks and activities. These recommendations take the form of information objects, other users of the recommendation system who are pursuing or have completed similar tasks or activities, or categories of information objects other users of the system have gathered in the past. The information objects recommended can be of many different types; in the example embodiment the invention given in the Detailed Description section below is adapted to a recommendation system for documents accessible via a data communications network such as the World-Wide Web or a company intranet. In general any uniquely identifiable object is recommendable.
The invention operates on the assumption that users group related objects together in categories or folders, in the normal course of their work while pursuing information seeking or tracking activities. The resulting categories correspond to groupings which are meaningful, intuitive, and useful to the users who created them.
In an embodiment, in order to provide recommendations to a “target” user for a particular “target” category they have created, the recommendation system of the present invention discovers categories created by users which are similar to the target category. Similarity between a discovered category and the target category is determined in part by the number of information objects which are in both the discovered and target categories. “Matching” categories have at least one information object which is also in the target category. From the set of discovered matching categories, recommendations can be made of information objects which are not already in the target category, as well as of the users who originally submitted the matching categories, and of the matching categories themselves. These recommendations can be delivered to the target user in the context of their target category. Thus if the user has a number of target categories, the grouping of the resulting recommendations will match the user's own intuitive grouping as exemplified by their target categories, rather than some predetermined categorization created by an administrator or editor.
Further configurations of the present invention allow a user to initiate communications with other users who have been recommended as relevant to a target category. Such communications are archived by the system and can be recommended in the same way as other information objects, since human expertise can often be more valuable than that codified in documents or information objects. In addition, users can submit relevant recommendations back to the recommendation system, as exemplars of relevant topics, so that further recommendations on more specific or different themes can be delivered.
REFERENCES:
patent: 4870579 (1989-09-01), Hey
patent: 4970681 (1990-11-01), Bennett
patent: 4996642 (1991-02-01), Hey
patent: 5583763 (1996-12-01), Atcheson et al.
patent: 5666442 (1997-09-01), Wheeler
patent: 5717913 (1998-02-01), Driscoll
patent: 5724567 (1998-03-01), Rose et al.
patent: 5749081 (1998-05-01), Whiteis
patent: 5754938 (1998-05-01), Herz et al.
patent: 5754939 (1998-05-01), Herz et al.
patent: 5790426 (1998-08-01), Robinson
patent: 5835087 (1998-11-01), Herz et al.
patent: 5857179 (1999-01-01), Vaithyanathan et al.
patent: 5867799 (1999-02-01), Lang et al.
patent: 6041311 (2000-03-01), Chislenko et al.
patent: 6049777 (2000-04-01), Sheena et al.
patent: 6092049 (2000-07-01), Chislenko et al.
patent: 6112186 (2000-08-01), Bergh et al.
Douglas W. Oard, et al., A Conceptual Framework for Text Filtering, Technical Report CLIS-TR-96-02, University of Maryland, College Park, College of Library and Information Services, 1996.
Loren Terveen, et al, PHOAKS: A System For Sharing Recommendations, Communications of the ACM, 40(3):56-62, Mar. 1997.
B. Miller, et al., Experience with GroupLens: Making Usenet Useful Again, pp. 1-17, Proceedings of the 1997 Usenix Winter Technical Conference, Jan. 1997.
Henry Kautz, et al., ReferralWeb: Combining Social Networks and Collaborative Filtering, Communications of the ACM, 40(3):63-65, Mar. 1997.
Joseph A. Konstan, et al., GroupLens: Applying Collaborative Filtering to Usenet News, Communications of the ACM, 40(3):77-87, Mar. 1997.
Paul Resnick, et al., GroupLens: An Open Architecture for Collaborative Filtering of Netnews, Proceedings of the ACM Conference on Computer Supported Cooperative Work, pp. 56-58,175-186, Chapel Hill, NC, 1994.
David Goldberg, et al., Using Collaborative Filtering to Weave an Information Tapestry, Communications of the ACM, 35(12):61-70, Dec. 1992.
Donald Fisk, An Application of Social Filtering to Movie Recommen
Polanco Marcos J.
Rucker James L.
Imana, Inc.
Jones Allston L.
Von Buhr Maria N.
LandOfFree
Software, method and apparatus for efficient categorization... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Software, method and apparatus for efficient categorization..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Software, method and apparatus for efficient categorization... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2588938