Data processing: financial – business practice – management – or co – Automated electrical financial or business practice or... – Electronic shopping
Reexamination Certificate
1999-01-22
2002-11-26
Coggins, Wynn (Department: 3625)
Data processing: financial, business practice, management, or co
Automated electrical financial or business practice or...
Electronic shopping
C705S001100, C705S014270, C705S027200
Reexamination Certificate
active
06487541
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to making a prediction about a user based on data provided in association with other users. In particular, a rating associated with a user is forecasted based on a predictability graph. Furthermore, the present invention relates to an automated system and method for providing recommendations of items for sale in an e-commerce environment.
BACKGROUND OF THE INVENTION
A computing environment offers an opportunity to collect information relating to the various preferences of users. User preferences may take the form of a rating which is associated with an item. A rating may be a quantitative measure of a users relative like or dislike of an item. For example, a user purchasing compact discs, books, or videos over the Internet may provide an integer, in a given range, indicating his/her satisfaction with a product available for sale.
Rating information provided by a group of users may be useful for recommending items to a user. In other words, the preference of a user may be predicted based on ratings for items provided by a group of users. One user may be found “similar” in his preferences to other users. Users with “similar” ratings may be used to predict each other's preferences. Deducing the preferences of a given user by examining information about the preferences of other “similar” users is often referred to as collaborative filtering. Users collaborate by each rating a subset of a set of items. The collective knowledge attained from the collaborative rating of items may then be used to make recommendations. For example, if, based on the collective knowledge, it is predicted that an item would be highly rated by a user, the item may be recommended to the user.
Collaborative filtering may be used to make recommendations to customers purchasing items for sale on the World Wide Web. Alternatively, collaborative filtering may be used to make marketing recommendation to vendors using the World Wide Web. In general, due to the nature of a distributed computer environment, collaborative filtering may be used to make recommendations and/or customize preferences of users involved in electronic transactions or e-commerce. Suppose, for example, that the amount of time a user spends viewing a Web page is regarded as an implicit rating of the Web page by the user. Users with “similar” viewing habits may be detected. Based on the observed viewing habits of users, recommendations may be calculated which estimate the interest a user may take in a particular Web page. In this way, a Web page estimated to be of interest to a particular user may be recommended to the user. For instance, if a user spends much time browsing Web pages detailing information relating to running, a Web page giving a comparative study of running shoes may be recommended to the user. The Web page giving the comparative study of running shoes may be recommended based on the amount of time other users, viewing running related Web pages, spend viewing the study.
Several collaborative filtering engines are currently available. A discussion of such collaborative filtering engines may be found, for example, in U.S. Pat. No. 4,870,579 and U.S. Pat. No. 4,996,642 both issued to John B. Hey (employed in collaborative filtering technology offered by LikeMinds Inc.), and Upendra Shardanand and Pattie Maes (a founder and former director of Firefly Network, Inc.), “Social Information Filtering: Algorithms for Automating Word of Mouth,” Proceedings of CHI '95, Denver, Colo., 1995,pages 210-217.
The technology employed by LikeMinds Inc., as disclosed in U.S. Pat. Nos. 4,870,579 and 4,996,642 involves random sampling of users. A measure of “agreement strength” between a current user and the randomly sampled users is computed. A subset is then chosen of the randomly sampled users. Each member of the subset corresponds to either a relatively high “measurement strength” or a relatively high item coverage. Item predictions are computed based on a “closeness” function of pairs of ratings by member of the subset.
The technology employed by Firefly Network, Inc. involves computing a Pearson r coefficient to measure the “similarity” between two users. Users who are “similar” to a given user are identified. Predictions are made by forming a weighted average of ratings provided by the “similar” group of users. The weighting factor used to form the average of ratings of items is made proportional to the Pearson coefficient.
SUMMARY OF THE INVENTION
A rating of a plurality of ratings is predicted. The rating is associated with a user of a plurality of users and the rating corresponds to an item of a plurality of items. One of the plurality of ratings, corresponding to at least one of the plurality of items, is provided for each of the plurality of users. A predictability relation between ones of the plurality of users and other ones of the plurality of users is calculated based on ratings provided by users. One of a plurality of nodes is assigned to each of the plurality of users. Ones of the plurality of nodes are connected with other ones of the plurality of nodes by a plurality of edges based on the predictability relation. A graph which includes the plurality of nodes and the plurality of edges is searched for a path from a node assigned to the user of the plurality of users to another node assigned to another user of the plurality of users. The rating of the plurality of ratings associated with the user of a plurality of users is calculated based on the path and the predictability relation.
REFERENCES:
patent: 4870579 (1989-09-01), Hey
patent: 4996642 (1991-02-01), Hey
patent: 5704017 (1997-12-01), Heckerman et al.
patent: 6189005 (2001-02-01), Chakrabarti et al.
patent: 2001/0013009 (2001-08-01), Greening et al.
patent: 2001/0014868 (2001-08-01), Herz et al.
patent: WO 98/33135 (1998-07-01), None
Ubois, It's a Jungle Out There, Upside v10n3 PP: 106-110 (pp. 1-2 of Abstract), Mar. 1998.*
Brandt, Internet Kamikazee: Yahoo's Tim Koogle, Upside v10nl PP: 100, 102+, Jan. 1998.*
Rapoza, GourpLens Sees The Future, PC Week, v15 n26 P46(1), Jun. 29, 1998.*
Rapoza, Alexa's Theory of Relativity, PC Week, v14 n35 p2(1) Aug. 18, 1997.*
Upendra, Shardanand et al., Social Information Filtering: Algorithms for Automating “Word of Mouth”, Proceedings of C '95, Denver, Colorado, pp. 210-217, 1995.*
Kautz et al., The Hidden Web, American Association for Artificial Intelligence, 1997 extracted from web site: http://jamesthornton.com on Apr. 26, 2002.*
Upendra Shardanand et al., “Social Information Filtering: Algorithms for Automating ‘Word of Mouth’”, Proceedings of CHI '95, Denver, Colorado, pp. 210-217, 1995.
Aggarwal Charu Chandra
Wolf Joel Leonard
Yu Philip Shi-Lung
Coggins Wynn
F. Chau & Associates LLP
Garg Yogesh C.
International Business Machines - Corporation
LandOfFree
System and method for collaborative filtering with... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for collaborative filtering with..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for collaborative filtering with... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2956837