International Business Machines Corporation

Yahoo! Inc.

Oath Inc.

Byron Edward Dom

Soumen Chakrabarti

Prabhakar Raghavan

David Andrew Gibson

Sridhar Rajagopalan

Andrew Stephen Tomkins

Shanmugasundaram Ravikumar

Alexandrin Popescul

Dmitry Pavlov

Deepa Paranjpe

John Samuel Batchelder

Martin Henk Van Den Berg

Robert S Jaffe

Ramnath Balasubramanyan

Raymond E Bonner

Rakesh Agrawal

Shivakumar Vaithyanathan

Jianchang Mao

Dragutin Petkovic

Jyh-Herng Chow

Kevin Snow McCurley

Bipin Suresh

Sunita Sarawagi

Eren Manavoglu

Tong Zhang

Jon Michael Kleinberg

Clifford A Brunk

Cliff Brunk

Hui Han

Abhishek Goyal

Piotr Indyk

Dao-I Lin

Techniques for statistically estimating a rating or other 'figure of merit' for a user are disclosed. According to one such technique, a first quantity of submissions that were submitted by a user is determined. A second quantity of submissions that (a) were submitted by the user and (b) obtained a particular rating from a rating mechanism also is determined. A user rating for the user is determined based at least in part on the first quantity, the second quantity, and a factor that is independent of both the first quantity and the second quantity&#x2014;such as the probability that an answer submitted by any answerer in a population will obtain the particular rating from the rating mechanism. The influence that the second quantity has on the user rating relative to the influence that the factor has on the user rating may depend at least in part on the first quantity.

Statistical credibility metric for online question answerers

Techniques are described herein for generating and displaying a confusion matrix wherein a data item belonging to one or more actual classes is predicted into a class. The classes in which the data item may be predicted (the 'predicted classes') are ranked according to a score that in one embodiment indicates the confidence of the prediction. If the data item is predicted into a class that is one of the top K ranked predicted classes, then the prediction is considered accurate and an entry is created in a cell of a confusion matrix indicating the accurate prediction. If the data item is not predicted into a class that is not one of the top K ranked predicted classes, then the prediction is considered inaccurate and an entry is created in a cell of a confusion matrix indicating the inaccurate prediction.

Confusion matrix for classification systems

A system is described for targeting data to a site referenced on a page based on a condition. The system may include a processor, a memory, and an interface. The memory may be operatively connected to the processor and the interface and may store a data, a site, a condition, and a page containing content. The interface may be operatively connected to the memory and the processor and may communicate the page to a user. The processor may identify the data, site, condition, and page containing content. The processor may add the data to the page if the content of the page satisfies the condition.

System for targeting data to sites referenced on a page

A system and method are disclosed for identifying a keyword that is a novel concept or anomaly based on prior search results for the keyword. Advertisements may be sold for the keyword, or the keyword may be purchased or recommended for purchase based on anticipation of increased future searches on the keyword.

Keyword bidding strategy for novel concepts

Techniques are described for assigning, to target categories of a target scheme, items that have been obtained from a plurality of sources. In situations in which one or more of the sources has organized its information according to a source scheme that differs from the target scheme, the assignment may be based, in part, on an estimate of the probability that items from a particular source category should be assigned to a particular target category. Such probability estimates may be based on how many training set items associated with the particular source category have been assigned to the particular target category. Source categories may be grouped into clusters. The probability estimates may also be based on how many training set items within the cluster to which the particular source category has been mapped, have been assigned the particular target category.

Assigning into one set of categories information that has been assigned to other sets of categories

Techniques are provided for automatic product categorization. In one aspect, the categorization is based on text and one or more other values associated with a product offering. In another aspect, a first categorization of a product offering is performed and, if the product category chosen is in a set of co-refinable product categories, then a second (or third, fourth and so on) categorization is performed among the set of co-refinable product categories. In a third aspect, products are categorized based on cost. In another aspect, after products are categorized, the products are flagged for further categorization processing if the cost for categorizing the product is beyond a predefined threshold.

Automatic product categorization

An improved system and method is provided for determining web page quality using collective inference based on local and global web page information. A classification engine may be provided for classifying a web page using local features of a seed set of web pages and global web graph information about the seed set of web pages. A dual algorithm based on graph regularization formulated as a well-formed optimization solution may be used in an embodiment for applying collective inference for binary classification of the web page using the local web page information and global web graph information of a web page, the local web page information and global web graph information of an authoritative set of web pages, and the local web page information and global web graph information of a non-authoritative set of web pages.

System and method for determining web page quality using collective inference based on local and global information

A Web server stores a table of Web page inlinks. When a Web page is accessed and a user wants to access other pages related to the accessed page, the user requests the table of inlinks, and from it generates a list of sibling links to the accessed page, the sibling links being outlinks of one or more of the inlinks in the table.

Method and system for distributed autonomous maintenance of bidirectional hyperlink metadata on the web and similar hypermedia repository

A system and method are provided for eliciting interesting structure from a collection of entities or resources with explicit and/or implicit, static and/or dynamic relations, called 'affinities,' between them. Interesting structure includes (1) notions of quality, authority, or definitiveness of information, (2) notions of relevance to a user's information need, (3) notions of similarity among the plurality of resources retrieved from a universe of resources by a query process, and (4) notions of similarity among the usages of resources by different users/servers. Similarities between entities are computed, based on similarities between the affinity values for the entities. That is, where the affinitiy values for two entities resemble each other, the two entities have a high degree of similarity. Using the similarities, the entities are ranked, clustered, etc., based on a significance derived from the similarities. The ranking, clustering, etc., makes up the interesting structure which is sought.

Method and system for filtering of information entities

A system and method for training an SVM in a scalable manner includes boosting the SVM during training. Specifically, individual SVMs in an ensemble of SVMs are trained using small subsets of a training data set, with data that earlier classifiers in the ensemble incorrectly classified being overrepresented in succeeding subsets. In this way, the speed with which the overall SVM is trained is increased and the memory requirements therefor are reduced, even for relatively large training data sets.

System and method for boosting support vector machines

A objective function based on a Bayesian statistical estimation framework is used to determine an optimal model selection by choosing both the optimal number of clusters and the optimal feature set. Heuristics can be applied to find the optimal (or at least sub-optimal) of this objective function in terms of the feature sets and the number of clusters, wherein the maximization of the objective function corresponds to the optimal model structure.

Model selection in machine learning with applications to document clustering

A focussed Web crawler learns to recognize Web pages that are relevant to the interest of one or more users, from a set of examples provided by the users. It then explores the Web starting from the example set, using the statistics collected from the examples and other analysis on the link graph of the growing crawl database, to guide itself towards relevant, valuable resources and away from irrelevant and/or low quality material on the Web. Thereby, the Web crawler builds a comprehensive topic-specific library for the benefit of specific users.

System and method for focussed web crawling

A method, apparatus, and article of manufacture for a computer implemented hypertext classifier. A new document containing citations to and from other documents is classified. Initially, documents within a neighborhood of the new document are identified. For each document and each class, an initial probability is determined that indicates the probability that the document fits a particular class. Next, iterative relaxation is performed to identify a class for each document using the initial probabilities. A class is selected into which the new document is to be classified based on the initial probabilities and identified classes.

Growing community of inventors

Los Gatos, CA, United States of America

Byron Edward Dom