Yahoo! Inc.

Overture Services, Inc.

Sathiya Keerthi Selvaraj

Philip Lewis Bohannon

Srujana Merugu

Sundararajan Sellamanickam

Arun Sundaresan Iyer

Olivier Chapelle

Dennis M DeCoste

Vinay Kakade

Arup Kumar Choudhury

Vipul Agarwal

John Anthony Tomlin

Cong Yu

Raghu Ramakrishnan

David Mark Ciemiewicz

Mridul Muralidharan

Nilesh Dalvi

Ashwin Kumar V Machanavajjhala

Ashwin Machanavajjhala

Alok S Kirpal

Priyanka Garg

Vikas Sindhwani

Arun Shankar Iyer

Kedar Bellare

Mani Abrol

A system and method is described for large scale entity-specific classification of each entity-specific set of candidates in a collection of candidates for each specific entity in a collection of entities. The collection of entities may comprise a specific category or domain of entities (e.g. schools, restaurants, manufacturers, products, events, people). Candidates may comprise webpages or other resources with resource identifiers. Entity specific sets of candidates may be found by leveraging search engine query results and user interaction therewith for queries based on entity-specific attributes. The relationship(s) or class(es) for which candidate resources are being classified relative to a specific entity may comprise an authoritative, official home page (OHP), or other class (e.g. fan page, review, aggregator) relative to a specific entity. A feature generator generates entity-specific features for candidates. In accordance with its features, one or more classifiers rank each candidate for a specific class for a specific entity.

Large scale entity-specific resource classification

A system is disclosed for obtaining and aggregating opinions generated by multiple sources with respect to one or more objects. The disclosed system uses observed variables associated with an opinion and a probabilistic model to estimate latent properties of that opinion. With those latent properties, the disclosed system may enable publishers to reliably and comprehensively present object information to interested users.

Opinion aggregation system

Methods, systems and computer readable mediums are provided for indexing network resources. One method includes accessing, using one or more computer systems, a data store of menu items. The method further includes accessing identification information associated with one or more food providers from one or more data sources. One or more network resources are crawled based on the identification information to search for one or more menu items in the data store of menu items associated with corresponding ones of the food providers. Using the one or more computing systems, an index feed is generated, the index feed comprising the identification information of one or more of the food providers, and one or more menu items associated with the identification information of corresponding food providers based on the crawl and search.

System and method for indexing food providers and use of the index in search engines

A classifier development process seamlessly and intelligently integrates different forms of human feedback on instances and features into the data preparation, learning and evaluation stages. A query utility based active learning approach is applicable to different types of editorial feedback. A bi-clustering based technique may be used to further speed up the active learning process.

Rapid iterative development of classifiers

Techniques are provided for the efficient location, processing, and retrieval of local product information derived from web pages generally locatable through form queries submitted to web pages often referred to as the 'deep' or 'hidden' web. In an embodiment, information such as product information and dealer-location information is located on a web page form such as a dealer-locator form. After location of a suitable web page form, editorial wrapping is performed to create an automated information extraction process. Using the automated information extractor, deep-web crawling is performed. A grid-based extraction of individual business records is performed, and matching and ingestion are performed in conjunction with a business listing database. Finally, metadata tags are added to entries in the business listing database. Metadata tags also may be added to entries in other databases.

Method and system for form-filling crawl and associating rich keywords

An improved system and method for generating a maximum utility slate of advertisements for online advertisement auctions is provided. Various utility factors for each advertisement that may be a candidate in a slate of advertisements may be applied within a framework in order to generate a maximum utility slate of advertisements. Either backward or forward dynamic programming may be applied to recursively evaluate the utility of subslates of advertisements in order to generate a maximum utility slate of advertisements. In an embodiment, a network with directed edges and associated costs may be defined, and the longest path may be found in the directed network for constructing a maximum utility slate of advertisements. Various utility factors may be applied for different objectives of an auctioneer and the framework presented may be extended for revenue ordering, exclusion of bidders, ordering slates according to first and second price utilities, and so forth.

System and method for generating a maximum utility slate of advertisements for online advertisement auctions

Methods and apparatus for performing computer-implemented extraction of temporal information for business entities and events are disclosed. In one embodiment, a sequence of text is obtained. A label is assigned to one or more of a plurality of segments of the text such that each of the one or more of the plurality of segments of the text is classified as temporal data in one of a plurality of classes of temporal data. One or more rules are applied to the one or more segments of the text that have been classified as temporal data to generate a structured representation of the temporal data, where the rules include one or more schematic rules. Each of the schematic rules pertains to one or more of the plurality of classes of temporal data and indicates a structure in which temporal data in the corresponding one or more of the plurality of classes is to be stored.

Extracting rich temporal context for business entities and events

In one embodiment, training a ranking model comprises: accessing the ranking model and an objective function of the ranking model; accessing one or more preference pairs of objects, wherein for each of the preference pairs of objects comprising a first object and a second object, there is a preference between the first object and the second object with respect to the particular reference, and the first object and the second object each has a feature vector comprising one or more feature values; and training the ranking model by minimizing the objective function using the preference pairs of objects, wherein for each of the preference pairs of objects, a difference between the first feature vector of the first object and the second feature vector of the second object is not calculated.

Efficient algorithm for pairwise preference learning

The present invention provides methods and systems for binary classification of items. Methods and systems are provided for constructing a machine learning-based and pairwise ranking method-based classification model for binary classification of items as positive or negative with regard to a single class, based on training using a training set of examples including positive examples and unlabelled examples. The model includes only one hyperparameter and only one threshold parameter, which are selected to optimize the model with regard to constraining positive items to be classified as positive while minimizing a number of unlabelled items classified as positive.

Pairwise ranking-based classifier

Generally, the present invention provides a method and computerized system for generating a classifier model, wherein the classifier model is operative to classify web content. The method and computerized system includes a first step of defining a plurality of predictive performance measures based on a leave one out (LOO) cross validation in terms of selectable model parameters. Exemplary predictive performance measures includes smoothened predictive measures such as F-measure, weighted error rate measure, area under curve measure, by way of example. The method and computerized system further includes deriving efficient analytical expressions for predictive performance measures to compute the LOO predictive performance and their derivatives. The next step is thereupon selecting a classifier model based on the LOO predictive performance.

System and method for generating a classifier model for classifying web content

An improved system and method is provided for training a multi-class support vector machine to select a common subset of features for classifying objects. A multi-class support vector machine generator may be provided for learning classification functions to classify sets of objects into classes and may include a sparse support vector machine modeling engine for training a multi-class support vector machine using scaling factors by simultaneously selecting a common subset of features iteratively for all classes from sets of features representing each of the classes. An objective function using scaling factors to ensure sparsity of features may be iteratively minimized, and features may be retained and added until a small set of features stabilizes. Alternatively, a common subset of features may be found by iteratively removing at least one feature simultaneously for all classes from an active set of features initialized to represent the entire set of training features.

System and method for training a multi-class support vector machine to select a common subset of features for classifying objects

Support vector machines (SVMs), though accurate, are not preferred in applications requiring great classification speed, due to the number of support vectors being large. To overcome this problem a primal system and method with the following properties has been devised: (1) it decouples the idea of basis functions from the concept of support vectors; (2) it greedily finds a set of kernel basis functions of a specified maximum size (d) to approximate the SVM primal cost function well; (3) it is efficient and roughly scales as O(nd) where n is the number of training examples; and, (4) the number of basis functions it requires to achieve an accuracy close to the SVM accuracy is usually far less than the number of SVM support vectors.

Building support vector machines with reduced classifier complexity

A computerized system and method for large scale semi-supervised learning is provided. The training set comprises a mix of labeled and unlabeled examples. Linear classifiers based on support vector machine principles are built using these examples. One embodiment uses a fast design of a linear transductive support vector machine using multiple switching. In another embodiment, mean field annealing is used to form a very effective semi-supervised support vector machine. For both these embodiments the finite Newton method is used as the base method for achieving fast training.

Large scale semi-supervised linear support vector machines

A system and method is provided for supervised learning. A training set is provided to the system. The system selects a training element from the provided training set, and adds the training element to a basis element set I. The system conducts an optimization test on the basis element set I with the selected training element to produce a selection score. The system determines whether the selection score indicates an improvement in optimization for the basis element set I. The system discards the selected element if the selection score does not indicate an improvement, and keeps the selected element if the selection score does indicate improvement. The process may then be repeated for other training elements until either the specified maximum number of basis functions is reached or improvement in optimization is below a threshold. At that point, the chosen set I should represent an optimized basis set.

Matching pursuit approach to sparse Gaussian process regression

The present invention provides a system and method for building fast and efficient support vector classifiers for large data classification problems which is useful for classifying pages from the World Wide Web and other problems with sparse matrices and large numbers of documents. The method takes advantage of the least squares nature of such problems, employs exact line search in its iterative process and makes use of a conjugate gradient method appropriate to the problem. In one embodiment a support vector classifier useful for classifying a plurality of documents, including textual documents, is built by selecting a plurality of training documents, each training document having suitable numeric attributes which are associated with a training document vector, then initializing a classifier weight vector and a classifier intercept for a classifier boundary, the classifier boundary separating at least two document classes, then determining which training document vectors are suitable support vectors, and then re-computing the classifier weight vector and the classifier intercept for the classifier boundary using the suitable support vectors together with an iteratively reindexed least squares method and a conjugate gradient method with a stopping criterion.

Growing community of inventors

Cupertino, CA, United States of America

Sathiya Keerthi Selvaraj