Microsoft Technology Licensing, LLC

Chris J Burges

John Carlton Platt

Cormac E Herley

Krysta Marie Svore

Xiao Li

Erin L Renshaw

Silviu-Petru Cucerzan

Yan Ke

Xiaodong Fan

Tao Tao

Denyong Zhou

Techniques for index building are described. Clickcounts of respective training URLs may indicate a number of times that corresponding training URLs were clicked in search engine results. A machine learning algorithm implemented on a computer computes a trained model that is then stored. The clickcounts and respective URLs are passed to the machine learning algorithm to train the model to predict probabilities based on feature vectors of URLs. An index of web pages is built for a set of URLs that identify the web pages. Feature vectors for the URLs are computed. Probabilities of the web pages of the URLs being searched in the future by users may be computed by processing the feature vectors with the trained model. The probabilities may be used to determine which of the URLs to include in the index.

User-driven index selection

A web spam page classifier is described that identifies web spam pages based on features of a search query and web page pair. The features can be extracted from training instances and a training algorithm can be employed to develop the classifier. Pages identified as web spam pages can be demoted and/or removed from a relevancy ranked list.

Web spam page classification using query-dependent data

A 'media stream customizer' customizes buffered media streams by inserting one or more media objects into the stream to maintain an approximate buffer level. Specifically, when media objects such as songs, jingles, advertisements, etc., are deleted from the buffered stream (based on some user specified preferences), the buffer level will decrease. Therefore, over time, as more objects are deleted, the amount of the media stream being buffered continues to decrease, thereby limiting the ability to perform additional deletions from the stream. To address this limitation, the media stream customizer automatically chooses one or more media objects to insert back into the stream, and ensures that the inserted objects are consistent with any surrounding content of the media stream, thereby maintaining an approximate buffer level. In addition, the buffered content can also be stretched using pitch preserving audio stretching techniques to further compensate for deletions from the buffered stream.

System and method for automatically customizing a buffered media stream

A computer-implementable method and system for performing a multi-level search. The method includes performing a primary search that involves executing a query submitted by a user, and returning primary search results (a list of documents, for example). The method further includes automatically performing a secondary search. The secondary search involves identifying at least one third-party source of information based on the query, and automatically assessing a semantic interpretation of the query. The secondary search utilizes the identified at least one third-party source of information and the semantic interpretation of the query to derive secondary search results, which are displayed along with the primary search results.

Multi-level search

A 'Media Identifier' operates on concurrent media streams to provide large numbers of clients with real-time server-side identification of media objects embedded in streaming media, such as radio, television, or Internet broadcasts. Such media objects may include songs, commercials, jingles, station identifiers, etc. Identification of the media objects is provided to clients by comparing client-generated traces computed from media stream samples to a large database of stored, pre-computed traces (i.e., 'fingerprints') of known identification. Further, given a finite number of media streams and a much larger number of clients, many of the traces sent to the server are likely to be almost identical. Therefore, a searchable dynamic trace cache is used to limit the database queries necessary to identify particular traces. This trace cache caches only one copy of recent traces along with the database search results, either positive or negative. Cache entries are then removed as they age.

System and method for speeding up database lookups for multiple synchronized data streams

Extracting features from signals for use in classification, retrieval, or identification of data represented by those signals uses a 'Distortion Discriminant Analysis' (DDA) of a set of training signals to define parameters of a signal feature extractor. The signal feature extractor takes signals having one or more dimensions with a temporal or spatial structure, applies an oriented principal component analysis (OPCA) to limited regions of the signal, aggregates the output of multiple OPCAs that are spatially or temporally adjacent, and applies OPCA to the aggregate. The steps of aggregating adjacent OPCA outputs and applying OPCA to the aggregated values are performed one or more times for extracting low-dimensional noise-robust features from signals, including audio signals, images, video data, or any other time or frequency domain signal. Such extracted features are useful for many tasks, including automatic authentication or identification of particular signals, or particular elements within such signals.

Growing community of inventors

Bellevue, WA, United States of America

Chris J Burges