Yahoo! Inc.

Nant Holdingsip, LLC

Fluential, LLC

Youssef Billawala

Farzad Ehsani

Demitrios Leo Master

Jun Huang

Yookyung Kim

Kevin Haas

Sudarshan Lamkhede

Yi-An Lin

Jan Pfeifer

The performance of traditional speech recognition systems (as applied to information extraction or translation) decreases significantly with, larger domain size, scarce training data as well as under noisy environmental conditions. This invention mitigates these problems through the introduction of a novel predictive feature extraction method which combines linguistic and statistical information for representation of information embedded in a noisy source language. The predictive features are combined with text classifiers to map the noisy text to one of the semantically or functionally similar groups. The features used by the classifier can be syntactic, semantic, and statistical.

Robust information extraction from utterances

An information retrieval system and computer-based method provide constructing a title for a search result summary of a document through title synthesis, wherein the title is suitable for use in assessing the relevance of the summarized document to a query. Meaningful keywords or key phrases (title components) about the document are Obtained. The title components are classified into pre-established title component classes. When a query is input to which the document is relevant, a title for the document is constructed by arranging title components selected from title component classes to maximize a title utility function. The title utility function may be a query-dependent grade. In addition to the query, the title utility function may also account for constraints under which the title is to be presented to a user.

Automatically constructing titles

Techniques for providing useful information to a user in response to a search query are provided. Based on the search query, one or more potential intents of the user are identified and a plurality of matching resources are identified. For at least one matching resource, a particular abstract template is selected based on the one or more potential intents. Each abstract (a) corresponds to a different intent than any other intent to which any other abstract template of the plurality of abstract templates corresponds, and (b) dictates a different manner of displaying information about a matching resource than any other manner of displaying dictated by any other abstract template of the plurality of abstract templates. A search results page is generated and sent to the user. The search results page includes an abstract for the at least one matching resource. The abstract is displayed based on the particular abstract template.

Intent driven search result rich abstracts

Growing community of inventors

Campbell, CA, United States of America

Youssef Billawala