Google Inc.

International Business Machines Corporation

Dandapani Sivakumar

Shumeet Baluja

Jay Yagnik

Yushi Jing

Andrew Stephen Tomkins

Shanmugasundaram Ravikumar

Andre Rohe

Todd Christopher Jackson

Sangsoo Sung

Andrew Ames Bunner

Luca De Alfaro

Justin Lewis Kosslyn

Ya Luo

Kevin Snow McCurley

Sasha Sobol

Ramanathan V Guha

Rohan Seth

Pranav Tiwari

Ronald Fagin

Deepak Ravichandran

Jasmine Novak

Mark M Sandler

Alexander Sobol

Sumit Kumar Sanghai

Daniel M Shiffman

Cynthia Dwork

John Anthony Tomlin

Uma Mahadevan

Simeon Naor

Moustafa A Hammad

Phokion Kolaitis

David Paul Williamson

Shanmugasundram Ravikumar

Raghunandan Hulikal Keshavan

Shachi Dave

Selvam Mahalingam

Deepak Raghuwanshi

Shripad V Thite

Implementations disclose crowdsourcing descriptor selection. A method includes receiving, by a processor, a reference to an internet resource from a device, wherein the reference to the internet resource is a uniform resource locator (URL) provided in a post interface, wherein the post interface presents posts to a social network, determining, by the processor, a plurality of descriptors indicative of content within the internet resource, sending, by the processor, the plurality of descriptors to the device for presentation in the post interface according to a priority order, receiving, via the post interface a selection of one of the plurality of descriptors, and generating, by the processor, a post to the social network, the generated post comprising the selected one of the plurality of descriptors.

Crowdsourcing descriptor selection

A system and method for generating a stream of content includes a content stream module that generates a model based on user input and/or prior activities using heterogeneous data sources. The heterogeneous data sources include search, entertainment, social activity and activity on third-party sites. The content stream module retrieves candidate content items that have interests that are similar to the user. The candidate content items are compared to the model and scored based upon interestingness of the content item to the user. The content stream module generates the stream of content from the candidate content items.

Scoring stream items with models based on user interests

A system and method for generating a real-time stream of content from heterogeneous data sources and a real-time index. The heterogeneous data sources include search, entertainment, social activity and activity on third-party sites. A fetching unit retrieves recent content that is indexed in the real-time index according to keywords. A model generation engine generates a model based on user activities. The mixer compares candidate content items from the heterogeneous data sources and the real-time index to the model to generate scores for each item and generates a stream of content based on the scores.

Scoring stream items in real time

Methods and apparatus for ranking suggestions for input of a user based on one or more attributes of the user that are associated with one or more of the suggestions. User attributes of the user's user profile may be matched to user attributes of a suggestion and the suggestion ranked based at least in part on bias measures associated with the matched user attributes of the suggestion. Some implementations are directed to determining, for each of a plurality of suggestions, multiple user attributes and associated bias measures to associate with the suggestion.

Ranking suggestions based on user attributes

Methods and apparatus for ranking search result documents for a query of a user based on one or more attributes of the user that are associated with one or more of the search result documents. User attributes of the user's user profile may be matched to user attributes of a search result document and the search result document ranked based at least in part on bias measures associated with the matched user attributes. Some implementations are directed to determining, for each of a plurality of documents, multiple user attributes and associated bias measures to associate with the document. A user attribute and associated bias measure for the document may be determined based on indicated interactions with the document by a set of users associated with the user attribute.

Ranking search result documents based on user attributes

In one aspect, a method is provided, including the following method operations: receiving a request to share a content page in a post to a social network site; analyzing the content page to identify a plurality of descriptive content; presenting a draft post, the draft post providing options for associating particular ones of the descriptive content for integration with the post; receiving a selection of at least one of the particular ones of the descriptive content; generating the post to the social network site based on the selection; wherein the method is executed by at least one processor.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for detecting local search queries. In one aspect, a method includes accessing a search query log that includes data specifying search queries corresponding to particular geographic regions and for at least one of the search queries corresponding to the particular geographic region generating a geo-query count that represents a total number of times that the search query was received over a specified period. The geo-query count is compared to a corresponding expected query count for the search query, where the expected query count is a baseline number of times that the query is expected to be received. In response to determining that the search query has a geo-query count that exceeds the corresponding expected query count by at least a threshold amount, the particular query is classified as a local query for the particular geographic region.

Geographically local query detection

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring resources. One method includes generating a resource graph, the graph including nodes representing resources, nodes representing queries, edges between nodes representing resources, and edges between nodes representing queries and nodes representing resources. The method further includes calculating a score for each of the resources, wherein the score for a particular resource is based on scores for resources represented by nodes with edges to the node representing particular resource in the graph and scores for queries represented by nodes that have an edge to the node representing particular resource in the graph, and ranking one or more of the resources in response to a first query received from a user based, at least in part, on the scores.

Augmented resource graph for scoring resources

Methods, systems, and apparatus, including computer program products, for determining rewrite source-rewrite target similarity scores. In one aspect the method includes receiving a rewrite source-rewrite target pair; identifying first queries that include the rewrite source and second queries that include the rewrite target; identifying a first web document referenced by a first search result responsive to the first query; identifying third queries for which the first web document was referenced by a third search result responsive to the third query; identifying a second web document that was referenced by a second search result responsive to the second query; identifying one or more fourth queries for which the second web document was referenced by a fourth search result responsive to the fourth query; and determining a similarity score for the rewrite source-rewrite target pair based on a measure of matching terms between third query terms and fourth query terms.

Determining rewrite similarity scores

The subject matter of this specification can be embodied in, among other things, a method that includes generating content-based keywords based on content generated by users of a social network. The method includes labeling nodes comprising user nodes, which are representations of the users, with advertising labels comprising content-based keywords that coincide with advertiser-selected keywords that are based on one or more terms specified by an advertiser. The method also includes outputting, for each node, weights for the advertising labels based on weights of advertising labels associated with neighboring nodes, which are related to the node by a relationship.

Advertiser and user association

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting keywords for resources. In one aspect, a method includes identifying a particular online resource that includes non-text content. Co-visitation data are obtained for the particular resource. The co-visitation data specify one or more co-requested online resources for the particular online resource. Each of the co-requested online resources were requested by a user device within a threshold period of the request for the particular online resource by the user device. Keywords are identified for each of the co-requested online resources, and can include keywords that were selected based on text content of the co-requested online resource. One or more of the identified keywords are selected as keywords for the particular resource.

Selecting keywords using co-visitation information

The subject matter of this specification can be embodied in, among other things, a method that includes inferring labels for videos, users, advertisements, groups of users, and other entities included in a social network system. The inferred labels can be used to generate recommendations such as videos or advertisements in which a user may be interested. Inferred labels can be generated based on social or other relationships derived from, for example, profiles or activities of social network users. Inferred labels can be advantageous when explicit information about these entities is not available. For example, a particular user may not have clicked on any online advertisements, so the user is not explicitly linked to any advertisements.

Video-related recommendations using link structure

Growing community of inventors

Cupertino, CA, United States of America

Dandapani Sivakumar