Yahoo! Inc.

Saurabh Sahni

Pankaj Kothari

Mani Kumar

John Sampson

Steve Ho

Todd Sampson

Ian Kennedy

Raymund Ramos

Chris Goffinet

Emanuel Miller

A method of gathering information relevant to the interests of a user includes receiving activity history associated with the user from a network data source, where the activity history includes a description of a user action initiated by the user, generating a first metadata item based upon the activity history; and storing the first metadata item as user profile information associated with the user in a user profile database. Generating the first metadata item may include extracting text from the activity history and generating the first metadata item based upon the text. The activity history may include a description of a user action. The method may further include providing a server plugin for performing a second method in response to access of web content on a server by the user, the second method comprising retrieving a content object from the server, determining if the first metadata item is relevant to the content object, adding a web link to the web content as a reference to the content object in response to the content object being relevant to the first metadata item; and providing the web content to the user.

Augmenting online content with additional content relevant to user interest

First data relating to a first user and second data relating to a second user are retrieved from a plurality of sources. A social connection is identified, by a computing device, between the first user and the second user using the first data and the second data. A label that describes the social connection is identified, by the computing device, using the first data and the second data. A first profile relating to the first user and a second profile relating to the second user is updated by the computing device to reflect the social connection and the label for the social connection.

Systems and methods for automated labeling of social connections

Embodiments are directed towards ranking candidate pieces of content for publication based on a weighted combination of an 'activity score' and a 'trending score'. In one embodiment, the activity score measures social activity related to the content, such as how much the content has been shared, viewed, or commented on. In one embodiment, the trending score indicates a measure of trending topics contained in the content. In one embodiment, trending topics are keywords derived from an analysis of social activities across the internet, such as blog posts, wall posts, tweets, comments, ratings, tags, videos, pictures, and the like. In one embodiment, a topic is trending if it surpasses a minimum absolute threshold of social activity as well as a percentage increase in social activity.

Method to increase content relevance using insights obtained from user activity updates

Embodiments are directed towards providing related keywords determined to be relevant to a search term using life-streaming data to identify the related keywords. Relevancy is determined by performing the related keyword analysis when life-stream services have been updated. Related keywords may be identified for a given search term using a combination of an aggregated life-stream analysis, a user profile based analysis, and/or a community based life-stream analysis. Each of these mechanisms identify a subset of updated life-streams or profiles to be analyzed to identify keywords that co-occur within each life-steam along with the given search term. The resulting co-occurring keywords may then be ordered based on a frequency of occurrence for the subset of life-streams. A weighted combination of the identified related keywords may be used to display related keywords to a user for use in modifying a search query.

Determining related keywords based on lifestream feeds

Embodiments are directed towards determining frequencies for adaptively crawling user activity feeds, by taking into account user activity patterns, as well as a type or nature of a service to be crawled. A statistical model of activity patterns is generated for each user and user activity source based on various parameters. The activity patterns may be weighted to account for variation in relevancy of user content. Various measures are generated that are adaptive by being updated at a rate determined sufficient to detect user activity pattern changes. The statistical model produces clusters of users for each service. A crawling frequency may be assigned to the various clusters such that different users for a given service may have content crawled at different frequencies.

Growing community of inventors

Indore, India

Saurabh Sahni