Tipbit Inc.

Ewan Ellis Mellor

Nathaniel Harper Ballou

Gordon John Mangione

Embodiments are directed towards employing an index of relevant items to determine and provide items that are relevant to a query. A data store that includes a plurality of items may be generated for each user. The user data store may be employed to generate an index of relevant items for each user. Relevant items may be pre-ranked in the index based on a degree of relevancy to a corresponding item. The index of relevant items may be employed to determine ranked relevant items associated with at least one item associated with a query. In response to the query, the relevant items may be further ranked or re-ranked based on at least a current time of the query and a time decay function associated the relevant items. A subset of the determined ranked relevant items may be provided to a client device for display to the user.

Inferring relevance based on user interactions with email

Embodiments are directed towards generating user isolated indexes for determining the content and relevance of relationships. User indexes may be generated that correspond to individual users of a larger population of users. Client applications may be accessed to determine items corresponding to the users. Information corresponding to the determined items may be listed in a user index corresponding to each user. Also, other information associated with the users may be stored in a data store separate from the user indexes. A context associated with one or more determined item may be determined based on an analysis of content included in the determined items, or other information. User indexes may be updated based on the determined contexts. Updating the user indexes may include updating the user index of the user associated with the determined items and updating other user indexes.

Growing community of inventors

San Francisco, CA, United States of America

Ewan Ellis Mellor