Microsoft Technology Licensing, LLC

Panayiotis Tsaparas

Rakesh Agrawal

Ariel Fuxman

Krishnaram N G Kenthapadi

Kannan Achan

Sreenivas Gollapudi

Livia Polanyi

Stelios Paparizos

Alexandros Ntoulas

Yue Lu

Indraneel Mukherjee

Panagiotis Papadimitriou

Lise C Getoor

User generated reviews and scores associated with the reviews may be used to train a review scoring model with textual features of the reviews. The review scoring model may be used to predict scores for newly received reviews. One or more constraints based on social networking application data associated with an author of a review may be used to adjust the predicted score of the review.

Assessing quality of user reviews

A query may be received at a computing device through a network. One or more attribute values that are preferences for a subset of the one or more terms of the query may be identified by the computing device. One or more products or services having associated attributes that have values that match a subset of the identified attribute values may be identified by the computing device, and a subset of the identified products or services may be presented by the computing device through the network. Implementations may also identify latent preferences, that is, preferences that are found for a query even where such a preference is not explicitly part of a term or token of the query.

Determining preferences from user queries

A tree structure has a node associated with each category of a hierarchy of item categories. Child nodes of the tree are associated with sub-categories of the categories associated with parent nodes. Training data including received queries and indicators of a selected item category for each received query is combined with the tree structure by associating each query with the node corresponding to the selected category of the query. When a query is received, a classifier is applied to the nodes to generate a probability that the query is intended to match an item of the category associated with the node. The classifier is applied until the probability is below a threshold. One or more categories associated with the nodes that are closest to the intent of the received query are selected and indicators of items of those categories that match the received query are output.

Determining query intent

As provided herein objects from a source catalog, such as a provider's catalog, can be added to a target catalog, such as an enterprise master catalog, in a scalable manner utilizing catalog taxonomies. A baseline classifier determines probabilities for source objects to target catalog classes. Source objects can be assigned to those classes with probabilities that meet a desired threshold and meet a desired rate. A classification cost for target classes can be determined for respective unassigned source objects, which can comprise determining an assignment cost and separation cost for the source objects for respective desired target classes. The separation and assignment costs can be combined to determine the classification cost, and the unassigned source objects can be assigned to those classes having a desired classification cost.

Object classification using taxonomies

Methods are provided for the classification of search engine queries and associated documents based on search engine query click logs. One or more seed documents or queries are provided that contain content that is representative of a category. A query click log containing information regarding queries entered by at least one user into the search engine and documents subsequently clicked in search engine results corresponding with the queries is analyzed to determine which one or more queries resulted in clicks on the seed documents. Information is stored associating the one or more queries with the category if they resulted in clicks on the seed documents.

Growing community of inventors

San Francisco, CA, United States of America

Panayiotis Tsaparas