A9.com, Inc.

Nicolas Bruce Trown

Matthew Hamilton Battles

Deepak Jain

Various embodiments identify candidate phrases users have used to reference a product in previous searches by analyzing query logs of a search engine. Upon identifying a candidate phrase within a volume of unstructured text, the candidate phrase is compared against a corpus of contemporary language. A frequency of each candidate phrase in contemporary language and the same in the query logs are determined and used as an indication to distinguish between genuine product references in the text or their normal use in common language in order to determine whether the volume of unstructured text is suitable, based on the number of references, for a hyperlink to a product page for the product of an electronic marketplace. Furthermore, behavioral data associated with user interaction with various aspects of the hyperlinks can be tracked over time to better identify suitable product references and tailor future hyperlink suggestions.

Growing community of inventors

Pleasanton, CA, United States of America

Nicolas Bruce Trown