Abstract: A system is configured to identify an entity referred to in speech or text by comparing the text of the entity mention to a database of the domain of the entity. The system may obtain a group of potential matches from the database and may then discriminatively rank those potential matches according to specific features identified for the context of the speech and/or text.

Abstract: Features are disclosed for processing user queries into a form that can produce relevant results. Spoken user queries can be transcribed into textual queries. Textual queries can be processed using a statistical model to identify entities within the queries. Running searches using the entities rather than the original search query can produce relevant results even when no result would have been obtained by running the original search query. In some embodiments, attributes may be identified and used during the search to narrow the results and potentially produce results that are more relevant.