Once the factual entity and context has been determined, entity and context identifier 109 generates a query and provides the query to the knowledge panel system 110. The knowledge panel system 110, in turn, processes the query and generates a responsive knowledge panel. A variety of appropriate information retrieval and relevance determining search algorithms can be used to process the query to identify responsive content.

For example, in FIG. 2, the entity and context identifier 109 may analyze the text behind the cursor 204 and identify multiple entities—bear, bee, bee hive, hungry, etc. However, the entity and context identifier 109, using semantic and linguist models, determines that the dominant intent of the paragraph is directed to “bear” and thus generates the query “bear” and sends the query to the knowledge panel system 110. Alternatively, the knowledge panel system could determine that the information of interest is related to the interactions of the entities mentioned in the document and how they relate to each other. In this example, the interaction may be that bears eat honey produced by bees, and thus a query for “bear diet” could be sent, even though the term “bear diet” does not itself appear in the document.