Content data items in a first electronic file that correspond to entities in a database comprising categorized entities are identified. Modified content data items of a second electronic file are generated, the modified content data items indicating how the content data items are to be interpreted by a natural language query answering system. Information in the second electronic file indicating how the content data items are to be interpreted by the natural language query answering system are modified in response to user input. The second electronic file is used by the natural language query answering system to facilitate analysis of a natural language query regarding content in the first electronic file. Results of the analysis using the natural language query answering system are output.