Processing incomplete temporal information in controlled natural language

If not all temporal information is available in a text, humans usually use additional background knowledge to answer questions about the text. In this paper, we first investigate how humans answer this kind of questions and then suggest two general strategies that we can use to answer these questions automatically in a controlled natural language context. We show how background knowledge about events and their effects can be made explicit in a controlled natural language and how this additional information can be translated together with the textual information into a formal notation for automated reasoning. For this purpose, we introduce an Answer Set Programming based version of the Event Calculus and use this reasoning framework as a starting point for answering questions over the formalised textual information.12 page(s