Sanda M. Harabagiu, Finley Lacatusu, and Paul Morarescu
The informative answer to a question needs sometimes to be formatted as a summary of a document or set of documents. This scenario requires a novel Question Answering architecture that relies on a combination of topical relations mined from WordNet and textual information available from documents related to the question’s topic. This paper presents a methodology of pointing out information to be included in a summary by using extraction templates. The paper also describes how the summary is generated using a small set of pasting operators.