A text mining approach as baseline for QA4MRE'12
Mathias Verbeke, Jesse J. Davis · Lirias · 2012
Abstract. This paper describes the participation of the KU Leuven DTAI team in the pilot task on machine reading of biomedical texts about the Alzheimer disease, which is part of the 2012 Question Answering for Machine Reading Evaluation campaign (QA4MRE’12). The main objective of our research was to develop a text mining system as a strong baseline for the task. Based on the outcome of this system, we want to investigate which types of questions can be answered based solely on the input text and the question string, and for which ones we need more advanced techniques that also consider the previously acquired background knowledge from the reference document collection. Furthermore this should give us some insights into the system behavior for specific question types and background information for the development of a tailored inference algorithm.