Comparison and evaluation of two approaches of a multilayered QA system applied to temporality

Abstract

This paper compares two approaches to a multilayered Question Answering (QA) architecture suitable for enhancing current QA capabilities with the possibility of processing complex questions. That is, questions whose answer needs to be gathered from pieces of factual information that is scattered in different documents. Specifically, we have designed a layer oriented to process the different types of temporal questions. In the first approach, complex temporal questions are decomposed into simple questions, according to the temporal relations expressed in the original question. In the same way, the answers of each resulting simple question are recomposed, fulfilling the temporal restrictions of the original complex question. In the second approach, temporal information is added to the sub-questions before being processed. Evaluation results show that the second approach outperforms the first one in a 30%.