The effect of answer validation on the performance of Question-Answering systems

Abstract

Question Answering (QA) systems help users to find relevant information. However, current QA systems still return several incorrect answers, which decreases the confidence of users in such systems. Some researchers have suggested the inclusion of validation in QA with the purpose of reducing wrong responses and, therefore, improving results. Nevertheless, there has not been a precise study about how validation contributes to QA results. In this paper, we include a new analysis of the impact of validation on QA results. We reuse data from real QA systems and show that validation enables QA systems to reduce incorrect answers and obtain better results. We think this study will allow researchers to develop better QA systems.