A Question Answering System Built on Domain Knowledge Base

Abstract

AbstractInteractive Question Answering (QA) system is capable of answering users’ questions with managing/understanding the dialogs between human and computer. With the increasing amount of online information, it is highly needed to answer users’ concerns on a specific domain such as health-related questions. In this paper, we proposed a general framework for domain-specific interactive question answering systems which takes advance of domain knowledge bases. First, a semantic parser is generated to parse users’ questions to the corresponding logical forms on basis of the knowledge base structure. Second, the logical forms are translated into query language to further harvest answers from the knowledge base. Moreover, our framework is featured with automatic dialog strategy development which relies on manual intervention in traditional interactive QA systems. For evaluation purpose, we applied our framework to a Chinese interactive QA system development, and used a health-related knowledge base as domain scenario. It shows promising results in parsing complex questions and holding long history dialog.KeywordsQuestion AnsweringBusiness LogicSPARQL QueryDialog SystemQuestion Answering SystemThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.