The state of the art in open domain complex question answering: a survey

Romina Etezadi, Mehrnoush Shamsfard

Abstract: Research on question answering (QA) systems has a long tradition. QA systems, as widely used systems in various applications, seek to find the answers to the given questions through the available resources. These systems are expected to be capable of answering various types of questions, including simple questions whose answers can be found in a single passage or sentence and complex questions which need more complicated reasoning to find the answer or their answer should be found by traversing several relations. Nowadays, answering complex questions from texts or structured data is a challenge in QA systems. In this paper, we have a comparative study on QA approaches and systems for answering complex questions. For this purpose, firstly, this paper discusses what a complex question is and surveys different types of constraints that may appear in complex questions. Furthermore, it addresses the challenges of these types of questions, the methods proposed to deal with them, and benchmark datasets used to evaluate their strengths and weaknesses.