News
[2021 APWeb-WAIM] Deep-gAnswer: A Knowledge Based Question Answering System
Yinnian Lin’s paper about knowledge-graph-based natural language question answering, Deep-gAnswer: A Knowledge Based
Question Answering System, has been accepted by APWEB-WAIM 2021.
Knowledge graph is a structural representation of entities and their relationship in the real world. Recently, knowledge-graph-based
natural language question answering (KBQA) has been a hot topic in NLP. A KBQA solution usually consist of entity recognizing, relation
detection, query path generation and result fetching.
This paper is based on our former work gAnswer. gAnwer relies on heuristic rules and priori knowledge for entity recognizing and
relation detection. In the paper the authors exploit the possibility for large-scale pre-trained DNN models to boost entity recognizing
and relation detection, replacing the toolkit used in gAnswer system. Accuracy of the two procedure is greatly improved.