Learning Representation Mapping for Relation Detection in Knowledge Base Question Answering

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 6130–6139 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 6130 Learning Representation Mapping for Relation Detection in Knowledge Base Question Answering Peng Wu1,2, Shujian Huang1,2 , Rongxiang Weng1,2, Zaixiang Zheng1,2, Jianbing Zhang1,2, Xiaohui Yan3, Jiajun Chen1,2 1National Key Laboratory for Novel Software Technology, Nanjing, China 2Nanjing University, Nanjing, China 3Poisson Lab, Huawei Technologies, Beijing, China {wup, wengrx, zhengzx}@nlp.nju.edu.cn {huangsj, zjb, chenjj}@nju.edu.cn yanxiaohui2@huawei.com Abstract Relation detection is a core step in many nat- ural language process applications including knowledge base question answering. Previous efforts show that single-fact questions could be answered with high accuracy. However, one critical problem is that current approaches only get high accuracy for questions whose relations have been seen in the training data. But for unseen relations, the performance will drop rapidly. The main reason for this prob- lem is that the representations for unseen re- lations are missing. In this paper, we propose a simple mapping method, named representa- tion adapter, to learn the representation map- ping for both seen and unseen relations based on previously learned relation embedding. We employ the adversarial objective and the re- construction objective to improve the mapping performance. We re-organize the popular Sim- pleQuestion dataset to reveal and evaluate the problem of detecting unseen relations. Ex- periments show that our method can greatly improve the performance of unseen relations while the performance for those seen part is kept comparable to the state-of-the-art.1 1 Introduction The task of Knowledge Base Question Answer- ing (KBQA) has been well developed in recent years (Berant et al., 2013; Bordes et al., 2014; Yao and Van Durme, 2014). It answers questions using an open-domain knowledge base, such as Freebase (Bollacker et al., 2008), DBpedia (Lehmann et al., 2015) or NELL (Carlson et al., 2010). The knowl- edge base usually contains a large set of triples. 1Our code and data are available at https://github. com/wudapeng268/KBQA-Adapter. Each triple is in the form of ⟨subject, relation, ob- ject⟩, indicating the relation between the subject entity and the object entity. Typical KBQA systems (Yao and Van Durme, 2014; Yin et al., 2016; Dai et al., 2016; Yu et al., 2017; Hao et al., 2018) can be divided into two steps: the entity linking step first identifies the tar- get entity of the question, which corresponds to the subject of the triple; the relation detection step then determines the relation that the question asks from a set of candidate relations. After the two steps, the answer could be obtained by extracting the corresponding triple from the knowledge base (as shown in Figure 1). Our main focus in this paper is the relation de- tection step, which is more challenging because it needs to consider the meaning of the whole question sentence (e.g., the pattern “where was ... born”), as well as the meaning of the candi- date relation (e.g., “place of birth”). For compari- son, the entity linking step benefits more from …

Content truncated to 500 words