Improved Neural Relation Detection for Knowledge Base Question Answering

Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pages 571–581 Vancouver, Canada, July 30 - August 4, 2017. c⃝2017 Association for Computational Linguistics https://doi.org/10.18653/v1/P17-1053 Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pages 571–581 Vancouver, Canada, July 30 - August 4, 2017. c⃝2017 Association for Computational Linguistics https://doi.org/10.18653/v1/P17-1053 Improved Neural Relation Detection for Knowledge Base Question Answering Mo Yu† Wenpeng Yin? Kazi Saidul Hasan‡ Cicero dos Santos† Bing Xiang‡ Bowen Zhou† †AI Foundations, IBM Research, USA ?Center for Information and Language Processing, LMU Munich ‡IBM Watson, USA {yum,kshasan,cicerons,bingxia,zhou}@us.ibm.com, wenpeng@cis.lmu.de Abstract Relation detection is a core component of many NLP applications including Knowl- edge Base Question Answering (KBQA). In this paper, we propose a hierarchi- cal recurrent neural network enhanced by residual learning which detects KB re- lations given an input question. Our method uses deep residual bidirectional LSTMs to compare questions and rela- tion names via different levels of abstrac- tion. Additionally, we propose a sim- ple KBQA system that integrates entity linking and our proposed relation detec- tor to make the two components enhance each other. Our experimental results show that our approach not only achieves out- standing relation detection performance, but more importantly, it helps our KBQA system achieve state-of-the-art accuracy for both single-relation (SimpleQuestions) and multi-relation (WebQSP) QA bench- marks. 1 Introduction Knowledge Base Question Answering (KBQA) systems answer questions by obtaining informa- tion from KB tuples (Berant et al., 2013; Yao et al., 2014; Bordes et al., 2015; Bast and Haussmann, 2015; Yih et al., 2015; Xu et al., 2016). For an input question, these systems typically generate a KB query, which can be executed to retrieve the answers from a KB. Figure 1 illustrates the process used to parse two sample questions in a KBQA system: (a) a single-relation question, which can be answered with a single <head-entity, relation, tail-entity> KB tuple (Fader et al., 2013; Yih et al., 2014; Bordes et al., 2015); and (b) a more complex case, where some constraints need to be handled for multiple entities in the question. The KBQA system in the figure performs two key tasks: (1) entity linking, which links n-grams in questions to KB entities, and (2) relation detection, which identifies the KB relation(s) a question refers to. The main focus of this work is to improve the relation detection subtask and further explore how it can contribute to the KBQA system. Although general relation detection1 methods are well stud- ied in the NLP community, such studies usually do not take the end task of KBQA into considera- tion. As a result, there is a significant gap between general relation detection studies and KB-specific relation detection. First, in most general relation detection tasks, the number of target relations is limited, normally smaller than 100. In contrast, in KBQA even a small KB, like Freebase2M (Bor- des et al., 2015), contains more than 6,000 relation types. Second, relation detection for KBQA often becomes a zero-shot learning task, since some …

Content truncated to 500 words