Bidirectional Attentive Memory Networks for Question Answering over Knowledge Bases

Proceedings of NAACL-HLT 2019, pages 2913–2923 Minneapolis, Minnesota, June 2 - June 7, 2019. c⃝2019 Association for Computational Linguistics 2913 Bidirectional Attentive Memory Networks for Question Answering over Knowledge Bases Yu Chen Rensselaer Polytechnic Institute cheny39@rpi.edu Lingfei Wu IBM Research wuli@us.ibm.com Mohammed J. Zaki Rensselaer Polytechnic Institute zaki@cs.rpi.edu Abstract When answering natural language questions over knowledge bases (KBs), different ques- tion components and KB aspects play differ- ent roles. However, most existing embedding- based methods for knowledge base question answering (KBQA) ignore the subtle inter- relationships between the question and the KB (e.g., entity types, relation paths and context). In this work, we propose to directly model the two-way flow of interactions between the questions and the KB via a novel Bidi- rectional Attentive Memory Network, called BAMnet. Requiring no external resources and only very few hand-crafted features, on the WebQuestions benchmark, our method significantly outperforms existing information- retrieval based methods, and remains com- petitive with (hand-crafted) semantic parsing based methods. Also, since we use attention mechanisms, our method offers better inter- pretability compared to other baselines. 1 Introduction With the rapid growth in large-scale knowledge bases (KBs) such as DBPedia (Auer et al., 2007) and FreeBase (Google, 2018), knowledge base question answering (KBQA) has drawn increasing attention over the past few years. Given questions in natural language (NL), the goal of KBQA is to automatically find answers from the underlying KB, which provides a more natural and intuitive way to access the vast underlying knowledge resources. One of the most prominent challenges of KBQA is the lexical gap. For instance, the same question can be expressed in various ways in NL while a KB usually has a canonical lexicon. It is therefore non- trivial to map an NL question to a structured KB. The approaches proposed to tackle the KBQA task can be roughly categorized into two groups: se- mantic parsing (SP) and information retrieval (IR) approaches. SP-based approaches address the prob- lem by constructing a semantic parser that converts NL questions into intermediate logic forms, which can be executed against a KB. Traditional semantic parsers (Wong and Mooney, 2007) require anno- tated logical forms as supervision, and are limited to narrow domains with a small number of logical predicates. Recent efforts overcome these limita- tions via the construction of hand-crafted rules or features (Abujabal et al., 2017; Hu et al., 2018) schema matching (Cai and Yates, 2013), and using weak supervision from external resources (Krish- namurthy and Mitchell, 2012). Unlike SP-based approaches that usually assume a pre-defined set of lexical triggers or rules, which limit their domains and scalability, IR-based ap- proaches directly retrieve answers from the KB in light of the information conveyed in the questions. These IR-based approaches usually do not require hand-made rules and can therefore scale better to large and complex KBs. Recently, deep neural networks have been shown to produce strong re- sults on many NLP tasks. In the field of KBQA, under the umbrella of IR-based approaches, many embedding-based methods (Bordes et …

Content truncated to 500 words