Learning to Rank Question-Answer Pairs Using Hierarchical Recurrent Encoder with Latent Topic Clustering

Proceedings of NAACL-HLT 2018, pages 1575–1584 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Learning to Rank Question-Answer Pairs using Hierarchical Recurrent Encoder with Latent Topic Clustering Seunghyun Yoon, Joongbo Shin and Kyomin Jung Dept. of Electrical and Computer Engineering Seoul National University, Seoul, Korea {mysmilesh,jbshin,kjung}@snu.ac.kr Abstract In this paper, we propose a novel end-to-end neural architecture for ranking candidate an- swers, that adapts a hierarchical recurrent neu- ral network and a latent topic clustering mod- ule. With our proposed model, a text is en- coded to a vector representation from an word- level to a chunk-level to effectively capture the entire meaning. In particular, by adapting the hierarchical structure, our model shows very small performance degradations in longer text comprehension while other state-of-the-art re- current neural network models suffer from it. Additionally, the latent topic clustering mod- ule extracts semantic information from tar- get samples. This clustering module is use- ful for any text related tasks by allowing each data sample to find its nearest topic cluster, thus helping the neural network model ana- lyze the entire data. We evaluate our mod- els on the Ubuntu Dialogue Corpus and con- sumer electronic domain question answering dataset, which is related to Samsung products. The proposed model shows state-of-the-art re- sults for ranking question-answer pairs. 1 Introduction Recently neural network architectures have shown great success in many machine learning fields such as image classification, speech recognition, ma- chine translation, chat-bot, question answering, and other task-oriented areas. Among these, the automatic question answering (QA) task has long been considered a primary objective of artificial intelligence. In the commercial sphere, the QA task is usually tackled by using pre-organized knowledge bases and/or by using information retrieval (IR) based methods, which are applied in popular intelligent voice agents such as Siri, Alexa, and Google Assis- tant (from Apple, Amazon, and Google, respec- tively). Another type of advanced QA systems is IBM’s Watson who builds knowledge bases from unstructured data. These raw data are also indexed in search clusters to support user queries (Fan et al., 2012; Chu-Carroll et al., 2012). In academic literature, researchers have in- tensely studied sentence pair ranking task which is core technique in QA system. The ranking task selects the best answer among candidates re- trieved from knowledge bases or IR based mod- ules. Many neural network architectures with end- to-end learning methods are proposed to address this task (Yin et al., 2016; Wang and Jiang, 2016; Wang et al., 2017). These works focus on match- ing sentence-level text pair (Wang et al., 2007; Yang et al., 2015; Bowman et al., 2015). There- fore, they have limitations in understanding longer text such as multi-turn dialogue and explanatory document, resulting in performance degradation on ranking as the length of the text become longer. With the advent of the huge multi-turn dialogue corpus (Lowe et al., 2015), researchers have pro- posed neural network models to rank longer text pair (Kadlec et al., 2015; Baudiˇs et al., 2016). …

Content truncated to 500 words