Improving Question Generation With to the Point Context

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3216–3226, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 3216 Improving Question Generation With to the Point Context Jingjing Li1∗ Yifan Gao1∗ Lidong Bing2 Irwin King1 Michael R. Lyu1 1 Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong 2 R&D Center Singapore, Machine Intelligence Technology Alibaba DAMO Academy 1{lijj, yfgao, king, lyu}@cse.cuhk.edu.hk 2l.bing@alibaba-inc.com Abstract Question generation (QG) is the task of gen- erating a question from a reference sen- tence and a specified answer within the sen- tence. A major challenge in QG is to iden- tify answer-relevant context words to finish the declarative-to-interrogative sentence transfor- mation. Existing sequence-to-sequence neural models achieve this goal by proximity-based answer position encoding under the intuition that neighboring words of answers are of high possibility to be answer-relevant. However, such intuition may not apply to all cases es- pecially for sentences with complex answer- relevant relations. Consequently, the perfor- mance of these models drops sharply when the relative distance between the answer frag- ment and other non-stop sentence words that also appear in the ground truth question in- creases. To address this issue, we propose a method to jointly model the unstructured sen- tence and the structured answer-relevant rela- tion (extracted from the sentence in advance) for question generation. Specifically, the struc- tured answer-relevant relation acts as the to the point context and it thus naturally helps keep the generated question to the point, while the unstructured sentence provides the full infor- mation. Extensive experiments show that to the point context helps our question genera- tion model achieve significant improvements on several automatic evaluation metrics. Fur- thermore, our model is capable of generating diverse questions for a sentence which conveys multiple relations of its answer fragment. 1 Introduction Question Generation (QG) is the task of automat- ically creating questions from a range of inputs, such as natural language text (Heilman and Smith, 2010), knowledge base (Serban et al., 2016) and ∗ These two authors contributed equally. Sentence: The daily mean temperature in January, the area’s coldest month, is 32.6 ◦F (0.3 ◦C); however, temperatures usually drop to 10 ◦F (-12 ◦C) several times per winter and reach 50 ◦F (10 ◦C) several days each winter month. Reference Question: What is New York City ’s daily January mean temperature in degrees celsius ? Baseline Prediction: What is the coldest temperature in Celsius ? Structured Answer-relevant Relation: (The daily mean temperature in January; is; 32.6 ◦F (0.3 ◦C)) Figure 1: An example SQuAD question with the base- line’s prediction. The answer (“0.3 ◦C”) is highlighted. image (Mostafazadeh et al., 2016). QG is an in- creasingly important area in NLP with various ap- plication scenarios such as intelligence tutor sys- tems, open-domain chatbots and question answer- ing dataset construction. In this paper, we focus on question generation from reading comprehension materials like SQuAD …

Content truncated to 500 words