Low-Resource Generation of Multi-hop Reasoning Questions

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6729–6739 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6729 Low-Resource Generation of Multi-hop Reasoning Questions Jianxing Yu, Wei Liu, Shuang Qiu, Qinliang Su∗, Kai Wang, Xiaojun Quan, Jian Yin School of Data and Computer Science, Sun Yat-sen University Guangdong Key Laboratory of Big Data Analysis and Processing, China {yujx26, liuw259, qiush9, suqliang}@mail.sysu.edu.cn {wangk73@mail2, quanxj3@mail, issjyin@mail}.sysu.edu.cn Abstract This paper focuses on generating multi-hop reasoning questions from the raw text in a low resource circumstance. Such questions have to be syntactically valid and need to logically cor- relate with the answers by deducing over mul- tiple relations on several sentences in the text. Specifically, we first build a multi-hop genera- tion model and guide it to satisfy the logical ra- tionality by the reasoning chain extracted from a given text. Since the labeled data is limit- ed and insufficient for training, we propose to learn the model with the help of a large scale of unlabeled data that is much easier to obtain. Such data contains rich expressive forms of the questions with structural patterns on syntax and semantics. These patterns can be estimat- ed by the neural hidden semi-Markov model using latent variables. With latent patterns as a prior, we can regularize the generation model and produce the optimal results. Experimental results on the HotpotQA data set demonstrate the effectiveness of our model. Moreover, we apply the generated results to the task of ma- chine reading comprehension and achieve sig- nificant performance improvements. 1 Introduction Question generation (QG) is a hot research topic that aims to create valid and fluent questions cor- responding to the answers by fully understanding the semantics on a given text. QG is widely used in many practical scenarios: including providing prac- tice exercises from course materials for education- al purposes (Lindberg et al., 2013), initiating the dialog system by asking questions (Mostafazadeh et al., 2017), and reducing the labor cost of creating large-scale labeled samples for the QA task (Duan et al., 2017). The mainstream QG methods can be summarized into the rule-based and neural-based models. The first method often transforms the input ∗Corresponding author. text into an intermediate symbolic representation, such as a parsing tree, and then convert the resulting form into a question by well-designed templates or general rules (Hussein et al., 2014). Since rules and templates are hand-crafted, the scalability and generalization of this method are limited. Respec- tively, the neural model usually directly maps the text to question based on neural network (Du and Cardie, 2017), which is entirely data-driven with far less labor. Such a model can be typically re- garded as learning a one-to-one mapping between the text and question. The mapping is mainly used to generate simple questions with a single sentence. However, due to the lack of fine-grained model- ing on the evidential relations on the text, such a method has minimal capability to form the multi- hop questions that require sophisticated …

Content truncated to 500 words