FuRongWang at SemEval-2017 Task 3: Deep Neural Networks for Selecting Relevant Answers in Community Question Answering

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 320–325, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics FuRongWang at SemEval-2017 Task 3: Deep Neural Networks for Selecting Relevant Answers in Community Question Answering Sheng Zhang, Jiajun Cheng, Hui Wang, Xin Zhang, Pei Li and Zhaoyun Ding College of Information Systems and Management, National University of Defense Technology, Changsha, P. R. China {zhangsheng,jiajun.cheng,huiwang, zhangxin,peili,zyding}@nudt.edu.cn Abstract We describe deep neural networks frame- works in this paper to address the com- munity question answering (cQA) rank- ing task (SemEval-2017 task 3). Convo- lutional neural networks and bi-directional long-short term memory networks are ap- plied in our methods to extract semantic information from questions and answers (comments). In addition, in order to take the full advantage of question-comment semantic relevance, we deploy interaction layer and augmented features before cal- culating the similarity. The results show that our methods have the excellent effec- tiveness for both subtask A and subtask C. 1 Introduction Answer selection is regarded as a key step in ques- tion answering tasks, especially for community question answering (cQA), which is greatly valu- able for user to retrieve information. Some cQA forums, such as Quora, Qatar Living and Stack Overflow, are quite open to users, providing a con- venient platform for asking and answering ques- tions. Hence, the development of cQA forums makes it urgent to answer questions automatically. SemEval-2017 (Bethard et al., 2017) task 3 (Nakov et al., 2017) is the task for selecting rele- vant answers (or comments) for questions in com- munity question answering (cQA). The data is col- lected from an online cQA forum, Qatar Living, which is close to some real application needs. The task consists of several subtasks, which are de- scribed briefly as follows: • Question-Comment Similarity: Given one question and ten candidates comments {c1, c2, · · · , cn}, the goal is to rank these !"#$% &'()**+", -)$./01-)%23.45 6)7%3. 8)#.)5)"%/%+3" 999 :+'+0/.+%; <+**)"1=/;). >3"7/%)"/%+3" Figure 1: One Basic Deep Learning cQA Frame- work. The question and the comment are mapped into fixed-length word vectors. After that, they are fed into the neural network framework to extract features. Through a fully connected neural net- work, the framework outputs the similarity of the question and the comment. comments according to their relevance to the question. • Question-External Comment Similarity: Given one question and ten candidate ques- tions, which also have ten candidate com- ments. The questions can be relevant or com- pletely irrelevant. As for the original ques- tion, the target is to rank the top ten relevant comments from these 100 candidate com- ments. The relevance between the question and the comment can be divided as “Good”, “Potential- lyUseful” and “Bad”. “PotentiallyUseful” and “Bad” do not make a clear distinction. “Good” comments are considered useful and should be ranked before “PotentiallyUseful” and “Bad” comments. From the above description, these tasks can be regarded as a binary classification tasks: the relation between the question and the comment is …

Content truncated to 500 words