SwissAlps at SemEval-2017 Task 3: Attention-based Convolutional Neural Network for Community Question Answering

Proceedings of the 11th International Workshop on Semantic Evaluations (SemEval-2017), pages 334–338, Vancouver, Canada, August 3 - 4, 2017. c⃝2017 Association for Computational Linguistics SwissAlps at SemEval-2017 Task 3: Attention-based Convolutional Neural Network for Community Question Answering Jan Deriu Zurich University of Applied Sciences deri@zhaw.ch Mark Cieliebak Zurich University of Applied Sciences ciel@zhaw.ch Abstract In this paper we propose a system for re- ranking answers for a given question. Our method builds on a siamese CNN archi- tecture which is extended by two attention mechanisms. The approach was evaluated on the datasets of the SemEval-2017 com- petition for Community Question Answer- ing (cQA), where it achieved 7th place ob- taining a MAP score of 86.24 points on the Question-Comment Similarity subtask. 1 Introduction Community Question Answering (cQA) describes the task of finding a relevant answer to a never- before seen question (Nakov et al., 2017). The cQA task in SemEval-2017 is subdivided into three subtasks: (a) Question-Comment Similar- ity, (b) Question-Question Similarity, and (c) Question-External Comment Similarity. We par- ticipated at the Question-Comment Similarity sub- task, which consists of re-ranking a set of 10 answers to a given question, such that all the relevant answers are ranked higher than the ir- relevant answers. We evaluated this system on the dataset provided by SemEval-2017 for the Question-Comment Similarity subtask, wich con- sits of approximately 2000 questions with 10 an- swers each. Our system ranked 7th place, achiev- ing a MAP score of 86.2 which was outperformed by 2 points by the 1st ranked system. In this paper we describe the implementation details of our sys- tem, which follows a siamese CNN architecture based on (Severyn and Moschitti, 2015) extended by the attention mechanisms introduced by (Yin et al., 2015). Siamese Architecture Siamese architectures usually consist of two parallel CNNs, each processing one sentence and then using the representations for the classification. Siamese ar- chitectures have been proposed for various tasks, e.g. (Bromley et al., 1993) used the structure for signature verification, and they have been shown to be very useful for modelling sentence pairs: (He et al., 2015), (Severyn and Moschitti, 2015), and (Tan et al., 2015) used the siamese architecture to generate representations for both sentences which then are used for classification. Attention Mechanisms Recently the notion of attention has been introduced in neural network architectures to mimic human behaviour, as we tend to focus on key parts of the sentences to extract relevant parts. Most of the work on attention mechanisms is focused on LSTMs: for instance in (Bahdanau et al., 2014) the authors use an attention mechanism for language translation, and in (Vinyals et al., 2015) the authors use it for generating parse trees. Regarding attention mechanisms for CNNs, we are only aware of (Yin et al., 2015), on which our system is based on. The rest of the paper is structured as follows: in Section 2 we present our model showing the siamese architecture augmented with two different attention mechanisms. In Section 3 we describe our experimental …

Content truncated to 500 words