Question Classification Based on MAC-LSTM
Bengong Yu, Qingtang Xu, Peihang Zhang · 2018
Question classification is very important for Question Answering (QA), and the result of question classification directly affects the quality of QA. Most of the question classification methods are based on supervised learning algorithms which require word embedding and does not consider the interrogative words features. However, question text is usually short and semantic information and word co-occurrence information are insufficient. To address the above problems, this paper proposes a multi-level Attention Convolution LSTM neural network (MAC-LSTM) for question classification. This approach uses the interrogative words attention mechanism to focus on the interrogative words features in the question contexts. At the same time, using the attention mechanism combined with the advantages of convolutional neural network and long-short memory model recurrent neural network (LSTM). MAC-LSTM is able to capture both local features of phrases as well as global and time-series features. Experiments show that, our approach achieves better performance than traditional machine learning method, ordinary convolutional neural network, and traditional LSTM on question classification tasks without the need of prior knowledge.