Volume 37, Number 3, June 2019
|Page(s)||558 - 564|
|Published online||20 September 2019|
Aspect Category Detection Based on Attention Mechanism and Bi-Directional LSTM
School of Computer Science, Northwestern Polytechnical University, Xi’an, 710072, China
2 Key Laboratory of Big Data Storage and Management, Northwestern Polytechnical University, Ministry of Industry and Information Technology, Xi’an, 710072, China
3 Unit 95806 of PLA, Beijing 100076, China
Online reviews play an increasingly important role in users' purchase decisions. E-commerce websites provide massive user reviews, but it is hard for individuals to make full use of the information. Therefore, it is an urgent task to classify, analyze and summarize the massive comments. In this paper, a model based on attention mechanism and bi-directional long short-term memory (BLSTM) is used to identify the categories of these review objects for the classification of the reviews. The model first uses BLSTM to train the review in the form of word vectors; then according to the part-of-speech, the output vectors of the BLSTM are given corresponding weights. The weights as prior knowledge can guide the learning of attention mechanism to enhance the classification accuracy; finally, the attention mechanism is used to capture category-related important features which are used for category determination. Experiments on the SemEval data set show that our model outperforms the state-of-the-art methods on aspect category detection.
在线评论在用户的购买决策中起到日益重要的作用，电商网站提供海量的用户评论，但是个体很难充分利用所有信息。因此，对这些评论进行分类、分析和汇总是很迫切的任务。首次提出一个基于注意力机制和双向LSTM（bi-directional long short-term memory，BLSTM）的模型来判定评论对象的类别，用于评论的分类。模型首先使用BLSTM对词向量形式的评论进行训练；然后根据词性为BLSTM的输出向量赋予相应权重，权重作为先验知识能指导注意力机制的学习；最后使用注意力机制捕捉与类别相关的重要信息用于类别判定。在SemEval数据集上进行了实验，结果表明，模型能有效提高评论对象类别判定的效果，优于其他算法。
Key words: user review / aspect category detection / attention mechanism / bi-directional long short-term memory / classification accuracy
关键字 : 用户评论 / 评论对象类别判定 / 注意力机制 / BLSTM
© 2019 Journal of Northwestern Polytechnical University. All rights reserved.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.