Recurrent Attention Network on Memory for Aspect Sentiment Analysis

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 452–461 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Recurrent Attention Network on Memory for Aspect Sentiment Analysis Peng Chen Zhongqian Sun Lidong Bing∗ Wei Yang AI Lab Tencent Inc. {patchen, sallensun, lyndonbing, willyang}@tencent.com Abstract We propose a novel framework based on neural networks to identify the sentiment of opinion targets in a comment/review. Our framework adopts multiple-attention mechanism to capture sentiment features separated by a long distance, so that it is more robust against irrelevant informa- tion. The results of multiple attentions are non-linearly combined with a recur- rent neural network, which strengthens the expressive power of our model for han- dling more complications. The weighted- memory mechanism not only helps us avoid the labor-intensive feature engineer- ing work, but also provides a tailor-made memory for different opinion targets of a sentence. We examine the merit of our model on four datasets: two are from Se- mEval2014, i.e. reviews of restaurants and laptops; a twitter dataset, for testing its performance on social media data; and a Chinese news comment dataset, for testing its language sensitivity. The experimental results show that our model consistently outperforms the state-of-the-art methods on different types of data. 1 Introduction The goal of aspect sentiment analysis is to iden- tify the sentiment polarity (i.e., negative, neutral, or positive) of a specific opinion target expressed in a comment/review by a reviewer. For exam- ple, in “I bought a mobile phone, its camera is wonderful but the battery life is short”, there are three opinion targets, “camera”, “battery life”, and “mobile phone”. The reviewer has a positive senti- ment on the “camera”, a negative sentiment on the ∗Corresponding author. “battery life”, and a mixed sentiment on the “mo- bile phone”. Sentence-oriented sentiment analysis methods (Socher et al., 2011; Appel et al., 2016) are not capable to capture such fine-grained senti- ments on opinion targets. In order to identify the sentiment of an individ- ual opinion target, one critical task is to model ap- propriate context features for the target in its orig- inal sentence. In simple cases, the sentiment of a target is identifiable with a syntactically nearby opinion word, e.g. “wonderful” for “camera”. However, there are many cases in which opinion words are enclosed in more complicated contexts. E.g., “Its camera is not wonderful enough” might express a neutral sentiment on “camera”, but not negative. Such complications usually hinder con- ventional approaches to aspect sentiment analysis. To model the sentiment of the above phrase- like word sequence (i.e. “not wonderful enough”), LSTM-based methods are proposed, such as target dependent LSTM (TD-LSTM) (Tang et al., 2015). TD-LSTM might suffer from the problem that af- ter it captures a sentiment feature far from the target, it needs to propagate the feature word by word to the target, in which case it’s likely to lose this feature, such as the feature “cost-effective” for “the phone” in “My overall feeling is that the phone, after …

Content truncated to 500 words