Context-Sensitive Lexicon Features for Neural Sentiment Analysis

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1629–1638, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics Context-Sensitive Lexicon Features for Neural Sentiment Analysis Zhiyang Teng, Duy-Tin Vo and Yue Zhang Singapore University of Technology and Design {zhiyang teng, duytin vo}@mymail.sutd.edu.sg yue zhang@sutd.edu.sg Abstract Sentiment lexicons have been leveraged as a useful source of features for sentiment anal- ysis models, leading to the state-of-the-art accuracies. On the other hand, most ex- isting methods use sentiment lexicons with- out considering context, typically taking the count, sum of strength, or maximum senti- ment scores over the whole input. We pro- pose a context-sensitive lexicon-based method based on a simple weighted-sum model, using a recurrent neural network to learn the sen- timents strength, intensification and negation of lexicon sentiments in composing the sen- timent value of sentences. Results show that our model can not only learn such operation details, but also give significant improvements over state-of-the-art recurrent neural network baselines without lexical features, achieving the best results on a Twitter benchmark. 1 Introduction Sentiment lexicons (Hu and Liu, 2004; Wilson et al., 2005; Esuli and Sebastiani, 2006) have been a useful resource for opinion mining (Kim and Hovy, 2004; Agarwal et al., 2011; Moilanen and Pulman, 2007; Choi and Cardie, 2008; Mohammad et al., 2013; Guerini et al., 2013; Vo and Zhang, 2015). Contain- ing sentiment attributes of words such as polarities and strengths, they can serve to provide a word-level foundation for analyzing the sentiment of sentences and documents. We investigate an effective way to use sentiment lexicon features. A traditional way of deciding the sentiment of a document is to use the sum of sentiment values of It’s an insignificant [criticism]−1→−0.5. Nobody gives a [good]+3→−1 performance in this movie She’s not [terrific]+5→+1 but not [terrible]−5→−1 either. It’s not a very [good]+3→−0.25 movie song! It removes my [doubts]−3→+1. Figure 1: Example sentiment compositions. all words in the document that exist in a sentiment lexicon (Turney, 2002; Hu and Liu, 2004). This simple method has been shown to give surprisingly competitive accuracies in several sentiment analysis benchmarks (Kiritchenko et al., 2014), and is still the standard practice for specific research commu- nities with mature domain-specific lexicons, such as finance (Kearney and Liu, 2014) and product re- views (Ding et al., 2008). More sophisticated sentence-level features such as the counts of positive and negative words, their total strength, and the maximum strength, etc, have also been exploited (Kim and Hovy, 2004; Wilson et al., 2005; Agarwal et al., 2011). Such lexicon fea- tures have been shown highly effective, leading to the best accuracies in the SemEval shared task (Mo- hammad et al., 2013). On the other hand, they are typically based on bag-of-word models, hence suf- fering two limitations. First, they do not explicitly handle semantic compositionality (Polanyi and Za- enen, 2006; Moilanen and Pulman, 2007; Taboada et al., 2011), some examples of which are shown in Figure 1. The composition effects can exhibit in- tricacies such …

Content truncated to 500 words