Coooolll: A Deep Learning System for Twitter Sentiment Classification

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 208–212, Dublin, Ireland, August 23-24, 2014. Coooolll: A Deep Learning System for Twitter Sentiment Classification∗ Duyu Tang†, Furu Wei‡, Bing Qin†, Ting Liu†, Ming Zhou‡ †Research Center for Social Computing and Information Retrieval Harbin Institute of Technology, China ‡Microsoft Research, Beijing, China {dytang, qinb, tliu}@ir.hit.edu.cn {fuwei, mingzhou}@microsoft.com Abstract In this paper, we develop a deep learn- ing system for message-level Twitter sen- timent classification. Among the 45 sub- mitted systems including the SemEval 2013 participants, our system (Coooolll) is ranked 2nd on the Twitter2014 test set of SemEval 2014 Task 9. Coooolll is built in a supervised learning framework by concatenating the sentiment-specific word embedding (SSWE) features with the state-of-the-art hand-crafted features. We develop a neural network with hybrid loss function 1 to learn SSWE, which en- codes the sentiment information of tweets in the continuous representation of words. To obtain large-scale training corpora, we train SSWE from 10M tweets collected by positive and negative emoticons, without any manual annotation. Our system can be easily re-implemented with the publicly available sentiment-specific word embed- ding. 1 Introduction Twitter sentiment classification aims to classify the sentiment polarity of a tweet as positive, nega- tive or neutral (Jiang et al., 2011; Hu et al., 2013; Dong et al., 2014). The majority of existing ap- proaches follow Pang et al. (2002) and employ ma- chine learning algorithms to build classifiers from tweets with manually annotated sentiment polar- ity. Under this direction, most studies focus on ∗ This work was partly done when the first author was visiting Microsoft Research. 1This is one of the three sentiment-specific word embed- ding learning algorithms proposed in Tang et al. (2014). This work is licensed under a Creative Commons At- tribution 4.0 International Licence. Page numbers and pro- ceedings footer are added by the organisers. Licence details: http://creativecommons.org/licenses/by/4.0/ designing effective features to obtain better clas- sification performance (Pang and Lee, 2008; Liu, 2012; Feldman, 2013). For example, Mohammad et al. (2013) implement diverse sentiment lexicons and a variety of hand-crafted features. To leverage massive tweets containing positive and negative e- moticons for automatically feature learning, Tang et al. (2014) propose to learn sentiment-specific word embedding and Kalchbrenner et al. (2014) model sentence representation with Dynamic Con- volutional Neural Network. In this paper, we develop a deep learning sys- tem for Twitter sentiment classification. First- ly, we learn sentiment-specific word embedding (SSWE) (Tang et al., 2014), which encodes the sentiment information of text into the continuous representation of words (Mikolov et al., 2013; Sun et al., 2014). Afterwards, we concatenate the SS- WE features with the state-of-the-art hand-crafted features (Mohammad et al., 2013), and build the sentiment classifier with the benchmark dataset from SemEval 2013 (Nakov et al., 2013). To learn SSWE, we develop a tailored neural net- work, which incorporates the supervision from sentiment polarity of tweets in the hybrid loss function. We learn SSWE from tweets, lever- aging massive tweets with emoticons as distant- supervised …

Content truncated to 500 words