Sentiment Analysis of Tweets using Heterogeneous Multi-layer Network Representation and Embedding

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 8932–8946, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 8932 Sentiment Analysis of Tweets using Heterogeneous Multi-layer Network Representation and Embedding Loitongbam Gyanendro Singh∗, Anasua Mitra∗, Sanasam Ranbir Singh Department of Computer Science and Engineering Indian Institute of Technology Guwahati, Assam, India {gyanendrol9, anasua.mitra, ranbir}@iitg.ac.in Abstract Sentiment classification on tweets often needs to deal with the problems of under-specificity, noise, and multilingual content. This study proposes a heterogeneous multi-layer network- based representation of tweets to generate mul- tiple representations of a tweet and address the above issues. The generated representations are further ensembled and classified using a neural-based early fusion approach. Further, we propose a centrality aware random-walk for node embedding and tweet representations suitable for the multi-layer network. From var- ious experimental analysis, it is evident that the proposed method can address the problem of under-specificity, noisy text, and multilin- gual content present in a tweet and provides better classification performance than the text- based counterparts. Further, the proposed cen- trality aware based random walk provides bet- ter representations than unbiased and other bi- ased counterparts. 1 Introduction With the growing popularity of Twitter, sentiment analysis of tweets has drawn the attention of sev- eral researchers from both academia and industry in recent times. Unlike other regular texts, senti- ment analysis on Twitter text poses plenty of chal- lenges because of various characteristics such as (i) under-specificity due to text limits, (ii) free-form writing such as the presence of user-defined hash- tags, mentions, emoticons, (iii) noisy texts due to the presence of short-form, long-form, multilin- gual, transliterated text, misspelling. Researchers try to address these problems by adopting vari- ous methods like task-specific representation learn- ing (Singh et al., 2020; Pham and Le, 2018; Fu et al., 2018; Tang et al., 2016; Kim, 2014), in- corporating additional information such as hash- ∗Equal contributions. tags (Alfina et al., 2017; Qadir and Riloff, 2014), re- lationship between users (Zhao et al., 2017), multi- source information (Zhou and Huang, 2017), en- sembling (Al-Twairesh and Al-Negheimish, 2019; Araque et al., 2017; Wang et al., 2014), etc. This paper proposes a novel approach to handle the above issues using a heterogeneous multi-layer network representation of a tweet. A multi-layer network is a network formulated by connecting different layers of networks. For example, a hetero- geneous multi-layer network can be formed by con- necting layers of networks of mentions, hashtags, and keywords. Multi-layer networks have shown to provide promising performance in other tasks like community detection and clustering (Hanteer and Rossi, 2019; Luo et al., 2020), node classifi- cation (Li et al., 2018; Zitnik and Leskovec, 2017; Ghorbani et al., 2019), representation learning in graphs (Cen et al., 2019; Zhang et al., 2018; Ni et al., 2018). A tweet or a collection of tweets can be represented by a multi-layer network. An ad- vantage of using network-based representation is that a network can be expanded by adding nodes or shrunk by removing nodes. …

Content truncated to 500 words