Public Sentiment Drift Analysis Based on Hierarchical Variational Auto-encoder

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 3762–3767, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 3762 Public Sentiment Drift Analysis Based on Hierarchical Variational Auto-encoder Wenyue Zhang1, Xiaoli Li2, Yang Li1, Suge Wang1, Deyu Li1, Liao Jian1, Jianxing Zheng1 1School of Computer and Information Technology, Shanxi University, Taiyuan 030006, China 2Institute for Infocomm Research, Singapore wsg@sxu.edu.cn Abstract Detecting public sentiment drift is a challeng- ing task due to sentiment change over time. Existing methods first build a classification model using historical data and subsequently detect drift if the model performs much worse on new data. In this paper, we focus on dis- tribution learning by proposing a novel Hierar- chical Variational Auto-Encoder (HVAE) mod- el to learn better distribution representation, and design a new drift measure to directly e- valuate distribution changes between historical data and new data. Our experimental results demonstrate that our proposed model achieves better results than three existing state-of-the- art methods. 1 Introduction Public sentiments, whose information hidden in a temporal sequence of documents, become increas- ingly valuable for real-world applications. Espe- cially, identifying when public drifts occur is of great importance to different stakeholders, such as government agencies, companies and news a- gencies, where they can take proactive actions to avoid damages and pay close attentions to new top- ics/sentiments etc (Hu et al., 2017). However, the dynamic nature of drifts/changes makes it a chal- lenging problem, although concept drift analysis can be applied to focus on detecting variation of data distributions over time. Given historical data and new incoming data, how to accurately detect drift based on distributional change is a critical issue. While existing methods are proposed for senti- ment analysis Wu et al. (2019); Fu et al. (2019); Kong et al. (2019); Hoang et al. (2019); Li et al. (2019), etc., many of them are not designed for sen- timent drift detection task. Xia et al. (2016) focuses on polarity shift detection, but it is at document- level instead of multi-document (or public) lev- el. Recently, some research have been conducted for stream sentiment classification Iosifidis et al. (2017). In particular, statistic processing control (SPC) Ross et al. (2012); Raza et al. (2015) built a detection mechanism which accumulates statis- tic information of drift indications, and Zhou et al. (2018) applied the mechanism in sentiment drift detection. In addition, Bifet and Gavalda (2007) proposed ADWIN method, which used the upper bound of Hoeffding’s inequality to mark drifts. Ad- ditionally, Nguyen et al. (2018) combined ADWIN with variational inference and built an online clas- sification system. Wang et al. (2013) proposed an opinion drift detection method which is threshold- based, restricting its applications. A novel frame- work, proposed by Liu et al. (2016), contains an opinion shift detector based on KL-divergence, while its detection performance is affected by its labeling results. Tsytsarau and Palpanas (2016) de- fined a novel concept of opinion contradictions and used it in a sentiment change detection experiment. However, the pair-wise method …

Content truncated to 500 words