RNN for Affects at SemEval-2018 Task 1: Formulating Affect Identification as a Binary Classification Problem

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 162–166 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics RNN for Affects at SemEval-2018 Task 1: Formulating Affect Identification as a Binary Classification Problem Aysu Ezen-Can SAS Inst. aysu.e.can@gmail.com Ethem F. Can SAS Inst. ethfcan@gmail.com Abstract Written communication lacks the multimodal features such as posture, gesture and gaze that make it easy to model affective states. Espe- cially in social media such as Twitter, due to the space constraints, the sources of informa- tion that can be mined are even more limited due to character limitations. These limitations constitute a challenge for understanding short social media posts. In this paper, we present an approach that uti- lizes multiple binary classifiers that represent different affective categories to model Twit- ter posts (e.g., tweets). We train domain- independent recurrent neural network models without any outside information such as af- fect lexicons. We then use these domain- independent binary ranking models to eval- uate the applicability of such deep learning models on the affect identification task. This approach allows different model architectures and parameter settings for each affect cate- gory instead of building one single multi-label classifier. The contributions of this paper are two-folds: we show that modeling tweets with a small training set is possible with the use of RNNs and we also prove that formulating affect identification as a binary classification task is highly effective. 1 Introduction Social media platforms allow users to share infor- mation, communicate with other users, learn about new products, and get latest news. The impor- tance of social media data is getting larger every day as social media usage grows every year (Dug- gan, 2015). Twitter is one such social media plat- form where users can write short posts as well as share links. Twitter is also used for getting news (Center, 2017). A large body of research has been conducted using Twitter data including analyzing user inten- tions (Java et al., 2007), determining influence of users (Romero et al., 2011), predicting retweet counts (Can et al., 2013), classifying sentiments of tweets (Jansen et al., 2009; Agarwal et al., 2011; Neethu and Rajasree, 2013; Kontopoulos et al., 2013; Pak and Paroubek, 2010). All of these studies have one goal in common: understand- ing/modeling information diffuse in Twitter. One aspect of modeling social media posts is focusing on emotional states of users. There has been plenty of efforts on determining affective states (Schwarz and Clore, 1983) and their effects to human behavior for different domains from ed- ucation (Sidney et al., 2005) to health care (Lisetti et al., 2003). For Twitter, this problem is even more challenging as the information source is lim- ited to the number of characters allowed in a single post and multimodal features (e.g., posture, ges- ture, and eye gaze) are not available. In this paper, we formulate affect identification task as a binary classification problem and investi- gate the applicability and effectiveness of domain- independent deep learning models …

Content truncated to 500 words