Word Emotion Induction for Multiple Languages as a Deep Multi-Task Learning Problem

Proceedings of NAACL-HLT 2018, pages 1907–1918 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Word Emotion Induction for Multiple Languages as a Deep Multi-Task Learning Problem Sven Buechel & Udo Hahn {sven.buechel|udo.hahn}@uni-jena.de Jena University Language & Information Engineering (JULIE) Lab Friedrich-Schiller-Universit¨at Jena, Jena, Germany http://www.julielab.de Abstract Predicting the emotional value of lexical items is a well-known problem in sentiment analy- sis. While research has focused on polarity for quite a long time, meanwhile this early focus has been shifted to more expressive emotion representation models (such as Basic Emo- tions or Valence-Arousal-Dominance). This change resulted in a proliferation of hetero- geneous formats and, in parallel, often small- sized, non-interoperable resources (lexicons and corpus annotations). In particular, the lim- itations in size hampered the application of deep learning methods in this area because they typically require large amounts of input data. We here present a solution to get around this language data bottleneck by rephrasing word emotion induction as a multi-task learn- ing problem. In this approach, the predic- tion of each independent emotion dimension is considered as an individual task and hidden layers are shared between these dimensions. We investigate whether multi-task learning is more advantageous than single-task learn- ing for emotion prediction by comparing our model against a wide range of alternative emo- tion and polarity induction methods featuring 9 typologically diverse languages and a total of 15 conditions. Our model turns out to out- perform each one of them. Against all odds, the proposed deep learning approach yields the largest gain on the smallest data sets, merely composed of one thousand samples. 1 Introduction Deep Learning (DL) has radically changed the rules of the game in NLP by dramatically boost- ing performance figures in almost all applications areas. Yet, one of the major premises of high- performance DL engines is their dependence on huge amounts of training data. As such, DL seems ill-suited for areas where training data are scarce, such as in the field of word emotion induction. We will use the terms polarity and emotion here to distinguish between research focusing on “se- mantic orientation” (Hatzivassiloglou and McKe- own, 1997) (the positiveness or negativeness) of affective states, on the one hand, and approaches which provide predictions based on some of the many more elaborated representational systems for affective states, on the other hand. Originally, research activities focused on polar- ity alone. In the meantime, a shift towards more expressive representation models for emotion can be observed that heavily draws inspirations from psychological theory, e.g., Basic Emotions (Ek- man, 1992) or the Valence-Arousal-Dominance model (Bradley and Lang, 1994). Though this change turned out to be really ben- eficial for sentiment analysis in NLP, a large vari- ety of mutually incompatible encodings schemes for emotion and, consequently, annotation formats for emotion metadata in corpora have emerged that hinder the interoperability of these resources and their subsequent reuse, e.g., on the basis of align- ments or mergers (Buechel and Hahn, 2017). As an alternative …

Content truncated to 500 words