Are Emotions Enumerable or Decomposable? And its Implications for Emotion Processing

Are Emotions Enumerable or Decomposable? And its Implications for Emotion Processing* Ying Chen, Sophia Y. M. Lee, and Chu-Ren Huang Department of Chinese & Bilingual Studies, The Hong Kong Polytechnic University {chenying3176, sophiaym, churenhuang}@gmail.com Abstract. Emotion is a complicated concept, and can be represented in different ways. In this paper, we discuss two kinds of emotion representations: the enumerative representation and the compositional representation. Compared to the enumerative representation, the compositional representation is the less rigid description of an emotion. However, from the perspective of emotion classification and detection, different representations often correspond to different emotion processing task. In the enumerative representation, emotion processing can be considered as single-label classification (detecting one and only one label); in the compositional representation, the task turns into the detection of a vector. In this paper, we explore the impact of these emotion representations in emotion processing, including the trade-off of these representations and the selection of technologies to process emotion. Keywords: emotion, emotion classification, emotion processing, multi-label classification Copyright 2009 by Ying Chen, Sophia Y. M. Lee, and Chu-Ren Huang 1 Introduction Emotions represent one of the most fundamental set of shared human experience, while the recognition and identification of emotions is one of the most crucial human cognitive ability. It is probably not an exaggeration to claim that most human activities are motivated by or designed to excite some emotion. And most events do activate emotion, regardless of whether they are designed to do so. Given the critical roles emotions play in human activities, it is not surprising that sentiment analysis, as coarse-grained account of emotional tendencies (positive, negative, and neutral) became one of the most popular topics in NLP and IE. What is surprising is that there were few studies on emotion computing, which would offer finer-grained information and will be universally applicable regardless of domain and product types. With regard to emotion processing, some works (Tokuhisa et al., 2008; Mihalcea and Liu, 2006) have been done on text, and most of them use the resource from web, i.e. web blog and analysis that can be explored, such as emotion detection (Tokuhisa et al., 2008), emotion classification (Mishne, 2005; Mihalcea and Liu, 2006), and emotion trend prediction (Mishne & Rijke, 2005; Balog & Rijke, 2006). In this paper, we discuss a basic yet important question in emotion analysis: How to classify and represent emotions? Although scientific study of emotion can be traced all the way back to early philosophers, both in the West and in the East, we still lack a standard theory of emotion classification today. In terms of emotion classification, the most urgent issue is the nature of emotion taxonomy. Should human emotions be treated as an enumerable, albeit rather large, set of atomic emotions? Or should human emotions be treated as decomposable as a set of primary emotions and their combinations? For example, in Turner's taxonomy (Turner, 2000), “pride” is decomposed into 92 23rd Pacific Asia Conference on Language, Information and Computation, pages 92–100 “happiness + fear”. This …

Content truncated to 500 words