SemEval-2007 Task 14: Affective Text

Proceedings of the 4th International Workshop on Semantic Evaluations (SemEval-2007), pages 70–74, Prague, June 2007. c⃝2007 Association for Computational Linguistics SemEval-2007 Task 14: Affective Text Carlo Strapparava FBK – irst Istituto per la Ricerca Scientifica e Tecnologica I-38050, Povo, Trento, Italy strappa@itc.it Rada Mihalcea Department of Computer Science University of North Texas Denton, TX, 76203, USA rada@cs.unt.edu Abstract The “Affective Text” task focuses on the classification of emotions and valence (pos- itive/negative polarity) in news headlines, and is meant as an exploration of the connec- tion between emotions and lexical seman- tics. In this paper, we describe the data set used in the evaluation and the results ob- tained by the participating systems. 1 Introduction All words can potentially convey affective mean- ing. Every word, even those apparently neutral, can evoke pleasant or painful experiences due to their semantic relation with emotional concepts or cate- gories. Some words have emotional meaning with respect to an individual story, while for many others the affective power is part of the collective imagina- tion (e.g., words such as “mum”, “ghost”, “war”). The automatic detection of emotion in texts is becoming increasingly important from an applica- tive point of view. Consider for example the tasks of opinion mining and market analysis, affective computing, or natural language interfaces such as e-learning environments or educational/edutainment games. Possible beneficial effects of emotions on memory and attention of the users, and in general on fostering their creativity are also well-known in the field of psychology. For instance, the following represent examples of applicative scenarios in which affective analysis would give valuable and interesting contributions: Sentiment Analysis. Text categorization according to affective relevance, opinion exploration for market analysis, etc. are just some exam- ples of application of these techniques. While positive/negative valence annotation is an ac- tive field of sentiment analysis, we believe that a fine-grained emotion annotation would in- crease the effectiveness of these applications. Computer Assisted Creativity. The automated generation of evaluative expressions with a bias on some polarity orientation are a key component for automatic personalized advertisement and persuasive communication. Verbal Expressivity in Human Computer Interaction. Future human-computer interaction, accord- ing to a widespread view, will emphasize naturalness and effectiveness and hence the incorporation of models of possibly many hu- man cognitive capabilities, including affective analysis and generation. For example, emo- tion expression by synthetic characters (e.g., embodied conversational agents) is considered now a key element for their believability. Affective words selection and understanding is crucial for realizing appropriate and expressive conversations. The “Affective Text” task was intended as an ex- ploration of the connection between lexical seman- tics and emotions, and an evaluation of various au- tomatic approaches to emotion recognition. The task is not easy. Indeed, as (Ortony et al., 1987) indicates, besides words directly refer- ring to emotional states (e.g., “fear”, “cheerful”) and for which an appropriate lexicon would help, there are words that act only as an indirect reference to 70 emotions depending on the context (e.g. “monster”, “ghost”). We can call …

Content truncated to 500 words