Abstract

Abstract: Emotion analysis of posts is still challenging becauseof the limited contextual information that they normally contain. In micro-blogenvironments, emoticons are frequently used and they have clear emotionalmeanings. They are important emotional signals for micro-blog emotion analysis.Existing studies typically use emoticons as noisy emotion labels or similaremotion indicators to effectively train classifier but overlook their emotionalpotentiality. I address this issue by constructing an emotional space as a featurerepresentation matrix and projecting emoticons and words into the emotionalspace based on the semantic composition. To improve the performance of emotionanalysis, I propose an Emotion Recognition on Twitter. Emoticon embedding is anemotional space projection operator. By projecting emoticons and words into anemoticon space, it can help identify subjectivity, polarity and emotion inmicro-blog environments. It is more capable of capturing emotion semantic thanother models, so it can improve the emotion analysis performance.

Keywords: EmotionRecognition, Twitter, Text Mining, Natural Language Processing (NLP), Hashtags,Natural Language Processing, Sentiment Analysis, Emoticons, Tweets