社交媒体中文本情感语义计算理论和方法

社交媒体是以社会网络为基础，互联网用户发表和分享信息为主要形式的在线交互媒体。在这些信息中包含大量的用户情感文本信息，并通过社交媒体影响现实世界。因此，对社交媒体中的情感文本进行全面、深度的语义分析有着重要的理论意义和应用价值。本课题结合文本情感表达和社交媒体的特点，探索社交媒体中文本情感在多角度下深层语义关系。首先提出了社交媒体情感文本语义表示模型，在此基础上，从情感文本表达层面研究显式和隐式情感分析，在社交媒体层面研究用户个性化情感分析和群体情感分析，在情感理解层面研究情感原因发现和情感预测，从不同层面对社交媒体中文本情感语义展开立体交叉研究。本课题的研究将建立一套面向社交媒体文本的情感语义计算理论、方法和框架，实现一个面向社交媒体的文本情感语义技术集成开放平台，为广大研究者和企业开发者提供技术支持，本课题研究成果将拓展社交媒体情感分析应用，为相关学科发展和国家重大需求提供有力支撑。

Social media is a collective of online communications channels that on the basis of social network and with the form of user generated content. In social media these exists a huge number of subjective text contexts expressing user's sentiment, which could have a strong influence on the real world. Therefore, it is of great importance in terms of both theoretical and practical contributions to thoroughly study sentiment analysis in social media and to understand the deep semantics of texts. In this project, we focus on understanding the deep semantics of texts in Chinese social media by capturing the characteristics of sentiment analysis and social media. Towards this goal, we first introduce a representation learning framework to represent the semantics of texts in social media. Based on the representation learning framework, we study the sentiment semantics of texts in social media from different perspectives, including explicit and implicit sentiment analysis of text, sentiment of individual and community in social media, and sentiment understanding through sentiment cause detection and sentiment prediction. Our goal is to build up a sentiment-specific semantic computational theory of texts in social media, and to develop computational approaches for sentiment computing. As a result, this project will broaden the application of sentiment analysis in social media, and provide technical support to social management, e-commerce and related domains such as artificial intelligence.