本项目面向社会媒体大数据背景下电子商务领域的商业管理模式创新对数据价值挖掘的迫切需求，在建立大规模真实产品评论文本语料库及相关资源库的基础上，综合运用计算语言学、统计数据建模、深度学习等领域的最新成果，系统地开展社会媒体评论文本中的隐式情感建模的相关理论与关键技术研究。具体内容包括：（1）隐式情感语义单元的识别与极性判别；（2）基于完备表示的隐式观点抽取；（3）隐式观点要素缺省项的识别与恢复；（4）情绪对观点的影响，并研制一个基于微博博文情感分析的产品推荐系统。本项目研究不仅对语言现象建模、丰富中文文本信息处理的基础理论与关键技术方面有着重要的科学意义，同时预期成果也在企业产品信誉深度分析、营销模式创新和大众消费行为指导等方面具有潜在应用价值。

Orienting to the urgent practical needs of data value mining for management model innovation in the field of e-commerce in the context of big data of social media, on the basis of constructing large scale corpus of real product review texts and some relevant resources, synthetically utilizing the latest research results in the fields of computing linguistics, statistic data modeling and deep learning, this project will systematically explore the new theory and key technology of modeling the implicit sentiments hidden in social media review texts. The main contents include: (1) identifying the semantic units of implicit sentiments and judging their polarities; (2) extracting implicit opinions based on its complete expression; (3) identifying and recovering the default options of essential factors of an opinion; (4) the influence of emotion to sentiment. And a product recommendation system will be also developed based on sentiment analysis of microblog texts. This project has an important scientific significance for linguistic phenomenon modeling which can enrich the basic theories and key technologies for Chinese text information processing. Meanwhile, the expectant achievements of this project have the potential application worthiness in deep analysis of enterprise product reputation, marketing pattern innovation and consumer consumption behavior guiding as well as.