语义信息是连接人类智能和海量的互联网社会媒体数据的重要桥梁，也是提高信息检索效率和准确度的重要工具。随着互联网社会化程度提高，社会媒体信息急速增多，语义提取方式上，人工方法效率不高，自动方法难以保证准确性；语义的分发上，其社会传播性较少被考虑。本项目针对以上问题，基于申请人前期工作基础，提出了一种新的语义提取与分发的思路。项目充分利用社会用户的群体智慧，基于众包计算进行语义提取；针对跨模态语义的关联性进行挖掘，对语义对象进行归一化描述；结合语义挖掘工具和主题模型进行数据处理，利用降维和相关反馈方法减少非主成分和噪音；针对语义的社会性，设计语义的分发机制，并提出匹配与检索算法。本项目旨在提出有效、完善的多源异构社会媒体语义特征提取、融合和分发方法，为社会媒体语义处理提供新的思路，为社会媒体知识挖掘提供新的交互式可视化框架，充分提高语义的知识蕴含度和表达效果。

Semantic information is an important bridge to connect the human intelligence and the mass Internet social media data, and it is also a critical tool to improve the efficiency and accuracy of information retrieval. With the explosion of the Internet social media, semantic extraction is facing challenges that manual annotation is inefficient and the automatic annotation is inaccurate, and the social characteristic is less considered in semantic distribution. To address these challenges, this project proposes a novel solution to semantic extraction and distribution based on our previous work. The project makes full use of the collective wisdom of the social users, and introduces the crowdsourcing computing into the semantic extraction. Firstly, the correlation of cross modal semantics is mined and the semantic objects are normalized. Secondly, based on semantic mining tools and topic models, we employ dimensionality reduction and relevance feedback approaches to reduce non principal components and noise. Finally, we research the storage and distribution mechanism, and design the matching and retrieval algorithms. This project aims to propose an effective and practical cross modal semantic extraction, fusion and distribution mechanism for multi-source and heterogeneous social media, provide a new idea for social media semantic processing, and design a new interactive visualization framework for social media knowledge mining and retrieval to fully improve the semantic knowledge implication and the effect of representation.