面向网络事件的跨平台异质媒体语义协同与挖掘

Various internet platforms, e.g., social network, news website, video sharing website, etc, become more and more important for information access and spreading. Web events are hot topics formed by the aggregation of large amounts of related web information, and reflect the interaction and fusion of web platforms and real society. However, because of the diversity of platforms, the heterogeneity of media, and the weak correlation of information, it causes great challenges to discover and mine web events comprehensively. This project aims to realize effective discovery and representation of web events through cross-platform cooperation. Therefore, we shall study several key problems with the cross-platform heterogeneous media semantic cooperation, such as semantic and structural consistency description, information relationship modeling and clustering, semantic cooperation and event discovery, as well as comprehensive and personal event content display. We aim to reveal the correlative relationship among heterogeneous media, understand the clustering and semantic cooperation of the cross-platform information. We plan to develop the key technologies of deep analysis and computation, multi-dimensional indexing, factor and entity mining for cross-platform heterogeneous web information. In addition, we will construct an evaluation system which targets at cross-platform heterogeneous media semantic cooperation and mining based on web events. This system will provide theoretical and technical support for the timely and precise discovery and panoramic display of web events. The project will play an important role in web information sensing and monitoring.

以社交网络、新闻网站、视频分享网站等为代表的各类网络平台已成为人们获取和传播信息的重要渠道。网络事件是大量相关联的网络信息聚合形成的热点主题，具有很强的突发性和扩散性，备受社会关注。由于平台的多样性、媒体的异质性和信息间的弱关联性，给准确、及时、全面发掘网络事件带来了很大挑战。本项目旨在通过跨平台间的协同，实现网络事件的高效发掘和呈现。为此，拟研究跨平台异质媒体语义协同中存在的语义结构一致性描述、信息关联与聚合、语义协同与事件发现、事件内容的全方位与个性化呈现等关键科学问题；揭示异质媒体相互间的关联关系、跨平台异质媒体语义信息的聚合机理、跨平台间信息的语义协同机制；突破跨平台异质网络信息的深度分析与计算、多维索引、因素与实体挖掘等关键技术，建立面向网络事件的跨平台异质媒体语义协同与挖掘验证系统，为网络事件的及时准确发掘和全景式呈现提供理论与技术支撑，更好地在网络信息感知和监管方面发挥作用。

网络事件是由大量相关联的网络信息聚合形成的热点主题，具有很强的突发性和扩散性，备受社会关注。该项目针对跨平台的网络事件，研究了跨平台异质媒体语义协同中的语义一致性描述、信息关联与聚合、语义协同与事件发现、事件内容的全方位与个性化呈现等关键科学问题，揭示了异质媒体相互间的关联关系，提出了跨平台异质媒体语义信息的聚合方法，建立了跨平台间信息的语义协同机制，突破了跨平台异质网络信息的深度分析与计算、快速索引、实体挖掘等关键技术，搭建了面向网络事件的跨平台异质媒体语义协同与挖掘验证系统，为网络事件的及时准确发掘和全景式呈现提供理论与技术支撑，可更好地在网络信息感知和监管方面发挥作用。发表及录用论文142篇(其中IEEE/ACM汇刊论文39篇，CCF A类会议33篇，一般SCI国际期刊35篇，国内一级学报2篇），申请国家发明专利12项，获选IEEE Fellow 1名，培养硕士/博士29名。获得教育部自然科学二等奖1项、黑龙江省科学技术二等奖1项及吴文俊人工智能科技进步二等奖1项。