项目名称: 基于知识迁移的网络舆论领袖识别方法及其适应性增强研究
项目编号: No.61472108
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 自动化技术、计算机技术
项目作者: 何慧
作者单位: 哈尔滨工业大学
项目金额: 80万元
中文摘要: 网络舆论领袖在话题观点的形成和舆论传播过程中起着至关重要的作用。发现和识别网络舆论领袖有助于掌握网络舆情中用户的意见趋势。跨越多维度(载体、领域以及时间)的网络舆论领袖识别是当前的挑战性问题。传统舆论领袖识别方法对特定载体与领域有较强依赖,不适合跨不同载体通道和不同主题事件的领袖发现。因此自适应、快速、准确、可迁移的网络舆论领袖建模与识别方法的研究具有重要的理论与应用价值。本项目将重点考虑上述因素,以跨领域舆论领袖自适应识别为核心科学问题,研究跨域领袖识别的自适应知识迁移基础理论和关键技术。力争在跨领域自适应舆情领袖识别模型,多领域知识融合下的舆论领袖识别方法,领袖识别模型的自适应界限等方面取得理论和技术突破,取得创新性成果。在国际上形成本项目研究的特色,并为我国舆论领袖识别技术提供基础理论支撑和实现方案。
中文关键词: 互联网舆情;舆论领袖识别;迁移学习;跨域;迁移界限
英文摘要: Network opinion leaders play a vital role in the establishment of the topic view and the dissemination of the leading opinion. Discovering and identifying the network opinion leaders can help to master public opinion trends. Identifying the opinion leaders across different physical carriers, topic areas and time dimensions, is one of the most challenging issues. However, traditional discovering methods rely heavily on the specific carrier and event, which has limitations to discover opinion leaders across different carriers, topic and time. Thus, adaptive, fast, accurate and portable network leader modeling and discovery methods have important research values in theory and practices. The core issue of this project is to adaptively discover the network opinion leaders across different areas, e.g. carriers, topic and time. The key method is the knowledge transfer learning. The main contents of this project include building the adaptively identifying modeling, identifying the public opinion leaders with multidisciplinary knowledge integration, and determining knowledge transfer learning thresholds, etc. Our project hopes to gain international reputation with innovation achievements and provides theoretical and practical supports for identifying the opinion leaders in China.
英文关键词: Internet public opinion;opinion leader identification;transfer learning;cross-domain;transfer bound