特定事件意见领袖挖掘

Identifying event-specific opinion leaders

为挖掘适用于特定事件的意见领袖,提出一种采用特定事件相关的新闻文本构建人物关系网络,结合社会网络分析方法挖掘意见领袖的方法。从特定事件相关的新闻文本中识别人名序列,采用滑动窗口和段落划分的方式确定人物之间的影响关系,计算影响关系的强弱,构建特定事件的人物关系网络。采用改进的LeaderRank算法计算人物关系网络中人物的影响力得分。实验结果表明,该方法能够有效识别特定事件的意见领袖,改进后的LeaderRank算法相比LeaderRank等算法能够更有效地识别意见领袖。

To identify opinion leaders on specific event,a method was proposed using event-specific news articles to build a network of people relationships,combined with social network analysis methods.The sequence of names was identified from news articles related to specific events,sliding window and paragraph partition were used to determine the influence relationship among people,and the strength of influence relationship between them was calculated for constructing the event-specific network of people relationships.An improved LeaderRank algorithm was proposed to calculate the influence scores of people in the network.Experimental results show that the proposed method can effectively identify event-specific opinion leaders,and the improved LeaderRank algorithm can identify opinion leaders more effectively than LeaderRank algorithm etc.