中文摘要：

由于网络用户讨论的主题变化频繁，因此在进行倾向性判定时，难以预先构造出满足各种情况的训练语料。针对上述问题，提出了一种意见领袖识别中的文本倾向性判定方法，进而建立考虑回复帖子倾向性的意见领袖发现模型。该模型建立在影响力扩散概率模型（IDPM）上，模型中引入了考虑帖子倾向性的有效系数。实验表明，该方法是有效的，其前50个的平均精确率相对分别提高了10．97％和5．45％。

英文摘要：

Because the topics changed rapidly which network users discussed, it was difficult to pre-constuct the training cor- pus to meet a variety of situations. To solve the problem above, this paper proposed a method to analyze the orientation in the opinion leaders identification, then established an opinion leader identification model which was based on the influence diffu- sion probability model （IDPM） and introducing the effective coefficient considering the orientation of the replied comments. The experiments show that the proposed method is effective, the average precisions of the top 50 numbers increased by 10. 97% and 5.45% respectively.