Identification of the influential users is one of the most practical analyses in social networks. The importance of this analysis stems from the fact that such users can affect their followers “/friends” viewpoints. This study aims at introducing two new indices to identify the most influential users in the Twitter social network. Four sets of features extracted from user activities, user profile, tweets, and actions performed on tweets are deployed to create the proposed indices. The available methods of detecting the most influential Twitterers either consider a limited set of features or do not accurately measure the effect of each feature. The indices proposed in this paper consider a comprehensive set of features and also provide a time-sensitive rank which can be used to measure the dynamic nature of influence. Moreover, the relative impact of each feature is computed and considered in the indices. We employ the indices to discover the influential Twitter users posting on Paris attacks in 2015, in a comprehensive analysis. The influence trend of users’ tweets in a 21-day period discloses that 76% of the users do not succeed in posting a second influential tweet. Results reveal that the proposed indices can detect both the publicly recognized sources (like celebrities) and also the less known individuals which gain credit by posting several influential tweets after a specific event. We further compare the proposed indices with other available approaches.

中文翻译：

---

在Twitter中衡量对时间敏感的用户影响

识别有影响力的用户是社交网络中最实用的分析之一。这种分析的重要性源于这样的事实，即这些用户可能会影响其关注者的“ /朋友”观点。这项研究旨在引入两个新索引，以识别Twitter社交网络中最具影响力的用户。从用户活动，用户配置文件，tweet和对tweet执行的操作中提取的四组功能被部署以创建建议的索引。检测最有影响力的Twitterer的可用方法要么考虑功能有限，要么不能准确衡量每个功能的效果。本文提出的指标考虑了一组全面的功能，还提供了一个对时间敏感的等级，可以用来衡量影响的动态性质。此外，计算每个特征的相对影响并在索引中加以考虑。我们使用这些索引来进行全面分析，以发现在2015年巴黎袭击事件中具有影响力的Twitter用户。用户在21天内发布的推文的影响趋势显示，有76％的用户未成功发布第二条有影响力的推文。结果表明，拟议的索引既可以检测到公众认可的来源（如名人），也可以检测到鲜为人知的个人，这些个人通过在特定事件后发布几条有影响力的推文获得信誉。我们进一步将建议的指标与其他可用方法进行比较。用户在21天内发布的推文的影响趋势显示，有76％的用户未成功发布第二条有影响力的推文。结果表明，拟议的索引既可以检测到公众认可的来源（如名人），也可以检测到鲜为人知的个人，这些个人通过在特定事件后发布几条有影响力的推文获得信誉。我们进一步将建议的指标与其他可用方法进行比较。用户在21天内发布的推文的影响趋势显示，有76％的用户未成功发布第二条有影响力的推文。结果表明，拟议的索引既可以检测到公众认可的来源（如名人），也可以检测到鲜为人知的个人，这些个人通过在特定事件后发布几条有影响力的推文获得信誉。我们进一步将建议的指标与其他可用方法进行比较。