【摘要】 GitHub社交平台是代码托管领域的主流平台，拥有超过7300万开发人员.评估GitHub社交网络中用户的影响力对开源成果的学习和应用有重要意义.针对PageRank算法及其改进方法在评估用户影响力时对用户交互行为与用户自身因素分析不全面的局限性，提出了一种引入用户行为权值分配策略的影响力计算方法CUIE(Comprehensive User Influence Evaluation)算法，将用户行为对其他用户影响力的贡献比例作为权值计算用户的CUIE值.基于真实数据的实验结果表明：将用户交互行为作为评价用户影响力分析的因素之一，能使模型获得更好的结果，与传统算法进行对比，在Top-500及以上的情况下，CUIE算法均取得最优的准确率和召回率.此方法能有效发现被传统方法忽略的部分核心用户，可作为传统方法的补充，在舆情分析与引导中也具有一定实用价值.更多还原

【Abstract】 GitHub social platform is the dominant platform in code hosting with more than 73 million developers. Evaluating the influence of users on GitHub’s social network is of great significance to the learning and application of open source achievements. In view of the limitations of PageRank algorithm and its improved method in the evaluation of user influence on user interaction behavior and user’s own factors, A Comprehensive User Influence Evaluation(CUIE) algorithm is proposed to calculate the user’s CUIE value by taking the proportion of the user’s contribution to the influence of other users as the weight. Based on real data, the experiment results show that taking user interaction as one of the factors to evaluate user influence analysis can make the model get better results. Comparing with the traditional method, in the case of a Top-500 and above, CUIE algorithm has achieved the optimal degree of accuracy and recall rate. This method can effectively find part of the core users overlooked by traditional method and can be used as a supplement to traditional methods,which has certain practical value in public opinion analysis and guidance.更多还原