论文部分内容阅读
转发行为是微博平台上信息传播的主要形式。目前已有的工作大多数聚焦在转发行为的分析和预测。针对给定的一条微博时如何找到其转发者这个问题并没有得到很好的解决。结合贝叶斯个性化排序优化标准(BPR-OPT)和分解机(FM),提出了一种通用的方法用于对微博转发者进行预测,并进一步对影响用户成为转发者的特征因素进行了细致分析,然后根据这些特征,在大规模真实数据集上对微博转发者进行了预测。实验证明该方法能够明显提高预测效果,同时也验证了基于pair-wise和特征相关的方法能更有效解决微博转发者预测问题。
Forwarding behavior is the main form of information dissemination on Weibo platform. Most of the work that has been done so far focuses on the analysis and prediction of forwarding behavior. The question of how to find its forwarder for a given microblog is not well addressed. Combined with BPR-OPT and decomposition machine (FM), a universal method is proposed for predicting the microblog booster’s transponders and further on the characteristic factors that affect the users as transponders Detailed analysis, and then based on these characteristics, the microblogging forwarders predicted in a large-scale real data set. Experiments show that this method can significantly improve the prediction performance, and also verify that the method based on pair-wise and feature correlation can more effectively solve the problem of the prediction of the microblogging transponders.