摘要
[目的/意义]微博转发预测对热点话题检测、网络营销及舆情监控等具有重要意义。[方法/过程]针对现有方法未充分考虑转发者个体差异的不足,提出一种融合多种转发习惯的微博转发预测方法。该方法依据转发者的历史数据从转发行为习惯、转发对象习惯和转发内容习惯三个方面挖掘其转发习惯,并利用具体的待预测微博及发布者的信息对转发习惯进行量化,得到对应的预测分量,最后利用逻辑回归算法实现微博转发预测。[结果/结论]在真实新浪微博数据集上的实验结果表明,相对于现有的经典方法,该转发预测方法的最大F值提高了约2.7%。
[Purpose/Significance]Microblog retweet prediction is important for hot topic detection, network marketing and public opinion monitoring.[Method/Process]To solve the problem that the existing methods do not fully consider the individual differences of forwarders, a new microblog retweet prediction method based on multiple retweet habits is proposed. The method uses user’s historical data to mine their retweet habits, including retweet behavior habit, retweet objects habit and retweet contents habit. And then quantify the retweet habits by the microblogs to be predicted and the information of publishers to obtain corresponding prediction components. Finally, the logistic regression algorithm is proposed to predict microblog retweet.[Result/Conclusion]The experimental results on real Sina microblog dataset show that compared to the existing classical method, the maximum F-Measure of the new retweet prediction method is increased by about 2.7%.
作者
徐建民
韩康康
何丹丹
吴树芳
Xu Jianmin;Han Kangkang;He Dandan;Wu Shufang(College of Cyberspace Security and Computer,Hebei University,Baoding 071002;School of Management,Hebei University,Baoding 071002)
出处
《情报杂志》
CSSCI
北大核心
2020年第3期123-129,155,共8页
Journal of Intelligence
基金
国家社会科学基金面上项目“网络信息治理视域下社交网络不可信用户识别研究”(编号:17BTQ068)研究成果之一
关键词
新浪微博
转发预测
转发习惯
个体差异
Sina microblog
retweet prediction
retweet habits
individual differences
