微博转发预测有助于热点话题检测、个性化微博推荐等，近些年引起了学术界和工业界的广泛关注。然而，现有的关于微博转发预测的研究工作没有充分利用用户之间的多重信任关系的影响。该文提出联合概率模型，把用户之间的多重信任关系融入传统的Bayesian Poisson因子分解（Bayesian Poisson factorization，BPF）模型，从而预测转发行为。该模型命名为TrustBPF，可以灵活地捕获用户之间的各种社交影响。该文进一步把用户之间的信任强度整合到一个框架中。在新浪微博数据集上验证结果表明：在NDCG@3和Precision@3指标上，TrustBPF模型比原始的BPF模型分别提升了90.91%和88.37%。
In recent years, predictions of retweet behavior have attracted growing attention from both academic and industrial fields since the retweets can help with a number of tasks such as hot topic detection and personalized tweet recommendations. However, existing studies on predictions of retweet behavior have not taken full advantage of the effects of multiple social relationships between users. This paper presents a probabilistic model which incorporates multiple trust relationships between users into a traditional Bayesian Poisson factorization (BPF) model to predict retweet behavior. This model can capture the effects of a variety of social influences and integrates user trust strengths into the framework. Tests on the Sina Weibo dataset demonstrate that the model gives 90.91% improvement in the NDCG@3 score and 88.37% improvement in the precision@3 score over a traditional BPF.
[1] SU S, WANG Y K, ZHANG Z B, et al. Identifying and tracking topic-level influencers in the microblog streams[J]. Machine Learning, 2018, 107(3):551-578.
[2] DE MAIO C, FENZA G, GALLO M, et al. Time-aware adaptive tweets ranking through deep learning[J]. Future Generation Computer Systems, 2017, doi:10.1016/j.future.2017.07.039.
[3] 仲兆满, 管燕, 胡云, 等. 基于背景和内容的微博用户兴趣挖掘[J]. 软件学报, 2017, 28(2):278-291. ZHONG Z M, GUAN Y, HU Y, et al. Mining user interests on microblog based on profile and content[J]. Journal of Software, 2017, 28(2):278-291. (in Chinese)
[4] 李洋, 陈毅恒, 刘挺. 微博信息传播预测研究综述[J]. 软件学报, 2016, 27(2):247-263. LI Y, CHEN Y H, LIU T. Survey on predicting information propagation in microblogs[J]. Journal of Software, 2016, 27(2):247-263. (in Chinese)
[5] 刘玮, 贺敏, 王丽宏, 等. 基于用户行为特征的微博转发预测研究[J]. 计算机学报, 2016, 39(10):1992-2006. LIU W, HE M, WANG L H, et al. Research on microblog retweeting prediction based on user behavior features[J]. Chinese Journal of Computers, 2016, 39(10):1992-2006. (in Chinese)
[6] JIANG B, LIANG J G, SHA Y, et al. Message clustering based matrix factorization model for retweeting behavior prediction[C]//Proceedings of the 24th ACM CIKM International on Conference on Information and Knowledge Management. New York:ACM, 2015:1843-1846.
[7] ZHANG Q, GONG Y Y, GUO Y, et al. Retweet behavior prediction using hierarchical dirichlet process[C]//Proceedings of the 29th AAAI Conference on Artificial Intelligence. Austin, Texas:AAAI, 2015:403-409.
[8] ZHANG J, TANG J, ZHONG Y Y, et al. Structinf:Mining structural influence from social streams[C]//Proceedings of the 31st AAAI Conferences on Artificial Intelligence. San Francisco, California, USA:AAAI, 2017:73-80.
[9] GOPALAN P, HOFMAN J M, BLEI D M. Scalable recommendation with hierarchical Poisson factorization[C]//Proceedings of the Thirty-First Conference on Uncertainty in Artificial Intelligence. Arlington, Virginia:AUAI Press, 2015:326-335.
[10] ZHANG J, TANG J, LI J Z, et al. Who influenced you? Predicting retweet via social influence locality[J]. ACM Transactions on Knowledge Discovery from Data (TKDD), 2015, 9(3):25.
[11] JIANG B, LIANG J G, SHA Y, et al. Retweeting behavior prediction based on one-class collaborative filtering in social networks[C]//Proceedings of the 39th SIGIR International ACM SIGIR Conference on Research and Development in Information Retrieval. New York:ACM, 2016:977-980.
[12] CEMGIL A T. Bayesian inference for nonnegative matrix factorisation models[J]. Computational Intelligence and Neuroscience, 2009, 2009:785152.
[13] HOFFMAN M, BLEI D M, WANG C, et al. Stochastic variational inference[J]. Journal of Machine Learning Research, 2013, 14(1):1303-1347.
[14] RENDLE S. Factorization machines[C]//Proceedings of the 10th IEEE International Conference on Data Mining. Sydney, NSW, Australia:IEEE, 2010:995-1000.
[15] CHANEY A J B, BLEI D M, ELIASSI-RAD T. A probabilistic model for using social networks in personalized item recommendation[C]//Proceedings of the 9th ACM Conference on Recommender Systems. New York:ACM, 2015:43-50.
