报告题目：Like Attracts Like? A Semi-supervised Probabilistic Ensemble for Online Review Spammers Detection

时间：2021年12月15日（星期三）下午14:00-15:30

方式：线上腾讯会议（会议码：659 914 083）

报告人：吴俊杰，北京航空航天大学

报告摘要：Review spammers can harm the trustworthy environment of online platforms by purposefully posting unauthentic ratings and comments for products or the online merchants. Though a vast majority of methods have been proposed to resolve the spammer detection issues in online platforms, several challenges such as the label scarcity and possible collusion between reviewers are still persistent, which limits the detection performances and the adaptations of the methods for the evolving cheating tactics of spammers. Building on the phenomenon of collusive spamming behaviors and the homophily theory, we introduce a reviewer network to account for the explicit co-review relations, and then the partially labeled spammers can be incorporated in the reviewer network to aid the estimations of the unknown reviewers' authenticity in a probabilistic way, in which the challenges of the scarcity of spammer labels can be addressed. We then propose a semi-supervised probabilistic model by collectively modeling both the individual behavioral features and the reviewer network, in which the probability of a user being a spammer can be identified by both the individual features and the connection structure with the labeled spammers within the reviewer network. The empirical evaluations have demonstrated the effectiveness of the model, and the results also suggest that the reviewer network can play a vital role in improving the detection performances under different network conditions with respect to an appropriate weight.

报告人简介：吴俊杰，北京航空航天大学教授、经济管理学院副院长、网安学院双聘教授。获国家自然科学基金委杰出青年基金、全国百篇优秀博士学位论文、中国电子学会技术发明一等奖、中国商业联合会科技进步一等奖。担任北航“数据智能研究中心”主任，国务院学位委第八届学科评议组（管理科学与工程学科）成员，国家自然科学基金委重大研究计划（大数据驱动的管理与决策）指导专家组成员，中国管理科学与工程学会“人工智能技术与管理应用”分会主任委员。长期从事管理科学、信息科学、社会科学的交叉创新研究，主要研究兴趣为数据挖掘、大数据计算、人工智能，并在社会治理、智慧城市、金融科技、智慧医疗等领域开展了广泛的应用