面向社交网络中多背景的信任评估模型
MCTE: a Trust Evaluation Model for Multiple Context in Social Networks

针对信任由背景敏感性导致的在社交网络中难以有效评估用户间接信任值的问题,提出了一种面向多交互背景的间接信任评估模型(multiple？？context trust evaluation,MCTE)。该模型利用相关性概念,通过对各背景下网络结构和用户信任关系的综合分析,建立覆盖在信任网络之上的相关性网络,进而利用交互背景的相关性计算跨背景用户的间接信任值。模型避免了多背景以及疏散网络中间接用户信任路径难以寻求,以及信任衰减对评估的影响,有针对性地为用户组建立相关网络,保证了预测的准确性及合理性。对真实社交网络的实验结果表明,MCTE模型不仅可以计算单一背景下用户的间接信任值,更适用于多交互背景下用户信任值的预测。与已有模型相比,评估准确度有较大的提高。
A novel multiple？？context trust evaluation model, named MCTE, is proposed for social networks to deal with the problem that it is difficult to effectively evaluate indirect trust value between users due to the context sensitivity of trust. A relevance network on top of trust networks is established by taking advantages of the relevant concepts, and through a comprehensive analysis of network structures and user’s trust relationship in each context. Then user’s indirect trust across the context is calculated by using the relevance of context. The Model avoids the effect of the trust attenuation on evaluation and the problem that the trust path between indirect users is difficult to find in multiple context and sparse networks, so that relevance networks of user groups can be built, and the evaluation accuracy and reasonableness are ensured. Experimental results on real social networks shows that the MCTE model can compute the indirect trust value in one single context, and is suitable for the prediction of user’s indirect trust in multiple context. Comparison with an existing model shows that the evaluation accuracy of the proposed model improves a lot