基于Hebbian一致性学习的P2P推荐算法

借鉴社会网络的概念,构建了一个基于信任权值的P2P(peer-to-peer)推荐网络,其中每个对等体作为一个用户代理负责维护其在推荐网络中的信任邻居关系。在此基础上,提出了一种基于Hebbian一致性学习的信任权重学习算法,并且基于相似用户发现机制、信任权重学习规则、潜在邻居调整策略等来自适应地调整用户与邻居用户的信任权重。实验数据证明该算法具有较高的推荐效率、社区构建效率和良好的可扩展性。

P2P Recommendation Algorithm Based on Hebbian Consistency Learning

Inspired by of the theory of Social Networks,constructs a P2P recommendation network based on trust weight,each peer behaves on a user agent acting on maintain its neighborhood.Consequently,this paper reports on a trust weight study algorithm based on the Hebbian consistency learning theory.With the adaptation of similar user exploitation mechanism,trust weight study rule,potential neighbours exchange,this algorithm can adjust the trust weight between neighbours automatically.Experiments have shown that this algorithm can improve recommendation efficiency and construction efficiency and much better scalability.