Providing recommendations in social networks by integrating local and global reputation

Highlights：

• We propose a novel approach to extend global reputation models with a local reputation.

• Local reputation is computed on the ego-network of the user, by means of an unsupervised approach.

• We performed an extensive experimental analysis on a data set extracted from a social network.

• Experiments are characterized by a sensitive analysis that consider the relevance given to local and global reputation, threshold to consider a user unreliable, and the dimension of the ego-networks.

• Experiments show that global reputation is useful only with small ego-networks, while the combined usage of global and local reputation leads to predict the expected trust with a high level of precision.