Narrow Results

In this paper, we empirically investigate the statistical properties of the user correlation network in terms of their common rated objects on MovieLens, and find that it has high clustering coefficient and ultra small average distance, which is close to the fully connected network. We argue that the above characteristics come from the fact that large-degree objects build lots of fully connected subnetworks by using the node projection method. By introducing the user global similarity, measured by the product of two users' similarity vectors, we present an effective way to identify users' specific interests by weakening the mainstream interests and noise interests. Numerical results show that we are able to obtain accurate and diverse recommendations by considering the second-order correlation redundant information simultaneously, which outperforms the state-of-the-art collaborative filtering (CF) methods. This work suggests that statistical properties of the user correlation network is an important factor to improve the performances of information filtering algorithms.

Communities in networks expose some intrinsic properties, each of them involves some influential nodes as its cores, around which the entire community grows gradually; the more the common neighbors that exist between a pair of nodes, the larger the possibility of belonging to the same community; the more the neighbors of any one node belong to a community, the larger the possibility that node belongs to that community too. In this paper, we present a novel method, which makes full utilization of these intrinsic properties to detect communities from networks. We iteratively select the node with the largest degree from the remainder of the network as the first seed of a community, then consider its first- and second-order neighbors to identify other seeds of the community, then expand the community by attracting nodes whose large proportion of neighbors have been in the community to join. In this way, we obtain a series of communities. However, some of them might be too small to make sense. Therefore, we merge some of the initial communities into larger ones to acquire the final community structure. In the entire procedure, we try to keep nodes in every community to be consistent with the properties as possible as we can, this leads to a high-quality result. Moreover, the proposed method works with a higher efficiency, it does not need any prior knowledge about communities (such as the number or the size of communities), and does not need to optimize any objective function either. We carry out extensive experiments on both some artificial networks and some real-world networks to testify the proposed method, the experimental results demonstrate that both the efficiency and the community-structure quality of the proposed method are promising, our method outperforms the competitors significantly.