۱- Introduction

۲- Background

۳- An evidential method for correcting noisy information in social networks

۴- Experimentations

۵- Conclusion

References

Abstract

Nowadays, social networks have become an important part of our daily lives. Hence, several researchers have been interested in the study and analysis of the interactions between the entities composing this type of networks. By modeling a social network, we can assign attributes to nodes and links based on network and community structure. These attributes which may be uncertain, imprecise or even noisy, involve obtaining a non-coherent network. In order to remedy this problem, we propose, in this paper, a method that corrects the noise in the network using the theory of belief functions.

Introduction

Nowadays, the use of computer technology and Internet has become essential. As a result, social networks became an important part of our daily lives. Therefore, it is interesting to study and analyze the types of relationships that exist in these networks. To do so, the study of the community structure as well as the nodes and links attributes represent main characteristics that must be taken into account to analyze these networks. In social network analysis [1,2], the observed attributes of social actors are understood in terms of patterns or structures of ties among the units. These ties may be any existing relationship between units; for example friendship, material transactions, etc. Currently, if we observe any social network, we will soon realize that the entities composing this network are grouped, for example, according to a center of interest, a category of age, a preference, etc. In his work, Santo Fortunato [3] explained that communities, also called clusters or modules, represent groups of vertices which probably share common properties and/or play similar roles within the graph. He argues also that the word community itself refers to a social context. In fact, people naturally tend to form groups, within their work environment, family or friends.