Influential nodes are capable of influencing the Social Network (SN) structure and functions. Influential nodes are not pre-defined and need to be identified to better understand and control the network. Trust of the nodes can be an important indicator of their influence and trusted nodes can also affect other nodes. However, analysis of trust and influence relationship in social network communities needs investigation. This study proposes a SNTrust model to find the trust of nodes in a network using a local and global trust and also investigates trust, influence and their relationship in SN communities. Different SN-based influence evaluation approaches named K-core, closeness centrality, eigenvector centrality, and page rank is used to find influential nodes. We have explored the trust and influence of nodes in their communities as well as in the network. Different standard datasets such as Consulting Company dataset, Freeman EIES dataset, Blogcatalogue dataset, and Facebook groups dataset are used for experimentation and Pearson’s correlation and level of significance (p-value) are used for results evaluation. We found a positive linear correlation between the local trust of a node and its influence. It is found that the nodes which are trusted in the network are highly trusted in their communities. There is a strong linear relationship between the influence of a node in the network and community. Furthermore, it is also observed that nodes that are close to each other in a community have high trust among them. The results show that the proposed SNTrust model performed better in finding trust, influence and their correlation.
Social Network (SN) is a collection of different types of social actors, their relationships and interactions. All users in a SN do not behave alike and have differences in their cultural, demographic, educational and professional characteristics. Also, they have different relationships of varying strengths with different people. Likewise, not all the users can have the equal capacity to influence others. To put it another way, the degree of importance of each node can vary in the network (Yang and Xie, 2016). Some users are more influential and have a greater impact on other users. Influential nodes are rare but effective to influence others in a SN (Zhao et al., 2017). Their opinions and actions can motivate people to follow them and to adapt their preferences. Such nodes are needed to control the SN (Yang and Xie, 2016), to spread influence to a larger audience (Zhao et al., 2016) and to build relationships (Zhang et al., 2010), for viral marketing, political campaigning, and brand advertisements (Araujo et al., 2017). Identification of influential nodes among all nodes in the SN is an interesting and intensively studied domain (Yang and Xie, 2016). It has been found that finding the most influential users for influence maximization is an NP-hard problem (Kempe et al., 2003; Wang et al., 2010). However, several techniques have been presented for identification of influential nodes as mentioned in section 3. In SN, users build relationships with each other. A user’s opinions can influence others in the decision-making process so influence relationships are significant to highlight important nodes. For example, an opinion of the “best friend” is considered more valuable to a person as compared to a “friend” relationship. The frequency of communication is also usually high in close friends other than casual relationships. Such SN data can be mined to analyze the trust of a person on another. Trust has different interpretations, and different representations which are widely used in different fields, such as Philosophy, Economics, Psychology, International Relations, Sociology, and Computer science (Cho et al., 2015). It is a multidimensional metric which represents consent of a trustor to have confidence in a trustee for his positive behavior. Trustor and trustee are two major entities of trust where a trustor is a person who evaluates how much he/she credits a trustee whereas a person who is being evaluated by trustor is termed as trustee (Baek and Kim, 2014). In SN, a number of nodes are part of the network, but not all nodes are equally significant or reliable. Their degree of importance is not predefined and it cannot be apparently estimated whether they are trusted or not. To avail the benefits of influential nodes, their credibility is also crucial. Most of the time, SN users come up with a situation where they have to accept friend request from an anonymous person or they have to follow an unknown person without any prior experience and knowledge about him/her. In the case of online shoppings, customers also have to think about the trustworthiness of the seller (Hamdi et al., 2016). Such situations enhance the need of a trust computation method which may help SN users to decide whether to proceed with another party or not.