دانلود رایگان مقاله انگلیسی پیدا کردن گره های نفوذ در شبکه های اجتماعی از طریق جستجوی عمومی

Finding influential nodes in a social network has many practical applications in such areas as marketing, politics and even disease control. Proposed methods often take greedy approaches to find the best k nodes to activate so that the diffusion of activation will spread to the largest number of nodes. In this paper, we study the effects of using a community finding approach to not only maximize the number of activated nodes but to also spread the activation to more segments of the network. After describing our approach we present experiments that explain the effects of this approach.

5 CONCLUSIONS

Finding influential nodes is an interesting problem that can be important to managers in marketing, politics and other diverse areas. Algorithms have been proposed that find an initial set of nodes to activate in order to maximize the number of nodes that will become activated after the initial set of nodes are used in the diffusion model. The problem itself has been previously shown to be NP-hard (Clauset et al., 2006). The approximation algorithms, while tractable are normally quite slow. They are designed to simply find an initial node set to maximize the spread of influence. An interesting extension to the problem is to not only maximize the spread of influence but to widen the spread by covering many different communities within the network. We propose in this paper to use community finding algorithms to not only find a large number of activated nodes but also to cover as many of the communities as possible. We have shown in the experiments that our approach is competitive in many data sets, with the results of the traditional greedy algorithm. While the greedy approach will almost always perform better using a community finding approach will often perform quite well.