利用團體異構性質在社群網路中尋找K個最具影響力的人

Mining Top-K Influencers in Social Networks on Heterogeneous Communities

吳祥彬(Hsiang-Pin Wu)

指導教授 ： 李新林

國立中正大學/工學院/資訊工程學系/碩士(2015年)

摘要

此一文章主要是探討在團體異構的社群網路中如何找尋K個最具影響力的人。在跨群傳遞影響力時，影響力會有降低的情況。基於此情況，我們的方法會選擇每個群中受歡迎的人物當作是影響者的候選人。當決定了候選人的組成人員後，我們希望這些人在其他群中也是受歡迎的。故會根據連到其他群的邊數做排序並選出最後的前K名當作是最後的影響者。而我們會透過熱傳導模型來推演此一集合所能影響的最大範圍，在兩組真實資料的驗證下，我們的方式於擴散範圍上優於其他的方法。

關鍵字

團體異構性 ； 影響最大化 ； 社群網路

並列摘要

In this thesis, we discuss how to find top-K influencers in social networks based on the different rates of diffusion between heterogeneous communities. The different rates of diffusion means that when influence spreads to other communities, the influence will be reduced. So our method wants to select persons which have good interpersonal relationships in those communities whose size is larger than the average size. It means intra-community edges and inter-community edges of the selected candidate node both larger than the average of communities. Following the popular persons are selected to be candidates, the candidates have stronger connections in other communities will be selected to be influencers. After selecting influencers, the spread of influence will be simulate by heat diffusion model. In experiments of the different rates of diffusion, two real-world data sets will be used to compare our method with others. The results show that our method has better performance on spread of influence in the large scale of network.

並列關鍵字

Heterogeneous Communities ； Influence Maximum ； Social Network

參考文獻

[18] Y. C. Chen, S. H. Chang, C. L. Chou, W. C. Peng, and S. Y. Lee, "Exploring Community Structures for Influence Maximization in Social Networks," ACM SIGKDD International Conference on Knowledge Discovery and Data Mining(KDD), Beijing, China, pp. 1-6, 2012.

[1] A. Clauset, M. E. J. Newman and C. Moore, "Finding community structure in very large networks," Phys. Rev. E, vol. 70, p. 066111, 2004.

[2] A. Lancichinetti, S. Fortunato and F. Radicchi, "Benchmark graphs for testing community detection algorithms," Phys. Rev. E, vol. 78, p. 046110, 2008.

[3] Barnett and Emma, "Facebook cuts six degrees of separation to four", Telegraph, 2012.

[4] B. H. Good, Y.-A. de Montjoye and A. Clauset, "The performance of modularity maximization in practical contexts," Phys. Rev. E, vol. 81, p. 046106, 2010.