Name
Abstract | ||
|---|---|---|
In social networks, link establishment among users is affected by diversity correlations. In this paper, we study the formation of links, map correlations into multidimensional network spaces and apply their behavioral and structural features to the problem of link prediction. First, by exploring user behavioral correlation and network structural correlation, we map them into three network spaces: following space, interaction space and structure space. With a hierarchical process, the coupling relationship between the spaces can be reduced and we can analyze the correlation in each space separately. Second, by taking advantage of the latent Dirichlet allocation (LDA) topic model for dealing with the polysemy and synonym problems, the traditional text modeling method is improved by Gaussian weighting and applied to user behavior modeling. In this way, the expression ability of topics can be enhanced, and improved topic distribution of user behavior can be obtained to mine user correlations in the following space and the interaction space. Moreover, the method can be extended using the hidden naive Bayesian algorithm which is good at reducing attribute independence. By quantifying the dependencies between common neighbors, we can analyze user correlations in the structure space and multiplex the correlations of the other two spaces to link prediction. The experimental results indicate that the method can effectively improve link prediction performance. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1007/s11432-017-9334-3 | SCIENCE CHINA Information Sciences |
Keywords | Field | DocType |
social networks, multidimensional network spaces, link prediction, LDA, hidden naive Bayes | Data mining,Latent Dirichlet allocation,Mathematical optimization,Weighting,Naive Bayes classifier,Multidimensional network,Correlation,Gaussian,Topic model,Structure space,Mathematics | Journal |
Volume | Issue | ISSN |
61 | 11 | 1674-733X |
Citations | PageRank | References |
0 | 0.34 | 31 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Yunpeng Xiao | 1 | 33 | 10.88 |
| Xixi Li | 2 | 0 | 0.68 |
| Yuanni Liu | 3 | 0 | 0.34 |
| Hong Liu | 4 | 34 | 2.69 |
| Qian Li | 5 | 1 | 3.39 |