Name
Abstract | ||
|---|---|---|
A new link prediction method using active learning technique, named HALLP, is proposed in this paper. The method provides the user with most useful examples from the large number of unlabeled examples (i.e. unlinked node pairs in the network) for query. Once labeled by users, these examples will be fed to the learner for the improvement of the link predictor in next round. The utility of an example is decided by its uncertainty measure calculated simultaneously by its local structure and its hierarchical structure in networks. Experiments indicate link prediction method can be improved with the use of active learning techniques and both the local structure and global structure are beneficial for selecting examples with high utility. |
Year | DOI | Venue |
|---|---|---|
2014 | 10.4304/jcp.9.3.551-556 | JOURNAL OF COMPUTERS |
Keywords | Field | DocType |
link prediction, active learning, link mining, social network analysis | Data mining,Global structure,Active learning,Pattern recognition,Computer science,Local structure,Link mining,Artificial intelligence,Machine learning | Journal |
Volume | Issue | ISSN |
9 | 3 | 1796-203X |
Citations | PageRank | References |
1 | 0.35 | 12 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Kejia Chen | 1 | 179 | 15.82 |
| Jingyu Han | 2 | 16 | 4.67 |
| Yun Li | 3 | 78 | 11.41 |