Research Article
An Exemplar-Based Clustering Model with Loose Constraints in Social Network
@INPROCEEDINGS{10.1007/978-3-030-82562-1_22, author={Bi Anqi and Ying Wenhao}, title={An Exemplar-Based Clustering Model with Loose Constraints in Social Network}, proceedings={Multimedia Technology and Enhanced Learning. Third EAI International Conference, ICMTEL 2021, Virtual Event, April 8--9, 2021, Proceedings, Part I}, proceedings_a={ICMTEL}, year={2021}, month={7}, keywords={Loose constraints Exemplar-based clustering model Message passing Social networks}, doi={10.1007/978-3-030-82562-1_22} }- Bi Anqi
Ying Wenhao
Year: 2021
An Exemplar-Based Clustering Model with Loose Constraints in Social Network
ICMTEL
Springer
DOI: 10.1007/978-3-030-82562-1_22
Abstract
Loose constraints have great effects on the study of message passing through social networks. This paper proposes a novel EEM-LC model who joints the pairwise loose constraints existing in social networks and the exemplar-based clustering model together, and also observes the application prospects of this model. Exemplar-based clustering model directly selects cluster centers from actual samples, so the structure and semantics of the comments on social networks would be preserved accordingly. Besides, EEM-LC unifies the two pairwise link constraints by one mathematical definition, and looses the restrictions of strong constraints. Moreover, on the basis of the Bayesian probability framework, EEM-LC implants loose pairwise constraints into its target function. That is to say, enhanced(\alpha )-expansion move algorithm is capable of optimizing this new model. Experimental results based on several real-world data sets have shown very convincing performance of the proposed EEM-LC model.
