Structural Change and Interaction Behavior in Multimodal Networks
Final technical rept. 17 Jul 2009-29 Jul 2010
DSO NATIONAL LABORATORIES SINGAPORE
Pagination or Media Count:
This work presents the results of research focused on mining information from multi-network interactions for the purpose of link prediction. Multi-networks are a generalization of multimodal networks. Multi-network link prediction was evaluated on the HEP-th theoretical high-energy physics authorship multinetwork. Achievements include 1 a novel iterative procedure for estimating unified multinetwork node similarity based only on the network structure information 2 label propagation algorithm to perform adjacency propagation through the similarity matrices to produce a ranking of potential new links. The work also researched modelling engagingness and responsiveness behaviors in email networks and messaging networks. Several quantitative models for measuring user engagingness and responsiveness behaviors were defined.
- Computer Systems