Machine Learning and Computing, International Conference on (2010)
Feb. 9, 2010 to Feb. 11, 2010
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICMLC.2010.25
Although there have been many recent studies of link prediction in co-authorship networks, few have tried to utilize the Semantic information hidden in abstracts of the research documents. We propose to build a link predictor in a co-authorship network where nodes represent researchers and links represent co-authorship. In this method, we use the structure of the constructed graph, and propose to add a semantic approach using abstract information, research titles and the event information to improve the accuracy of the predictor. Secondly, we make use of the fact that researchers tend to work in close knit communities. The knowledge of a pair of researchers lying in the same dense community can be used to improve the accuracy of our predictor further. Finally, we test out hypothesis on the DBLP database in a reasonable time by under-sampling and balancing the data set using decision trees and the SMOTE technique.
Link Prediction, Graph Mining, Social Networks, Data Mining, Machine Learning
Mrinmaya Sachan, Ryutaro Ichise, "Using Abstract Information and Community Alignment Information for Link Prediction", Machine Learning and Computing, International Conference on, vol. 00, no. , pp. 61-65, 2010, doi:10.1109/ICMLC.2010.25