A Fast Algorithm for Predicting Topics of Scientific Papers Based on Co-authorship Graph Model
- 1.1k Downloads
This paper focuses on the problem of predicting the topic of a paper based on the co-authorship graph. Co-authorship graph is an undirected graph in which paper is represented by a node and two nodes are linked together by a link if they share at least one common author. The approach of link-based object classification (LBC) is based on the assumption that papers in the same neighbourhoods of the co-authorship graph tend to have same topic, and the predicted topic for one node in the graph depends on the topics of the another nodes that linked to it. In order to solve LBC, we have a traditional relaxation labeling to be proposed by Hoche, S., and Flach. Based on this algorithm, we propose an improvement of this algorithm. Our proposed algorithm has the processing speed faster than the traditional one. We test the performance of the proposed algorithm with the ILPnet2 database and compare the experimental result with the traditional algorithm.
Keywordsco-authorship topic relaxation labeling social network
Unable to display preview. Download preview PDF.
- 2.Hoche, S., Flach, P.: Predicting Topics of Scientific Papers from Co-Authorship Graphs: a Case Study. In: Proceedings of the 2006 UK Workshop on Computational Intelligence (UKCI 2006), pp. 215–222 (September 2006)Google Scholar
- 3.The ILPnet2 (online retrieved), http://www.cs.bris.ac.uk/~ILPnet2/Tools/Reports/
- 4.Lu, Q., Getoor, L.: Link-based classification. In: Proceedings of International Conference on Machine Learning (2003)Google Scholar
- 5.Macskassy, S., Provost, F.: A simple relational classifier. In: Workshop on Multi-Relational Data Mining, pp. 64–77 (2003)Google Scholar