Discovery of Relationships between Interests from Bulletin Board System by Dissimilarity Reconstruction
In this paper, we propose a new method to analyze people’s interests by simulating the gradually changing and transferring mechanism among interests. Different boards in Bulletin Board System (BBS) which focus on various topics serve as representation of people’s interests. A technique named Dissimilarity Reconstruction (DSR) is put forward to discover relationships between the interests. DSR tries to grasp the intrinsic structure of the data set by the following steps. First, Vector Space Model (VSM) representations of the interests are obtained by taking users in BBS as terms and the numbers of messages they post as weights. Second, dissimilarities are calculated from the interest vectors. Finally, the nonlinear technique Isomap is engaged to map the interests into the intrinsic dimensional space of the data set where Euclidean distance between two interests well represents their relationship.
KeywordsDimensionality Reduction Geodesic Distance Vector Space Model Neighborhood Graph Geodesic Path
Unable to display preview. Download preview PDF.
- 1.Kou, Z.B., Zhang, C.S.: Reply Networks on a Bulletin Board System. Phys. Rev. E 67, 36117 (2003)Google Scholar
- 2.Balabanovic, M.: Learning to Surf: Multiagent Systems for AdaptiveWeb Page Recommendation. Ph.D. dissertation, Stanford University: Department Of Computer Science (1998)Google Scholar
- 3.Billsus, D., Pazzani, M.J.: Learning collaborative information filters. In: Proc. Fifteenth International Conference on Machine Learning, Madison, WI, pp. 46–53 (1998)Google Scholar
- 7.Steyvers, M.: Multidimensional Scaling. In: Encyclopedia of Cognitive Science, Macmillan, London (2002)Google Scholar