Context Data Clustering Based On Modified Fuzzy Possibilistic C-Means Algorithm for Efficient Context-Aware Computing Services

Minimizing the context acquisition task helps context-aware computing services to operate more efficiently. In particular, using current context history to predict future context is useful since it decreases the frequency or need of context sensing. Context-prediction is complicated because context data is extremely diverse: various kinds and types of data, irregular set of correlation among relevant attributes, and quickly changing values. One way of predicting future context is to accurately recognize context patterns. In this paper, we aim to develop a modified fuzzy possibilistic clustering algorithm based on the conventional Fuzzy Possibilistic C-Means (FPCM) in order to obtain higher quality clustering results. For the experiment, we developed a prototype system, SiteGuide, to recommend tourism information in an amusement park. The results of the numerical simulation show that the proposed clustering algorithm gives more accurate prediction results than the Fuzzy C-Means (FCM) and the FPCM methods.