A dimension-reduction framework for human behavioral time series data
Santi Phithakkitnukoon, Ram Dantu · Open Research Online (The Open University) · 2009
Human-machine interaction has become one of the most active research areas, and influenced several new paradigms of computing such as Social computing,Mo-bile computing, and Pervasive/Ubiquitous computing, which are typically concerned with the study of hu-man user’s behavior to facilitate behavioral modeling and prediction. Human behavioral data are usually high-dimensional time series, which need dimension-reduction strategies to improve the efficiency of com-putation and indexing. In this paper, we present a dimension-reduction framework for human behavioral time series. Generally, recent behavioral data are much more interesting and significant in understanding and predicting human behavior than old ones. Our ba-sic idea is to reduce to data dimensionality by keep-ing more detail on recent behavioral data and less de-tail on older data. We distinguish our work from other recent-biased dimension-reduction techniques by em-phasizing on recent-behavioral data and not just recent data. We experimentally evaluate our approach with synthetic data as well as real data. Experimental results show that our approach is accurate and effective as it outperforms other well-known techniques.