Deflection Analysis for Spatially Propagated Web Sensors

We experience or forecast various phenomena (e.g., rain, snow, and earthquake) in the physical world, while we carry out various actions (e.g., blogging, searching, and e-shopping) in the Web world. Many researches have tried to mine the Web for knowledge about various phenomena in the physical world, and also Web services with Web-mined knowledge have been made available for the public. However, there are not enough investigations on how accurately Web-mined data reflect physical-world data. It is socially-problematic to utilize Web-mined data in public Web services without ensuring their accuracy sufficiently. The previous work has introduced various kinds of “Web Sensors” with Temporal Shift and Spatio-Temporal Propagation to extract spatiotemporal numerical data about a targeted physical phenomenon from Web documents searched by linguistic keyword(s) representing the physical phenomenon, and compared them based on their correlation coefficients with Japan Meteorological Agency’s physically-sensed spatiotemporal statistics. To analyze the deflection of Spatially-Propagated Web Sensors, this paper introduces their novel definition based on not only distance but also azimuthal angle between geographic spaces.