| Summary: | 碩士 === 國立中興大學 === 資訊科學與工程學系 === 103 === Over the recent years smart devices have become an ubiquitous medium supporting various forms of functionality and are widely accepted for common users. One distinguishing feature for smart devices is the ability of positioning the physical location of a device, and numerous applications based on user location information have been proposed. While the potentials have been foreseen, location based services fundamentally suffer from the problem of lacking an effective and scalable mechanism to bridge the gap between the machine-observed locations and the human understandable places. For example, GPS-based positioning technique produces a GPS coordinate (a longitude and a latitude) for a location, which is difficult to be understood by humans. A mapping between the low-level GPS coordinates and the high-level place semantics is required, as high-level semantics are the basis for most applications. In this thesis, we contribute on this fundamental problem. Differing from the existing solutions on this subject, we start from a novel perspective; we propose to address the place semantic understanding problem by casting it as a classification problem and employ machine learning techniques to automatically infer the types of the places. The key observation is that human behaviors are not random, e.g., people visit restaurants around noon, go for work in the daytime, and stay at home at night. Namely, by properly selecting features, a mechanism for automatically inferring place type semantics can be achieved. In addition to the temporal features, spatial correlation between Wi-Fi access points can be explored for predicting place types. The observation is that due to limited signal coverage by a Wi-Fi access point (typically a range of a few tens of meters), two Wi-Fi access points observed together by a user at the same time, may have similar place type semantics. In other words, one may infer the type of a location from its neighbors. This thesis summarizes our treatment
and findings of leveraging the observations to infer the type of a place. Also, we perform extensive experiments and using real logs collected from recruited participants to evaluate the performance of the proposed techniques. The experiment results demonstrate the effectiveness of the proposed techniques.
|
|---|
