Modeling space-time activities and places for a smart space —a semantic approach

• Main Author: Fan, Junchuan
• Other Authors: Bennett, David A.
• Format: Others
• Language: English
• Published: University of Iowa 2017
• Subjects: Deep Learning; Human mobility; RDF; Semantic Web; Smart space; Space-time Activity; Geography
• Online Access: https://ir.uiowa.edu/etd/5752; https://ir.uiowa.edu/cgi/viewcontent.cgi?article=7229&context=etd

The rapid advancement of information and communication technologies (ICT) has dramatically changed the way people conduct daily activities. One of the reasons for such advances is the pervasiveness of location-aware devices, and people’s ability to publish and receive information about their surrounding environment. The organization, integration, and analysis of these crowdsensed geographic information is an important task for GIScience research, especially for better understanding place characteristics as well as human activities and movement dynamics in different spaces. In this dissertation research, a semantic modeling and analytic framework based on semantic web technologies is designed to handle information related with human space-time activities (e.g., information about human activities, movement, and surrounding places) for a smart space. Domain ontology for space-time activities and places that captures the essential entities in a spatial domain, and the relationships among them. Based on the developed domain ontology, a Resource Description Framework (RDF) data model is proposed that integrates spatial, temporal and semantic dimensions of space-time activities and places. Three different types of scheduled space-time activities (SXTF, SFTX, SXTX) and their potential spatiotemporal interactions are formalized with OWL and SWRL rules. Using a university campus as an example spatial domain, a RDF knowledgebase is created that integrates scheduled course activities and tweet activities in the campus area. Human movement dynamics for the campus area is analyzed from spatial, temporal, and people’s perspectives using semantic query approach. The ontological knowledge in RDF knowledgebase is further fused with place affordance knowledge learned through training deep learning model on place review data. The integration of place affordance knowledge with people’s intended activities allows the semantic analytic framework to make more personalized location recommendations for people’s daily activities.