Enhancing Ontology-based Context Modeling with Temporal Vector Space for Ubiquitous Intelligence Shermann S. M. Chan Media Research Institute, Waseda University, Japan shermann@aoni.waseda.jp Qun Jin Dept. Human Informatics and Cognitive Sciences, Waseda University, Japan jin@waseda.jp Abstract Context is the information, which is created and obtained from the surrounding environment for the interaction between humans and computational services. A generic model is a key accessor to the context in any context-aware applications for ubiquitous computing. In the past decades, a number of context modeling techniques have been proposed e.g. markup scheme based, logic-based, graphical, and ontology-based. Since ontology in its nature is a promising tool to specify concepts and interrelations, it has been widely adopted in context modeling. However, in the rapid changing environments, semantics may vary according to the time factors and dynamic group of users. In this paper, we propose an ontology-based context model with temporal vector space in order to complement this deficiency. 1. Introduction The increasing power and availability of distributed computing brings us to the era of ubiquitous computing [1]. Context is the key information, which is created and obtained from the surrounding environment, gathering for the interaction between humans and computational services; we can use context to characterize a situation related to the interaction between users, applications, and the ever- changing environment [2]. Nowadays, a number of context-awareness applications have been developed for a variety of situations and purposes such as (a) using RFID tags to tag objects for reminding [3], (b) using microphones to collect sound source for localization in a home environment [4], (c) using RFID tags, motion detectors, break-beam sensors, pressure mats and contact switches to track people and to recognize activity for automatic health monitoring [5], (d) using RFID augmented objects, infra-red range finder, proximity sensor, and a 2x4 LED-based abstract display etc, to display information by an augmented mirror [6], (e) using microphone and Philips Inca camera [7] for voice recognition, sound localization, object identification and tracking to develop a domestic user-interface robot [8]. In fact, a generic model is a key accessor to the context in any context-aware applications. Some programming frameworks are available for rapid development of context-aware applications e.g. Context Toolkit [2], JCAF [9], DAIDALOS [10], and Siljee et al. [11]. In addition to the programming frameworks, a number of context modeling approaches have been proposed e.g. key-value based, markup scheme based, graphical, object-oriented, logic based, and ontology based [12]. From the W3C [13] and the Semantic Web [25], a number of standard initiatives and tool developments exist for modeling ontology for the web such as RDF (Resource Description Framework) Schema [14, 15], DAML (DARPA Agent Markup Language) [16], OIL (Ontology Inference Language) [17], DAML-S (DAML Services) [18], and OWL (Web Ontology Language) [20] (formerly, DAML+OIL [19]). Since ontology in its nature is a promising tool to specify concepts and interrelations [21, 22], ontologies have been widely adopted in the semantic web community and in the research of context modeling for knowledge representation, knowledge sharing, logic inference, and knowledge reuse e.g. Semantic Spaces [23, 24], CoBrA (Context Broker Architecture) [26], GAS (Gadgetware Architectural Style) Ontology [30, 31], Preuveneers et al. [32], CoOL (Context Ontology Language) [33], and MUPE (Multiple User Profile Merging) [34]. However, in the rapid changing environments, semantics may vary according to time and dynamic user groups. In this paper, we propose an ontology- based context model with temporal vector space in order to complement this deficiency. Proceedings of the 20th International Conference on Advanced Information Networking and Applications (AINA’06) 1550-445X/06 $20.00 © 2006 IEEE