Ontology-Based Context Representation and Reasoning Using OWL and SWRL

Recently, quite a few researches upon context-awareness, ubiquitous computing and automatic surveillance have emerged. Their objectives are to appropriately react to current context situation over the web, rather than in a single machine. This requires exchanging the context of extremely dynamic web environment, as well as reasoning about the context to achieve goal. Thus, context representation and context reasoning play significant roles. This paper presents an ontology-based approach to: 1) employing web ontology language (OWL) to describe context knowledge, and 2) using semantic web rule language (SWRL) to represent rule-based inference rules for context reasoning. OWL describes context knowledge in details by defining relationships among knowledge. As an XML-based language, OWL enjoys good interoperability for knowledge sharing and reuse over the web. Further, SWRL represents rule-based first-order logic (FOL) inference rules, which are expressed in terms of predefined OWL taxonomic context knowledge. By doing so, SWRL rules perfectly integrate with OWL, making it easy to develop and maintain application systems. Besides, as SWRL is also XML-based, it is not so dependent upon inference engine. That is, inference engine can be changed without rewriting the rules for engine-specific syntax. This enhances system maintainability.

I. Introduction

Recently, with the wide use of world-wide web, services are not provided locally. Thus, data should be easy to transmit over the web. Further, knowledge should be well defined so that various hosts can understand the data in real time. Quite a few researches are currently being conducted on context-awareness [8][10], ubiquitous computing [9][11] or automatic security surveillance [6][7] [12][13]. Basically, they extract data from outside environment (called context knowledge), infer or analyze the data, and then respond in real time to environmental situations by providing suitable services to user. As this is done in a context-driven manner, the way context is represented is rather important in developing such systems. The issues below are raised in [5]: a)

Context knowledge is usually represented differently in various systems without a standard, causing poor interoperability, reusability, and distributed composition [5]. Further, as the current representations are lack of semantics to fully represent the hierarchy of context knowledge, it is difficult to infer the knowledge.

b)

Context-awareness systems usually involve some complex situations that cannot be represented by web ontology language (OWL) features alone. Thus, rule representation may be needed.

c)

Poor compatibility between rule and context knowledge makes rule development difficult. In addition, high coupling between rule and inference engine incurs rule re-development cost when engine changes.