SCATEAgent: Context-Aware Software Agents for Multi-Modal Travel (original) (raw)

2005, Whitestein Series in Software Agent Technologies

In this paper we describe our research on an agentbased intelligent, flexible, and context-aware multimodal traveler information system, SCATEAgent. The work targets the representation and manipulation of core user, preference, and context models which facilitate highly-customized and adaptable agents playing key roles in an agent-based Advanced Traveler Information System (ATIS). This ATIS finds and fuses information from multiple sources on routes, congestion, incidents, weather, alternative transit modes and schedules, provides proactive real-time updates and context-specific guidance as the user travels, and is vigilant about impacting events. A highly intelligent, personalized, and user-friendly assistant to the traveler, especially in case of transitions among different travel modes, SCATEAgent will promote effective individual traveling and also help to smooth the transportation load in general. The expected contributions include the design of user, preference, context and travel ontologies; user, context, task models based on these ontologies, a set of representations to drive agent behavior and communication, and a compatible integration of rules, machine learning, information retrieval, and semantic web. Currently, we have completed the first stage of our research, producing first pass ontologies and models, an initial prototype of a small-scale test-bed incorporating GPS-enabled cell-phones, and multiple mechanisms for agents to handle and adapt to such models and travel related events.