Distributed agents for online spatial searches

Abstract As the availability and utilisation of online data blossoms, automated online searches—whether to answer a simple question, seek specific sensor readings, or investigate research in a particular domain—have raised a number of issues. Simple search tools do not access the deep web of services and online forms, and cannot handle knowledge domain-specific search problems, but specialist search tools can have a narrow domain and applicability. Some online tools circumvent these problems by putting more filter controls into the hands of users, but this leads to more complex interfaces which can raise usability barriers. A distributed approach, where specialised search agents act autonomously to find contextualised information, can provide a useful compromise between a simple, general search interface and specialist searches. This paper outlines work in progress on design and use of specialist search agents, with a case study to find public transportation bus stops within a spatial region. The approach is demonstrated with a proof of concept web interface, developed to interpret a text query to find and show bus stop locations within a named boundary by coordinating multiple online search agents. Search agents were designed to follow a common model to allow for future development of agent types, including specialist agents used in the case study to search standard open web services and extract spatial features.