Calendar Agents on the Semantic Web

pages in HTML quickly captured public interest, and it evolved at a phenomenal rate into a vast knowledge base. During this growth, however, Web pages' information structure , which is crucial to making the information machine understandable, was sacrificed in favor of presentation and physical design. However, this lack of structure has not deterred agents from using information on the Web. Many agent systems track users' navigation habits as they click throughout the Web so they can suggest new, potentially interesting Web pages to them; others automate access to information sites, replicating human-oriented queries and parsing the resulting pages. Agents have also harvested tax-onomies that guide human navigation toward clustered topics of interest. However, in each case, these agents have had to use bespoke parsers (code written specifically to parse a given Web page, such as Yahoo's restaurant information), and content scrapers (languages for defining rules used to extract information from and manipulate Web pages) to pull out the text before using keyword recognition or natural language recognition techniques to understand the content. 1 The emergence of the Semantic Web 2 has simplified and improved knowledge reuse on the Web. Agents can parse Semantic Web pages to elicit relevant information. They can now understand and reason about information and use it to meet users' needs and provide assistance through reference to ontologies, axioms, and languages such as DARPA Agent Markup Language. 3 The emerging symbio-sis of knowledge about users' desires, preferences, and habits with information garnered from the Semantic Web results in superior agent-based assistance than that provided by existing agents. In Technology Review's " A Smarter Web, " 4 Tim Berners-Lee describes a travel scenario in which a user instructs an agent to organize a trip to a conference. To achieve this, the agent requires knowledge about the user's flight preferences , current schedule, and details such as credit card information. The agent obtains the conference's dates and locations from the semantic markup at the conference Web site. After booking a flight, the agent then stores the resulting itinerary in the user's calendar and cancels or reschedules pending meetings while the user is away. It highlights conference presentations and sessions that might interest the user and sets reminders to notify him or her (through PDA or phone, for example) of an impending talk. RCAL To address many of these tasks, we've developed the RETSINA (Reusable Environment for Task-Structured Intelligent …