Text understanding agents and the Semantic Web

We discuss the challenges involved in adapting the OntoSem natural language processing system to the Web. One set of tasks involves processing Web documents, translating their computed meaning representations from the OntoSem's native KR language into the Semantic Web language OWL, and publishing the results as Web pages and RSS feeds. Another set of tasks works in reverse -- querying the Web for facts needed by OntoSem, translating them from OWL into OntoSem's native KR language and importing the results. A central problem underlying both sets of tasks is that of translating knowledge between OntoSem's KR language and ontologies and those of the Semantic Web. OntoSem2OWL has been developed as a translation system to support these translations. We describe SemNews, an implemented prototype application that demonstrates the process. It monitors RSS feeds of news stories, applies OntoSem to understand the text, and exports the computed facts back to the Web in OWL.