Knowledge-based conversational agents and virtual storytelling

  • Authors:
  • Paul Tarau;Elizabeth Figa

  • Affiliations:
  • University of North Texas, Denton, Texas;University of North Texas, Denton, Texas

  • Venue:
  • Proceedings of the 2004 ACM symposium on Applied computing
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We describe an architecture for building speech-enabled conversational agents, deployed as self-contained Web services, with ability to provide inference processing on very large knowledge bases and its application to voice enabled chatbots in a virtual storytelling environment. The architecture integrates inference engines, natural language pattern matching components and story-specific information extraction from RDF/XML files. Our Web interface is dynamically generated by server side agents supporting multi-modal interface components (speech and animation). Prolog refactorings of the WordNet lexical knowledge base, FrameNet and the Open Mind common sense knowledge repository are combined with internet meta-search to provide high-quality knowledge sources to our conversational agents. An example of conversational agent with speech capabilities is deployed on the Web at http://logic.csci.unt.edu:8080/wordnet_agent/frame.html. The agent is also accessible for live multi-user text-based chat, through a Yahoo Instant Messenger protocol adaptor, from wired or wireless devices, as the jinni_agent Yahoo IM "handle".