TR2001-25

Learning Task Models for Collaborative Discourse (subsumed by TR2002-04)

    •  Andrew Garland, Neal Lesh, and Candy Sidner, "Learning Task Models for Collaborative Discourse (subsumed by TR2002-04)", Tech. Rep. TR2001-25, Mitsubishi Electric Research Laboratories, Cambridge, MA, July 2001.
      BibTeX TR2001-25 PDF
      • @techreport{MERL_TR2001-25,
      • author = {Andrew Garland, Neal Lesh, and Candy Sidner},
      • title = {Learning Task Models for Collaborative Discourse (subsumed by TR2002-04)},
      • institution = {MERL - Mitsubishi Electric Research Laboratories},
      • address = {Cambridge, MA 02139},
      • number = {TR2001-25},
      • month = jul,
      • year = 2001,
      • url = {https://www.merl.com/publications/TR2001-25/}
      • }
  • Research Areas:

    Artificial Intelligence, Data Analytics

Abstract:

Combining general principles about collaboration with a task model for a specific environment allows an agent to adapt its utterences based upon the history of interactions with the user. However, developing models that can be used by a collaborative agent is a significant engineering challenge. Learning techniques that infer an accurate model for a given task from annotated examples can lessen this burden considerably. However, there are is still a noticeable disparity between an accurate model and a model that results in dialogs that a human user is comfortable with.