CSCI 5582: Artificial Intelligence

Fall 1996

Semester Project Proposal Excerpts

Projects with an Eliza-like structure


Joe Fisher: jfisher@bighorn.dr.lucent.com
Carrie Oswald: carrie@bighorn.dr.lucent.com
Bryan Welch: bryan@bighorn.dr.lucent.com

For our CSCI 5582 artificial intelligence project, we will be working on a custom version of Eliza. This will be an investigative project where we will implement a custom Eliza such that ours will act and respond as would the manager in the Dilbert cartoons when the user makes comments similar to those by Dilbert himself. We plan to choose several themes found in the Dilbert cartoon for implementation, and we will try to include several themes that also lend themselves to work here at Lucent Technologies. For comments that don't fit any theme we've implemented, this Eliza will make more generic, but appropriate, responses. Initial design work will be based on the Eliza information in chapter ten of the book Programming in MacScheme by Mike Eisenberg. Other resources of Eliza implementations and logic will be utilized as they are found, but the basic approach of implementing a generic response engine with several themes will be followed throughout the project.


Jeremy Bargen: nvb@nougat.interealm.com

The goal of the Conspiracy Theorist project is to produce a system that, like the STUDENT system by Daniel Bobrow or the ELIZA system by Joseph Weizenbaum, will read user input and use a pattern-matching algorithm to determine an appropriate response. The Conspiracy Theorist will take as input statements about current events, and will incorporate those statements into a conspiracy theory involving the U.N., the C.I.A., and the space aliens, among other groups. Thus, a simple statement about the weather could cause the Conspiracy Theorist to give a lecture on the secret goverment weather control installation above the Arctic Circle. The ideal goal would be to be able to give the Conspiracy Theorist input consisting of actual newspaper stories on a highly restricted set of topics. In practice, however, this goal may not be realizable.


Tavares Ford: Tavares.Ford@Colorado.EDU

The program I am proposing for my semester project in Artificial Intelligence will respond to textual input from a user with textual output that exhibits emotional reaction. The operation of the program will be by keyword scan; however, it will not be constricted by a specific grammar or set of words. If the discourse by the user goes outside of accepted norms of speech (a loose grammar, if you will) then the program will respond with an admonition that it cannot understatnd the input.

Back to proposals page