A General Architecture for a Real-Time Discourse Agent
and a Case Study in Oral Reading Tutoring

Schedule

Design and Implementation

Expected Date	Description	Status	Relevant Links
Nov 1 1996	Task model (with persistent credit)	Integrated	(Aist 1997)
Jan 31 1997	Event heirarchy	Written	System documentation: Event heirarchy
Jan 31 1997	Discourse module	Integrated	System documentation: Discourse module
Feb 15 1997	Interface objects	Integrated	System documentation: Interface objects
Jan 31 1997	Turn-taking rules	Integrated	System documentation: Turn-taking rules
Jan 20 - Feb 15 1997	Expanded interventions (e.g. choral reading)	Integrated
Feb 15 1997	Full duplex mode	Integrated

Integration

Expected Date	Description	Status	Relevant Links
Jan 20 1997	Switch to enable choice of old or new architecture	Integrated
Nov 7 1996	Prompt: respond to long periods of inactivity	Integrated
Jan 31 1997	Unstick behavior: backchannel or help on word	Integrated
Feb 5 1997	TakeTurn: give feedback on first uncredited word	Integrated
Feb 5 1997	Usability test: Klaus Zechner	Completed
Feb 10 1997	Usability test: CMU Outreach students	Completed
Feb 10 1997	Usability critique: CMU HCI Seminar Class	Completed
Mar 15 1997	Fix major usability bugs	Completed

Evaluation

Expected Date	Description	Status	Relevant Links
Mar 31 1997	Evaluation, Argumentative: Literature review of human-human and human-computer interaction, especially the role of time-sensitive behavior	Completed, needs revision
Mar 31 1997	Evaluation, Argumentative: Comparison of architecture to proposed rules governing human-human interaction	Conceptualized
Feb 15 1997	Evaluation, Existential: Existence and demonstration of prototype system	Tested
Mar 18-19 1997	Evaluation, Qualitative: Videotape students using tutor at Fort Pitt Elementary	Completed
Apr 15 1997	Evaluation, Qualitative: Videotape expert critiqueing student-tutor interaction	Completed
Mar 18-19 1997	Evaluation, Quantitative -- Descriptive: Collect data from tests with Fort Pitt Elementary students	Completed
Apr 20 1997	Evaluation, Quantitative -- Descriptive: Example of "invisible experiment" on effect of backchanneling	Completed
TBA	Evaluation, Quantitative -- Comparative: Deployment at Fort Pitt	Canceled (1)
TBA	Evaluation, Quantitative -- Comparative: Comparison of data with old tutor to data with new tutor	Canceled

(1) The Reading Tutor is deployed, but no comparative evaluation is planned.

Presentation and Documentation

Expected Date	Description	Status	Relevant Links
Apr 25 1997	User Manual	Outlined
Apr 25 1997	System Documentation	Designed	System Documentation
Apr 25 1997	Final Presentation
Apr 25 1997	System Demonstration
May 5 1997	Final report	Conceptualized

Status: one of {Conceptualized, Postponed, Designed, Written, Tested, Integrated}