Design team members: Justina Chan, Benjamin Gin

Supervisor: Prof. C. MacGregor

Background

Currently in the Internet age, vast amounts of information are available from numerous sources on all topics. However, to access the “right” information when the user desires it is becoming more difficult. The design of information retrieval systems matching the desire of the user is a challenging problem. When users are seeking or retrieving information, they might be looking for information that is conceptually related, and often, this information may not be lexically related. While in other cases, similar pieces of information may be described using different terms. Algorithms recognizing the underlying mappings of concepts are particularly useful. Disciplines such as Soft Computing aims to tackle this challenge. Neural Networks, one of the areas of Soft Computing, not only introduces technical advantages such as efficiency, accuracy and tolerance of errors, etc., as Harold Thimbleby puts it, neurocomputing makes “the computer become humble” [1]. Unlike conventional computing, neural computing system has to learn and adapt to the user rather the other way around. This makes the study of Human-Computer Interaction (HCI) more important in the development process than ever.

Project description

This workshop aims to examine the interrelationships between artificial intelligence and human-computer interaction. More specifically, it explores the area of incorporating human factors with the design of intelligent agent algorithms which provide filtered information related to the electronic calendar appointment entries entered by the user, based on the user’s profile generated by its preferences and activities.

The objectives of this workshop can be broken down into two areas:

1. The design and implementation of the intelligent algorithms:

Build a personal profile of a user based on previous activities through web browser history, email activities, address books and previous appointments.
Creation of an intelligent agent that collects URLs for appointments in the calendar based on the user’s profile.
Design of a functional graphical user interface (GUI) for the Intelligent Calendar to display appropriate amount of information to the users, as well as the controls of the features of the agent (e.g. turning on and off the searching of the URLs).

2. Research questions related to user’s interactions with the calendar and its effects on the algorithm:

Given the data and profile for a specified user, what type of information retrieved by the agent is useful from the user’s perspective?
What should the evaluation and validation procedure be to compare the suggestions generated by the intelligent agent to the user’s “real” preference?
What possible actions or tasks might the user perform that leads to more reliable data on a user’s preference (e.g. if the user deletes email from person A, does that mean that the appointment with person is not important?)?

Design methodology

This workshop adopts the spiral design model, which consists of the stages of analysis, design, implementation and testing. It is expected that at least four iterations will be executed.

First iteration

  • Analyze the problem and design high-level model of the system.

  • Present project proposal to professors and graduate students from the disciplines of HCI and Artificial Intelligence.

Second iteration

  • Analyze and research on area of intelligent agents and implementation tools.

  • Perform task analysis and research on human factors related to artificial intelligence.

  • Propose mock-up design of the user interface and perform design walkthroughs.

Third iteration

  • Evaluate alternatives of intelligent agent algorithm.

  • Perform functional testing on the final algorithms.

  • In parallel, implement standalone modules integrating with the chosen platform to identify potential roadblocks.

  • Refine user interface based on design walkthrough findings.

  • Conduct a study to collect information on how people relate and organize certain information conceptually.

Fourth iteration

  • Analyze and incorporate findings of previous testing in terms of efficiency and accuracy.

  • Implement the rest of the algorithms and features including the user interface.

  • Examine mapping of the intelligent agent system results to the user testing results.

  • Further design refinements and recommendations will be made until the criteria of the project are met.

Reference

[1] Beale, R., Finlay, J. Neural Networks and Pattern Recognition in Human-Computer Interaction. West Sussex, England: Ellis Horwood Limited, 1992.