FORM OF THE CLASS
15 motivated students engage in high-level discussions about software
agents.
Each class will focus on a different kind of software agent application or
technique. Demonstrations and videos of recent applications will also be used
to facilitate class discussion.
In each class, two student volunteers will lead class discussion based on two
readings from the literature. All students are expected to volunteer to help
run at least one discussion. These students decide (together with Pattie and
Brad) upon the readings, which are distributed to the class one week ahead.
These two students are also expected to do some additional reading and research
into their theme, so that they have an in-depth understanding of the topic and
can answer questions from other students. They will lead class discussion,
drawing from questions submitted by students via email on Friday.
The other students are expected to read the papers beforehand and to send (via
email) a short list of questions/issues for discussion to the TA. In summary,
weekly, all students are expected to:
-
- Read the papers for the next Monday class meeting
- Send (via email) one or more questions or issues that would be interesting to
discuss to the class TA. These will be forwarded to the two discussion leaders
for that week.
- Participate in class discussions.
In addition to the weekly meetings, students have to do a programming project
in which they build a software agent. Works-in-progress will be shown a month
before the end of the term, with final projects due at the end of the term.
During exam week, students present their project and results to the rest of the
class.
TOPICS FOR SESSIONS:
(these are flexible if students are interested in topics not listed
below):
APPLICATIONS:
-
- Eager assistant type agents (Maxims, Meeting Scheduler, Eager, ...)
- Agent applications for wearables computers (remembrance agent,
communications, forget-me-not)
- commerce applications (bargain finder, fido, kasbah, papers on negotiating
agents, etc.)
- matchmaking agents (Yenta, ATT work, etc.)
- filtering & guiding agents (Letizia, Amalthaea, HOMR, Newt, ...)
- Agents currently already available commercially / business models for
commercial agents (Autonomy, Open Sesame, Bargain finder, Lifestyle finder,
Netangels, ...)
- resource allocation by multiple agents
- Things That Think, software agents in your environment
TECHNIQUES:
-
- learning agents & programming by example
- collaborative filtering
- personification of agents: good or bad idea?
- mobile agents (telescript, TCL, Sodabot,...)
- multi-agent systems (KQML, ...)
- market based models
- knowledge-based agents
ON-LINE SITES OF INTEREST:
- - the MIT Media Lab
Agents Group home page. Most of our papers can be downloaded from this site.
- - links from Microsoft on
Lifelike Computer Characters.
- -
University of Maryland agents page; also has many links to other resources.
- - Microsoft User Interface projects.
- - links to and information about artificial life research.
- -
Artificial Intelligence & Machine Learning Best Resources.
- - Papers and photos from the
CIKM Intelligent Information Agents Workshop (Baltimore, Dec. 1-2, 1995).
- -
a newsletter published monthly by Tim Finin of UMBC about developments in software agents of all
kinds.
- - IBM's
intelligent agents information and pointers to other sites and
companies doing agent-related work.
- -
Experimental Knowledge Systems Lab and Distributed AI Lab at University of
Massachusetts, specializing in AI-based agents.
- - Todd Neller (Stanford) maintains a
list of pointers to agent-related pages.
- - Steve Gant maintains a
bibliography of publications on Intelligent Software Agents.
- -
online archive of the software-agents mailing list (includes subscription
information).
- - Electronic Commerce
LIST OF DEMOS/SOFTWARE TO LOOK AT:
-
- Firefly
- Letizia
- Remembrance Agent
- Amalthaea
- Kasbah
- Autonomy
- Netangels
- Bargain Finder
- Fido
[to be completed]
LIST OF VIDEOS:
- Knowledge navigator video, Apple
- Starfire, Sun
- 1995, HP
- Guides, Apple
[to be completed]
Back to Software Agents Seminar class page.