Looking for at least three references for research that addresses the
following issue:

Given that there were extensive efforts during the 1990s toward
developing software agents (aka "intelligent agents", "bots",
"personal assistants", etc.), have these efforts declined?  If so, are
there indications of why, when, how various projects encountered
obstacles?  In contrast, were there major successes other than search
engines?

Each reference must be either a PhD thesis at a major university, or
an article in a computer industry journal written by a primary
researcher in the field.

A general category for this kind of work would be: 
://www.google.com/Top/Computers/Artificial_Life/Agents/

NB: our findings so far tend to show that the rise of "Web 2.0"
companies during the past few years has provided more general purpose
tools - mostly browser-based - which obviate the need for personal
agents.  For example, many of the implicit personalization services
provided by search engines, Amazon, eBay, del.icio.us, Technorati,
etc., have displaced or subsumed potential features which could have
been provided by explicit and distinct software agents.

For example, there is an interesting collection of links to agent
research at the UMBC AgentWeb http://agents.umbc.edu/ - however at
this point many of the links are either broken or their activity
appears to have ceased since 2000.

Another example is the collection at BotSpot http://www.botspot.com/ -
however most of the actual examples are for desktop activities
(text-to-speech, translation) but do not provide much "intelligence".

One wonders, "where are the success stories for agents?"

---

Clarification of Question by ceteri-ga on 21 Jun 2006 10:38 PDT

We are looking for references which can be verified online.  We want
to avoid having to purchase articles prior to marking this question as
"answered".

---

Request for Question Clarification by pafalafa-ga on 21 Jun 2006 11:39 PDT

ceteri-ga,

This is an interesting question, but I have to say, the concept of an
'agent' is so fluid that it's difficult to pin-down anything regarding
trends.

Please have a look at this link:


http://scholar.google.com/scholar?as_q=&num=100&btnG=Search+Scholar&as_epq=agent+based&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=2004&as_yhi=2007&as_allsubj=all&hl=en&lr=&safe=active


It would help to focus my research if you could pinpoint any articles
at the above link that are particularly useful (or particularly
useless!) for your needs.

Thanks for whatever feedback you can provide.

pafalafa-ga

---

Clarification of Question by ceteri-ga on 22 Jun 2006 22:45 PDT

True, the question certainly needs more constraints. That Google
Scholar search produces almost 10K hits.

I'll need to work through a few different Google Scholar search
results to make notes about what to avoid, what to explore in depth.

Some immediate issues:

   * "agent-based simulation" generally refers to computer simulation
of representations of human actions, as in economic simulation
experiments.  These should be filtered out.

   * searches on "intelligent agent" tend to include quite a large
number of hits related to "Intelligent Design", which is off-topic.

   * "agent oriented methodology" - probably related to the simulation
work above, and to be disregarded.


Ideally we'd prefer to focus on criticisms of R&D related to:

   * commercial implementations of software agents
   * agents that present some aspects of virtual embodiment (humanoid,
etc.), some conversation, potentially some affect as well
   * agents that are not simply a Google Alert type of email notification

To try to be current, let's also constrain to papers or articles
written within the past 3 years.


There is a reasonable good Wikipedia topics about distinguishing
agents from programs or objects (though much of the rest of that
article seems rather vague):
   http://en.wikipedia.org/wiki/Software_agent#What_is_not_an_agent_...

And then this paper embellishes on a similar topic, proposing "A
Taxonomy for Autonomous Agents":
   http://www.msci.memphis.edu/~franklin/AgentProg.html


Here's an example of what might be considered as a commercially
successful software agent (though not embodied or affective):
   http://www.blackpearl.com/solutions/solutions_overview.html


Here's an example of a paper close to the sources that we need (albeit
from 1999, so too far out of date):
   http://scholar.google.com/url?sa=U&q=http://www.csrc.nist.gov/staff/jansen/IEEEaftm.pdf


Working along the lines of a Google Scholar query, these searches
bring the result size down to 100-500 instead of 10K, though the
results are not all sufficient for our constraints:

http://scholar.google.com/scholar?as_q=criticism+commercial+success&num=100&btnG=Search+Scholar&as_epq=intelligent+agent&as_oq=&as_eq=intelligent-design&as_occt=any&as_sauthors=&as_publication=&as_ylo=2004&as_yhi=2007&as_allsubj=some&as_subj=eng&as_subj=soc&hl=en&lr=&safe=off
http://scholar.google.com/scholar?num=100&hl=en&lr=&safe=off&q=criticism+commercial+success+%22autonomous+agent%22+-intelligent-design&as_ylo=2004&as_yhi=2007&btnG=Search&as_allsubj=some&as_subj=eng&as_subj=soc
http://scholar.google.com/scholar?as_q=criticism+commercial+success+&num=100&btnG=Search+Scholar&as_epq=software+agent&as_oq=&as_eq=intelligent-design+agent-based&as_occt=any&as_sauthors=&as_publication=&as_ylo=2004&as_yhi=2007&as_allsubj=some&as_subj=eng&as_subj=soc&hl=en&lr=&safe=off