We are making available, our recent paper on Cooperative Information
Gathering that deals with networked information discovery. For
researchers in the IR/Information Discovery community, the paper
serves to highlight how they can benefit by taking a Distributed
Problem Solving view of their domain. It also discusses certain aspects
of Information Gathering that make it a good domain for researchers
in Multi-agent Systems to test their theories.
-- Naghi
**********************************************************************
Cooperative Information Gathering:
A Distributed Problem Solving Approach
Tim Oates, M. V. Nagendra Prasad and Victor R. Lesser
Department of Computer Science
University of Massachusetts, Amherst, MA 01003.
{oates,nagendra,lesser}@cs.umass.edu
UMass Computer Science Technical Report 94--66
With the proliferation of electronically available information, an
additional burden has been placed on the implementors of information
gathering (IG) systems. The set of data that represents the best
response to a query may be the aggregation of data acquired from
distributed, heterogeneous information sources. In such environments,
we distinguish between two approaches to the problem of information
gathering that may be characterized as {\em distributed processing}
and {\em distributed problem solving}(DPS). The former is
characteristic of existing IG systems while the latter is the raison
d'\^{e}tre for Multi-Agent Systems (MAS). We examine features of
problems that point to the need for a DPS approach, and discuss the
benefits of viewing information gathering as distributed problem
solving (which subsumes distributed processing). This approach,
called Cooperative Information Gathering (CIG), involves concurrent
and asynchronous access and composition of associated information
spread across a network of information servers by a group of
intelligent agents. Top level queries drive the creation of partially
elaborated information gathering plans, resulting in the employment of
multiple semi-autonomous, cooperative agents for the purpose of
achieving goals and subgoals within those plans. Finally, we briefly
survey current work on distributed and agent-based approaches to
Information Gathering.
There are two ways to access the technical report:
ftp to ftp.cs.umass.edu and get /pub/lesser/oates-94-66.ps
> ftp ftp.cs.umass.edu
> Name:anonymous
> Password: <give your email address here>
> cd /pub/lesser
> binary
> get oates-94-66.ps
> bye
or through WWW link:
http://dis.cs.umass.edu/research/cig.html