[Corpora-List] CfP: Resources and Evaluation for Identity Matching, Entity Resolution and Entity Management

[Miller, Keith J.]

Call for Papers 

Workshop on Resources and Evaluation for Identity Matching, Entity
Resolution and Entity Management

LREC 2008 Workshop (http://www.lrec-conf.org/lrec2008/)
31 May 2008


Structured repositories of data about people are being created through
information extraction from unstructured text as well as from sources
that may themselves be structured documents such as passports or
customer transactions. Problems arise in managing these structured
repositories and integrating information from diverse sources. For
example, newly added information must be consistent with existing
information, must avoid duplication, and must be associated with an
existing entity when that is appropriate. Researchers have addressed
these problems in different contexts with goals such as name or record
matching, identity resolution, and entity disambiguation. In this
workshop, researchers with different perspectives will focus on the
development of resources, algorithms and evaluation methodologies to
improve the technology for managing structured repositories of identity
data.

Evaluation measures for tasks that integrate person information are
especially challenging. Whereas it is generally accepted that entity
extraction systems can be evaluated using MUC scoring metrics, the case
is less clear for "follow-on" technologies. Even a seemingly simple
task such as matching person names in a database context is deceptively
complex, and although measures like precision and recall have been used
to evaluate name and record matching, there are methodological issues
to resolve before we can refer to a "standard" evaluation methodology
for this task. Moreover, it is much less clear how to effectively
evaluate identity matching, resolution, and management systems, or even
what it means to perform an effective identity match, particularly in
the context of data containing identity attributes of varying quality
and in which we have varying degrees of confidence.


We solicit papers that address the following areas:

1. Position papers which:

* Discuss metrics for evaluation of the above-mentioned technologies
* Discuss resources that can be brought to bear on these tasks
* Descriptions of the use cases for these technologies and/or
discussion of which evaluation methods are appropriate for different
use cases
* Discuss approaches for dealing with uncertainty and assigning
confidence values, particularly with respect to manual ground truth
annotation, system output and evaluation metrics

2. Papers discussing state-of-the-art systems and resources for
performing or evaluating name and record matching, identity resolution,
and identity management. Suggested topics include, but are not limited
to:

* Research in name / record matching
* Research in entity disambiguation / resolution / deconfliction
* Descriptions of systems or best practices for entity management
* Evaluation of record matching, entity disambiguation, or entity
management in different use cases
* Resources for identity matching, identity resolution, or identity
management which can contribute to improvements in system performance
* Resources for evaluation of these technologies

Although we are distributing a call for papers to be presented, our
vision is to organize an interactive and dynamic event rather than a
"mini-conference." Presentations will serve to introduce topics for
discussion and to shape the day. However, a significant portion of time
will be devoted to interactive exercises, brainstorming, and other
"work" activities. The workshop will be organized such that all
attendees (including the organizers) come out better informed about the
problems associated with managing structured identity data and with
initial ideas on a methodology for evaluating techniques designed to
solve these problems. In particular, we hope to have made progress
toward an understanding of contextualized and principled evaluation of
identity matching, resolution, and management systems, and to connect a
group of researchers interested in collaborating on this research.

Organizing Committee:

Keith J. Miller (The MITRE Corporation)
Mark Arehart (The MITRE Corporation)
Sherri Condon (The MITRE Corporation)
Jason Duncan (U.S. Department of Defense)
Louise Guthrie (University of Sheffield)
Richard Lutz (The MITRE Corporation)
Massimo Poesio (Universita' di Trento)
 

Important Dates:

Deadline for Paper Submission: Friday, 8 February 2008
Notification to Authors: Friday, 29 February 2008
Submission of Final Version: Friday, 14 March 2008
Workshop at LREC 2008: Saturday, 31 May 2008 

Papers should be submitted in MS Word, PDF, or text (ASCII) format to
Keith J. Miller at keith at mitre.org, with a copy to Mark Arehart at
marehart at mitre.org. Please include the text "LREC Workshop Submission"
in your subject line.

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora