Learning Statistical Models from Relational Data

AAAI 2000 Workshop
General

AAAI Workshop
Summer 2000

Pointers
Updcoming Events
Papers
Software
Data

Contact Information
Information
Mailing List

 The AAAI 2000 Workshop on Learning Statististical Models from Relational Data was held on July 31, 2000 in Austin, Texas. The Workshop brought together researchers from diverse research areas, including machine learning, inductive logic programming, statistics, and databases. The workshop included nine paper presentations and two invited talks. The workshop closed with a roundtable discussion of potential application domains. Additional details on the schedule are given below. The collected papers from the workshop are available as a AAAI Press technical report. 

The bulk of the research presented at the workshop shared a common motivation: to uncover patterns and make predictions from structured data. However, there are multiple paths toward the common goal of statistical relational learning (SRL). One path begins with machine learning and statistical methods for "flat" or attribute-value representations, and expands these approaches to incorporate relational structure. However, a key assumption of many existing learning techniques -- independent and identically distributed instances -- may no longer hold, so the naive approach of flattening structured data may introduce important statistical errors. A second path extends techniques for relational learning in nonprobabilistic domains, especially inductive logic programming, to incorporate stochastic models. This is an active research area and several new languages and learning algorithms have been proposed. 

There was general consensus that a longer workshop should be held in the near future, allowing more time for discussion and synthesis of the many different approaches and applications. 

Workshop Co-chairs

Lise Getoor Stanford University getoor@cs.stanford.edu
David Jensen University of Massachusetts 
Amherst		 jensen@cs.umass.edu

Workshop Committee

Daphne Koller Stanford University koller@cs.stanford.edu
Heikki Mannila Helsinki University/
Nokia Research 		 Heikki.Mannila@hut.fi
Tom Mitchell Carnegie Mellon University  Tom.Mitchell@cmu.edu
Stephen Muggleton University of York stephen@cs.york.ac.uk

Schedule and Papers

9:00-9:10 Welcome
9:10-9:30 Learning Stochastic Logic Programs
Stephen Muggleton 
9:30-9:50 Interpreting Bayesian Logic Programs
Kristian Kersting, Luc De Raedt, and Stefan Kramer
9:50-10:10 A Bayesian Language for Cumulative Learning
Avi Pfeffer
10:10-10:30 Discussion
10:30-10:45 Coffee Break
10:45-11:30 Invited Talk: Relational Learning: Lessons from the Web
Tom Mitchell and Sean Slattery
11:30-11:50 Efficient Mining of Graph-Based Data
Jesus Gonzalez, Istvan Jonyer, Lawrence B. Holder and Diane J. Cook
11:50-12:10 Iterative Classification in Relational Data
Jennifer Neville and David Jensen
12:10-12:25 Discussion
12:25-1:50 Lunch
1:50-2:35 Invited Talk: Learning Probabilistic Relational Models
Daphne Koller
2:35-2:55 Learning Probablistic Relational Models with Structural Uncertainty
Lise Getoor, Daphne Koller, Benjamin Taskar, and Nir Friedman
2:55-3:15 Rule Induction from Noisy Examples
Laura Firoiu
3:15-3:30 Discussion
3:30-3:45 Coffee Break
3:45-4:05 Learning Mappings between Data Schemas
AnHai Doan, Pedro Domingos, Alon Y. Levy
4:05-4:25 Using Hierarchies, Aggregates and Statistical Models to Discover Knowledge from Distributed Databases
Ronen Pairceir, Sally McClean, and Bryan Scotney
4:25-4:40 Discussion
4:40-5:40 Roundtable
Applications for Learning from Relational Data
Henry Goldberg, Chris Manning, Stephen Muggleton, Ted Senator, Lyle Ungar
5:40-6:00 Wrapup
Version 3.0
Updated 2000.09.26