ABSTRACT
We use clustering to derive new relations which augment database schema used in automatic generation of predictive features in statistical relational learning. Entities derived from clusters increase the expressivity of feature spaces by creating new first-class concepts which contribute to the creation of new features. For example, in CiteSeer, papers can be clustered based on words or citations giving "topics", and authors can be clustered based on documents they co-author giving "communities". Such cluster-derived concepts become part of more complex feature expressions. Out of the large number of generated features, those which improve predictive accuracy are kept in the model, as decided by statistical feature selection criteria. We present results demonstrating improved accuracy on two tasks, venue prediction and link prediction, using CiteSeer data.
- M. Craven and S. Slattery. Relational learning with statistical predicate invention: Better models for hypertext. Machine Learning, 43(1/2):97--119, 2001. Google ScholarDigital Library
- Saso Dzeroski and Nada Lavrac. An introduction to inductive logic programming. In Saso Dzeroski and Nada Lavrac, editors, Relational Data Mining, pages 48--73. Springer-Verlag, 2001. Google ScholarDigital Library
- Dean Foster and Lyle Ungar. A proposal for learning by ontological leaps. In Proc. of Snowbird Learning Conference, Snowbird, Utah, 2002.Google Scholar
- L. Getoor, N. Friedman, D. Koller, and A. Pfeffer. Learning probabilistic relational models. In Saso Dzeroski and Nada Lavrac, editors, Relational Data Mining, pages 307--338. Springer-Verlag, 2001.Google ScholarCross Ref
- L. Kaufman and P.J. Rousseeuw. Finding Groups in Data: An Introduction to Cluster Analysis. Wiley-Interscience, 1990.Google Scholar
- Mathias Kirsten, Stefan Wrobel, and Tamas Horvath. Distance based approaches to relational learning and clustering. In Saso Dzeroski and Nada Lavrac, editors, Relational Data Mining, pages 213--230. Springer-Verlag, 2001. Google ScholarDigital Library
- S. Kramer, N. Lavrac, and P. Flach. Propositionalization approaches to relational data mining. In Saso Dzeroski and Nada Lavrac, editors, Relational Data Mining, pages 262--291. Springer-Verlag, 2001. Google ScholarDigital Library
- Stephen Muggleton and Luc De Raedt. Inductive logic programming: Theory and methods. Journal of Logic Programming, 19,20:629--679, 1994.Google ScholarCross Ref
- Dmitry Pavlov, Alexandrin Popescul, David M. Pennock, and Lyle H. Ungar. Mixtures of conditional maximum entropy models. In Proc. of ICML-2003, 2003.Google Scholar
- Claudia Perlich and Foster Provost. Aggregation-based feature invention and relational concept classes. In Proc. of KDD-2003, 2003. Google ScholarDigital Library
- Alexandrin Popescul. Statistical Learning from Relational Databases, PhD thesis. University of Pennsylvania, 2004. Google ScholarDigital Library
- Alexandrin Popescul, Gary Flake, Steve Lawrence, Lyle H. Ungar, and C. Lee Giles. Clustering and identifying temporal trends in document databases. In Proc. of the IEEE Advances in Digital Libraries, 2000. Google ScholarDigital Library
- J.R. Quinlan and R.M. Cameron-Jones. Induction of logic programs: FOIL and related systems. New Generation Computing, 13:287--312, 1995.Google ScholarDigital Library
- G. Salton and M.J. McGill. Introduction to Modern Information Retrieval. McGraw Hill, 1983. Google ScholarDigital Library
- Gideon Schwartz. Estimating the dimension of a model. The Annals of Statistics, 6(2):461--464, 1979.Google ScholarCross Ref
- E. Shapiro. Algorithmic Program Debugging. MIT Press, 1983. Google ScholarDigital Library
Index Terms
- Cluster-based concept invention for statistical relational learning
Recommendations
Relational visual cluster validity (RVCV)
The assessment of cluster validity plays a very important role in cluster analysis. Most commonly used cluster validity methods are based on statistical hypothesis testing or finding the best clustering scheme by computing a number of different cluster ...
Aggregation-based feature invention and relational concept classes
KDD '03: Proceedings of the ninth ACM SIGKDD international conference on Knowledge discovery and data miningModel induction from relational data requires aggregation of the values of attributes of related entities. This paper makes three contributions to the study of relational learning. (1) It presents a hierarchy of relational concepts of increasing ...
Some connectivity based cluster validity indices
Identification of the correct number of clusters and the appropriate partitioning technique are some important considerations in clustering where several cluster validity indices, primarily utilizing the Euclidean distance, have been used in the ...
Comments