Welcome to ProbReM!

ProbReM is a framework for modelling relational data using Probabilistic Relational Models (PRMs).

Several frameworks have so far been proposed in Statistical Relational Learning, we hope to contribute to these efforts. ProbReM allows to model relational domains in the form of a directed graphical model which is specified in XML; the data lives in a relational database.

• It is based on the Directed Acyclic Probabilistic Entity Relationship (DAPER) model
• It supports discrete variables
• The parameters of the model can be learned using a Maximum Likelihood (ML) estimate
• Generic aggregation functions (e.g. average, min, max, mode) are used for nodes with multiple parents for one dependency
• Markov Chain Monte Carlo (MCMC) methods are used for approximate inference

ProbReM is being developed by Fabian Kaelin at McGill University and the National Institute of Informatics. This software is free to use for academic research only.

Content

The documentation is organized as follows:

• About ProbReM
• Using ProbReM
• Example model
• Extensions
• Reference

Indices and tables

• Index
• Module Index
• Search Page