relational-classification

Implements a log-linear classification model based on relational features from Bayesian network learning. Under Construction.

Input:

1. A relational database
2. A target instance (e.g. genre(BraveHeart) in IMDB)
3. A learned Bayesian network (using FactorBase).

Output: A probability distribution over values of the target instance.

Implements our relational dependency network formula from the paper Fast Learning of Relational Dependency Networks. See also the tutorial.