Implements a log-linear classification model based on relational features from Bayesian network learning. Under Construction.
Input:
- A relational database
- A target instance (e.g. genre(BraveHeart) in IMDB)
- 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.