ScholarWorks@UMassAmherst (University of Massachusetts Amherst) · 2009 · 198 citations · 21 references
Open access
ComputingComputer ScienceMachine LearningNatural Language ProcessingProgramming LanguagesFirst-order LogicNatural Language UnderstandingRelational Data ModelDeclarative SemanticsFactor GraphsGraphical ModelsJoint SegmentationMarkov Logic NetworksProbabilistic ComputationProbabilistic ProgrammingSegmentation
Discriminatively trained undirected graphical models have shown empirical success, but defining their repeated structure and tied parameters flexibly for complex relational data remains a challenge. The authors propose using an imperative language to express model structure, inference, and learning instead of declarative approaches. They implement imperatively defined factor graphs in FACTORIE, a library that blends declarative factor‑graph semantics with imperative construction and operation, enabling mixed domain knowledge and computational efficiencies. Compared to Markov Logic Networks on joint segmentation and coreference, the approach is 3–15× faster and reduces error by 20–25%, setting a new state of the art.
Discriminatively trained undirected graphical models have had wide empirical success, and there has been increasing interest in toolkits that ease their application to complex relational data. The power in relational models is in their repeated structure and tied parameters; at issue is how to define these structures in a powerful and flexible way. Rather than using a declarative language, such as SQL or first-order logic, we advocate using an imperative language to express various aspects of model structure, inference, and learning. By combining the traditional, declarative, statistical semantics of factor graphs with imperative definitions of their construction and operation, we allow the user to mix declarative and procedural domain knowledge, and also gain significant efficiencies. We have implemented such imperatively defined factor graphs in a system we call FACTORIE, a software library for an object-oriented, strongly-typed, functional language. In experimental comparisons to Markov Logic Networks on joint segmentation and coreference, we find our approach to be 3-15 times faster while reducing error by 20-25%—achieving a new state of the art.
21
Matthew Richardson, Pedro Domingos · Machine Learning · 2006 · 2.7K citations · Full text
Inductive Logic Programming: Theory and methods
Stephen Muggleton, Luc De Raedt · The Journal of Logic Programming · 1994 · 1.6K citations · Full text
Discriminative probabilistic models for relational data
Ben Taskar, Pieter Abbeel, Daphne Koller · 2002 · 637 citations
Probabilistic Horn abduction and Bayesian networks
David Poole · Artificial Intelligence · 1993 · 593 citations