OCaml library for semi-Markov conditional random fields Author: Mike Lin [email protected] http://github.com/mlin/CRF

Semi-Markov conditional random fields (SMCRFs) are a type of probabilistic graphical model used to produce segmentations of sequential data, such as gene finding in DNA sequences or named-entity recognition in natural language text. SMCRFs are a discriminative generalization of hidden Markov models and hidden semi-Markov models. This OCaml library provides a generic infrastructure for building SMCRFs, estimating their parameters based on training data, and applying them to segment new examples of interest.

Status: the code is pretty stable, but there is no tutorial documentation yet, which is sorely needed for a complex topic like this. Working on it.

Requires: GNU Scientific Library, ocamlgsl and ocaml+twt