BEYOND CFG

Software

rparse

rparse is a data-driven parser for Probabilistic Linear Context-Free Rewriting Systems (PLCFRS). It has been developed at the Emmy Noether group of Prof. Dr. Laura Kallmeyer at the University of Tübingen. At this time, it is maintained at the project "Beyond CFG". 

See the rparse homepage for more information.

uparse

uparse is a shift-reduce parser for discontinuous structures, written in Java.

uparse is available on github.

Hierarchical Aligner

The Bottom-up Hierarchical Aligner verifies whether a given word alignment can be generated with a Synchronous Linear Context-Free Rewriting System (SLCFRS) of specified fan-out, including ITG/SCFG. Please refer to the following publication for details:

Miriam Kaeshammer (2013): Synchronous Linear Context-Free Rewriting Systems for Machine Translation.Seventh Workshop on Syntax, Semantics and Structure in Statistical Translation (SSST-7), NAACL-HLT 2013 Workshop. Atlanta, Georgia, USA. [pdf] [bib] [code]

RegAligner

RegAligner is a tool for word alignment, meant as a replacement for GIZA++. It implements the models IBM1-4 and HMM, which can optionally be combined with regularity terms. It is free to use and modify for all kind of research purposes. A recent version can be found in the Git repository.

RegAligner was written by Thomas Schoenemann in roughly equal parts in his free time and at Lund University, Sweden. A few refinements were made at the University of Pisa.

treetools

treetools is a toolbox for manipulating constituency treebank trees, written in Python.

treetools is available on github.