yap

yap - Yet Another Parser

yap is yet another parser written in Go. It was implemented to test the hypothesis of my MSc thesis on Joint Morpho-Syntactic Processing of MRLs in a Transition Based Framework at IDC Herzliya with my advisor, Reut Tsarfaty. A paper on the morphological analysis and disambiguation aspect for Modern Hebrew and Universal Dependencies was accepted to COLING 2016

yap is currently provided with a model for Modern Hebrew, trained on a heavily updated version of the SPMRL 2014 Hebrew treebank. We hope to publish the updated treebank soon as well.

yap contains an implementation of the framework and parser of zpar from Z&N 2011 (Transition-based Dependency Parsing with Rich Non-local Features by Zhang and Nivre, 2011) with flags for precise output parity (i.e. bug replication), trained on the morphologically disambiguated Modern Hebrew treebank.

yap is under active development and documentation.

DO NOT USE FOR PRODUCTION

Requirements

• http://www.golang.org
• bzip2
• 4-16 CPU cores
• ~4.5GB RAM for Morphological Disambiguation
• ~2GB RAM for Dependency Parsing

Compilation

• Download and install Go
• Setup a Go environment:
  • Create a directory (usually per workspace/project) mkdir yapproj; cd yapproj
  • Set $GOPATH environment variable to your workspace: export GOPATH=path/to/yapproj
  • In the workspace directory create 3 subdirectories: mkdir src pkg bin
  • cd into the src directory cd src
• Clone the repository in the src folder of the workspace, then:

go get .
go build .
./yap

• Bunzip the Hebrew MD model: bunzip2 data/hebmd.b32.bz2
• Bunzip the Hebrew Dependency Parsing model: bunzip2 data/dep.b64.bz2

You may want to use a go workspace manager or have a shell script to set $GOPATH to <.../yapproj>

Processing Modern Hebrew

Currently only Pipeline Morphological Analysis, Disambiguation, and Dependency Parsing of pre-tokenized Hebrew text is supported. For Hebrew Morphological Analysis, the input format should have tokens separated by a newline, with another newline to separate sentences.

The lattice format as output by the analyzer can be used as-is for disambiguation.

For example:

עשרות
אנשים
מגיעים
מתאילנד
...

כך
אמר
ח"כ
...

Commands for morphological analysis and disambiguation:

./yap hebma -raw input.raw -out lattices.conll
./yap md -in lattices.conll -om output.conll

The output of the morphological disambiguator can be used as input for the dependency parser. Command for dependency parsing:

./yap dep -inl output.conll -oc dep_output.conll

Citation

If you make use of this software for research, we would appreciate the following citation:

@InProceedings{moretsarfatycoling2016,
  author = {Amir More and Reut Tsarfaty},
  title = {Data-Driven Morphological Analysis and Disambiguation for Morphologically Rich Languages and Universal Dependencies},
  booktitle = {Proceedings of COLING 2016},
  year = {2016},
  month = {december},
  location = {Osaka}
}

HEBLEX, a Morphological Analyzer for Modern Hebrew in yap, relies on a slightly modified version of the BGU Lexicon. Please acknowledge and cite the work on the BGU Lexicon with this citation:

@inproceedings{adler06,
    Author = {Adler, Meni and Elhadad, Michael},
    Booktitle = {ACL},
    Crossref = {conf/acl/2006},
    Editor = {Calzolari, Nicoletta and Cardie, Claire and Isabelle, Pierre},
    Ee = {http://aclweb.org/anthology/P06-1084},
    Interhash = {6e302df82f4d7776cc487d5b8623d3db},
    Intrahash = {c7ac3ecfe40d039cd6c9ec855cb432db},
    Keywords = {dblp},
    Publisher = {The Association for Computer Linguistics},
    Timestamp = {2013-08-13T15:11:00.000+0200},
    Title = {An Unsupervised Morpheme-Based HMM for {H}ebrew Morphological
        Disambiguation},
    Url = {http://dblp.uni-trier.de/db/conf/acl/acl2006.html#AdlerE06},
    Year = 2006,
    Bdsk-Url-1 = {http://dblp.uni-trier.de/db/conf/acl/acl2006.html#AdlerE06}}

License

This software is released under the terms of the [https://www.apache.org/licenses/LICENSE-2.0](Apache License, Version 2.0).

The Apache license does not apply to the BGU Lexicon. Please contact Reut Tsarfaty regarding licensing of the lexicon.

Contact

You may contact me at mygithubuser at gmail or Reut Tsarfaty at reutts at openu dot ac dot il