https://github.com/dpressel/arcs-py

Arc-Eager and Arc-Hybrid Greedy Dependency Parser with Dynamic Oracle in Python (with no Dependencies!)
https://github.com/dpressel/arcs-py

nlp nlp-dependency-parsing

Last synced: 4 months ago
JSON representation

Arc-Eager and Arc-Hybrid Greedy Dependency Parser with Dynamic Oracle in Python (with no Dependencies!)

• Host: GitHub
• URL: https://github.com/dpressel/arcs-py
• Owner: dpressel
• Created: 2015-02-21T17:22:17.000Z (almost 10 years ago)
• Default Branch: master
• Last Pushed: 2020-09-29T02:28:51.000Z (over 4 years ago)
• Last Synced: 2024-10-12T04:49:34.750Z (4 months ago)
• Topics: nlp, nlp-dependency-parsing
• Language: Python
• Homepage:
• Size: 785 KB
• Stars: 20
• Watchers: 4
• Forks: 11
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
arcs-py

=======

Python implementation of Arc-Eager and Arc-Hybrid Greedy Dependency Parsing trained with a dynamic oracle described in:

__Goldberg, Yoav, and Joakim Nivre. "Training Deterministic Parsers with Non-Deterministic Oracles." (2013)__

__Goldberg, Yoav, and Joakim Nivre. "A Dynamic Oracle for Arc-Eager Dependency Parsing" (2012)__

The sample data comes from Question Bank and a sample of PTB provided by NLTK in the corpora section, which I converted to a labeled dependency CONLL file using LTH converter and David Vadas' patches.  The sample data also contains non-projective parses, which are ignored by the sample driver program.

If you are looking for a modern flexible dependency parser that supports non-projective parses and is near SoTA, you probably want to use a neural parser architecture like the one in [mead](https://github.com/dpressel/mead-baseline)