English

Parsing Universal Dependencies without training

Computation and Language 2017-01-13 v1

Abstract

We propose UDP, the first training-free parser for Universal Dependencies (UD). Our algorithm is based on PageRank and a small set of head attachment rules. It features two-step decoding to guarantee that function words are attached as leaf nodes. The parser requires no training, and it is competitive with a delexicalized transfer system. UDP offers a linguistically sound unsupervised alternative to cross-lingual parsing for UD, which can be used as a baseline for such systems. The parser has very few parameters and is distinctly robust to domain change across languages.

Keywords

Cite

@article{arxiv.1701.03163,
  title  = {Parsing Universal Dependencies without training},
  author = {Héctor Martínez Alonso and Željko Agić and Barbara Plank and Anders Søgaard},
  journal= {arXiv preprint arXiv:1701.03163},
  year   = {2017}
}

Comments

EACL 2017, 8+2 pages

R2 v1 2026-06-22T17:47:54.205Z