Multilingual L-PCFG Models

Description

The Rainbow Parser is a phrase-structure syntactic parser. At its core, there is the use of a latent-variable PCFG model. Its training procedure is based on spectral methods of learning.

The parser is now available on github.

(December 2017) All models for all SPMRL languages and German NEGRA corpus are available here.

Paper

Click here for the following paper.


@inproceedings{narayan-16b,
  title={Optimizing Spectral Learning for Parsing},
  author={Shashi Narayan and Shay B. Cohen},
  booktitle={Proceedings of {ACL}},
  year={2016}
}

Some notes

Below we include the table of results on the test sets from the SPMRL shared task to parse morphologically rich languages. For a legend, see the paper (Tables 2 and 3).


language CL van. CL opt. SP van. SP opt. Berkeley
alt text Basque 79.6 81.4 79.9 80.5 74.7
alt text French 74.3 75.6 78.7 79.1 80.4
alt text German (NEGRA) 76.4 78.0 78.4 79.4 80.1
alt text German (TiGeR) 74.1 76.0 78.0 78.2 78.3
alt text Hebrew 86.3 87.2 87.8 89.0 87.0
alt text Hungarian 86.5 88.4 89.1 89.2 85.2
alt text Korean 76.5 78.4 80.3 80.0 78.6
alt text Polish 90.5 91.2 91.8 91.8 86.8
alt text Swedish 76.4 79.4 78.4 80.9 80.6