Skip to content

Latest commit

 

History

History
44 lines (30 loc) · 1.46 KB

README.md

File metadata and controls

44 lines (30 loc) · 1.46 KB
Raw

Words-as-Classifiers (Python)

Requirements

  • scikit-learn
  • pandas
  • pandasql
  • sqlite3
  • numpy

You can import as follows:

from wac import WAC

then you can train and evaluate a model. See example in wac_test.py

By default, the composition method is prod (avg and sum also available)

By default, the WAC classifier is LogisticRegression, but it can use any type in the classifier_spec=(classifier,arguments) as long as the classifier has predict_proba.

Notes

Obviously, wac is a simple application of a classifier for each word and is therefore easy to implement. This code helps with taking care of all the vocabulary, training data, and evaluation. It's not the same code or dataset in the original paper (see below); rather, it's a generalized implemenetation.

Reference

@InProceedings{kennington-schlangen:2015:ACL-IJCNLP,
  author    = {Kennington, Casey  and  Schlangen, David},
  title     = {Simple Learning and Compositional Application of Perceptually Grounded Word Meanings for Incremental Reference Resolution},
  booktitle = {Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)},
  month     = {July},
  year      = {2015},
  address   = {Beijing, China},
  publisher = {Association for Computational Linguistics},
  pages     = {292--301},
  url       = {http://www.aclweb.org/anthology/P15-1029}
}