Brill++

Rule-based part-of-speech tagging with different baseline taggers and different set of rules.

Required Packages:

python 2.7 and NLTK

python

>>> import nltk

>>> nltk.download('punkt')

>>> nltk.download('treebank')

>>> nltk.download('brown')

>>> nltk.download('averaged_perceptron_tagger')

Rule Generation:

python ap-tagging-rule-gen.py <#rules> <corpus = 'brown' or 'treebank'>

The generated rules, which are converted to regular expressions, are witten in "regex.txt" file.

The input to the automata is written in "input.txt".

Rule to MNRL format:

To convert the rules to MNRL format, use hscompile with the following commad:

pcre2mnrl regext.txt regex.mnrl