Unknown symbol in nltk pos tagging for Arabic

Question

I have used nltk to tokenize some arabic text

However, i ended up with some results like

(u'an arabic character/word', '``') or (u'an arabic character/word', ':')

However, they do not provide the `` or : in the documentation.

hence i would like to find out what is this

from nltk.toeknize.punkt import PunktWordTokenizer 

z = "أنا تسلق شجرة"
tkn = PunkWordTokenizer
sen = tkn.tokenize(z)
tokens = nltk.pos_tag(sent)

print tokens

Answer 1

The default NLTK POS tag is trained on English texts and is supposedly for English text processing, see http://www.nltk.org/_modules/nltk/tag.html. The docs:

An off-the-shelf tagger is available.  It uses the Penn Treebank tagset:

    >>> from nltk.tag import pos_tag  # doctest: +SKIP
    >>> from nltk.tokenize import word_tokenize # doctest: +SKIP
    >>> pos_tag(word_tokenize("John's big idea isn't all that bad.")) # doctest: +SKIP
    [('John', 'NNP'), ("'s", 'POS'), ('big', 'JJ'), ('idea', 'NN'), ('is',
    'VBZ'), ("n't", 'RB'), ('all', 'DT'), ('that', 'DT'), ('bad', 'JJ'),
    ('.', '.')]

And the code for pos_tag:

from nltk.data import load


# Standard treebank POS tagger
_POS_TAGGER = 'taggers/maxent_treebank_pos_tagger/english.pickle'
def pos_tag(tokens):
    """
    Use NLTK's currently recommended part of speech tagger to
    tag the given list of tokens.

        >>> from nltk.tag import pos_tag # doctest: +SKIP
        >>> from nltk.tokenize import word_tokenize # doctest: +SKIP
        >>> pos_tag(word_tokenize("John's big idea isn't all that bad.")) # doctest: +SKIP
        [('John', 'NNP'), ("'s", 'POS'), ('big', 'JJ'), ('idea', 'NN'), ('is',
        'VBZ'), ("n't", 'RB'), ('all', 'DT'), ('that', 'DT'), ('bad', 'JJ'),
        ('.', '.')]

    :param tokens: Sequence of tokens to be tagged
    :type tokens: list(str)
    :return: The tagged tokens
    :rtype: list(tuple(str, str))
    """
    tagger = load(_POS_TAGGER)
    return tagger.tag(tokens)

This works for me to get Stanford tools working in python on Ubuntu 14.4.1:

$ cd ~
$ wget http://nlp.stanford.edu/software/stanford-postagger-full-2015-01-29.zip
$ unzip stanford-postagger-full-2015-01-29.zip
$ wget http://nlp.stanford.edu/software/stanford-segmenter-2015-01-29.zip
$ unzip /stanford-segmenter-2015-01-29.zip
$ python

and then:

from nltk.tag.stanford import POSTagger
path_to_model= '/home/alvas/stanford-postagger-full-2015-01-30/models/arabic.tagger'
path_to_jar = '/home/alvas/stanford-postagger-full-2015-01-30/stanford-postagger-3.5.1.jar'

artagger = POSTagger(path_to_model, path_to_jar, encoding='utf8')
artagger._SEPARATOR = '/'
tagged_sent = artagger.tag(u"أنا تسلق شجرة")
print(tagged_sent)

[out]:

$ python3 test.py
[('أ', 'NN'), ('ن', 'NN'), ('ا', 'NN'), ('ت', 'NN'), ('س', 'RP'), ('ل', 'IN'), ('ق', 'NN'), ('ش', 'NN'), ('ج', 'NN'), ('ر', 'NN'), ('ة', 'PRP')]

If you have java problems when using Stanford POS tagger, see DELPH-IN wiki: http://moin.delph-in.net/ZhongPreprocessing