def __init__(self, lang='nl', mdir=None):
"""Load the embeddings in the proper language."""
from reach import Spreach
if not mdir:
mdir = os.path.abspath(os.path.dirname(__file__)) + '/datasets/'
if lang == 'nl':
mdir += 'sparse-cow.txt'
self.r = Spreach(mdir)
class WordEmbeddings(object):
"""WordEmbeddings container interacting with reach.
Loads sparse word embeddings from pre-existing word embedding files. These
can be either found under `datasets` in Omesa, or created using for
example gensim, and https://github.com/mfaruqui/sparse-coding.
Parameters
----------
lang : str, optional, default 'nl'
Language, currently only Dutch (nl) is provided in Omesa.
mdir : str, optional, default None
If you want to load a custom embedding file, state path.
Attributes
----------
r : class
Spreach class initialization. More info can be found in README of
https://github.com/stephantul/reach.
"""
def __init__(self, lang='nl', mdir=None):
"""Load the embeddings in the proper language."""
from reach import Spreach
if not mdir:
mdir = os.path.abspath(os.path.dirname(__file__)) + '/datasets/'
if lang == 'nl':
mdir += 'sparse-cow.txt'
self.r = Spreach(mdir)
def transform(self, raw, parse=None):
"""Convert sentences to tokens to vectors."""
# FIXME: reach does an extra split if just a string is provided
return self.r.transform(raw.split(' '))
class WordEmbeddings(object):
"""WordEmbeddings container interacting with reach.
Loads sparse word embeddings from pre-existing word embedding files. These
can be either found under `datasets` in Omesa, or created using for
example gensim, and https://github.com/mfaruqui/sparse-coding.
Parameters
----------
lang : str, optional, default 'nl'
Language, currently only Dutch (nl) is provided in Omesa.
mdir : str, optional, default None
If you want to load a custom embedding file, state path.
Attributes
----------
r : class
Spreach class initialization. More info can be found in README of
https://github.com/stephantul/reach.
"""
def __init__(self, lang='nl', mdir=None):
"""Load the embeddings in the proper language."""
from reach import Spreach
if not mdir:
mdir = os.path.abspath(os.path.dirname(__file__)) + '/datasets/'
if lang == 'nl':
mdir += 'sparse-cow.txt'
self.r = Spreach(mdir)
def transform(self, raw, parse=None):
"""Convert sentences to tokens to vectors."""
# FIXME: reach does an extra split if just a string is provided
return self.r.transform(raw.split(' '))
def __init__(self, lang='nl', mdir=None):
"""Load the embeddings in the proper language."""
from reach import Spreach
if not mdir:
mdir = os.path.abspath(os.path.dirname(__file__)) + '/datasets/'
if lang == 'nl':
mdir += 'sparse-cow.txt'
self.r = Spreach(mdir)
