

In [1]:

    
%pylab inline









    



Populating the interactive namespace from numpy and matplotlib

1.8 CBOW/Skip-gram model

Literature: here and here.



In [1]:

    
import nltk
from tethne.readers import zotero
import matplotlib.pyplot as plt
from helpers import normalize_token, filter_token
import gensim
from itertools import groupby, islice



In [2]:

    
text_root = '../data/EmbryoProjectTexts/files'
zotero_export_path = '../data/EmbryoProjectTexts'

corpus = nltk.corpus.PlaintextCorpusReader(text_root, 'https.+')
metadata = zotero.read(zotero_export_path, index_by='link', follow_links=False)



In [3]:

    
years = []
documents = []
for fileid in corpus.fileids():
    for sentence in corpus.sents(fileids=[fileid]):
        years.append(metadata[fileid].date)
        documents.append([normalize_token(token) 
                          for token in sentence 
                          if filter_token(token)])



In [4]:

    
len(documents)









    Out[4]:





51943



In [5]:

    
model = gensim.models.Word2Vec(documents, size=200, window=5)

Similary is measured using cosine distance in the low-dimensional embedding/vector space.



In [31]:

    
model.most_similar(positive=['cell'])









    Out[31]:





[(u'nucleus', 0.6756647825241089),
 (u'derived', 0.6577451825141907),
 (u'adult', 0.6551524996757507),
 (u'differentiate', 0.5955120921134949),
 (u'embryonic', 0.5923236012458801),
 (u'interaction', 0.5920150279998779),
 (u'line', 0.5903056263923645),
 (u'differentiated', 0.5851104259490967),
 (u'tissue', 0.5801264643669128),
 (u'registry', 0.5774797201156616)]



In [39]:

    
model.similarity('egg', 'ovum'), model.similarity('egg', 'embryo'), model.similarity('egg', 'brain')









    Out[39]:





(0.68291990417474357, 0.33644428780437841, 0.091136195140101212)



In [61]:

    
model.syn0.shape     # Words x Dimensions.









    Out[61]:





(9769, 200)



In [63]:

    
len(model.index2word)









    Out[63]:





9769

PCA on lower-dimensional projection



In [7]:

    
from sklearn import decomposition



In [52]:

    
focal_words = ['egg', 'ontogeny', 'ovum', 'fetus', 'ivf', 'uterus', 'brain', 
               'neuron', 'cell', 'transplantation', 'culture', 'stem', 'nucleus']



In [53]:

    
focal_indices = [model.index2word.index(w) for w in focal_words]



In [54]:

    
pca = decomposition.PCA(n_components=2)



In [55]:

    
t = pca.fit_transform(model.syn0[focal_indices, :])



In [56]:

    
plt.scatter(t[:, 0], t[:, 1])
for i in xrange(len(focal_indices)):
    w = focal_words[i]
    plt.text(t[i, 0], t[i, 1], w)



In [ ]: