def test():
dao_de_jing = text.Text.from_file(os.path.join('..', 'data', 'ddj.txt'))
lunyu = text.Text.from_file(os.path.join('..', 'data', 'lunyu.txt'))
wenyan_corpus = corpus.Corpus(text_model=model.BinaryModel())
wenyan_corpus.add_text(lunyu)
wenyan_corpus.add_text(dao_de_jing)
#print(wenyan_corpus.vector_space)
clustering = wenyan_corpus.clustering(n_clusters=2)
display.render_html(wenyan_corpus, clustering, 'output.html')
def setUp(self):
self.corpus = corpus.Corpus(min_occurrences=2)
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))
def setUp(self):
self.corpus = corpus.Corpus()
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))
self.corpus.add_text(text.Text(DDJ_END, 'ddj_end'))
def setUp(self):
self.corpus = corpus.Corpus()
def setUp(self):
self.corpus = corpus.Corpus(text_model=model.BagOfWordsModel())
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))