def test_build_lsh_forest(self):
# Create a CompanyCorpus instance, and initialize it with some data
test_corpus = [['company_1', 'Provider of software'], ['company_2', 'Provider of hardware']]
test_input = pd.DataFrame(test_corpus, columns=['domain', 'description'])
cc = CompanyCorpus(test_input)
number_to_remove = 1
cc.build_idf(description_column_name='description')
cc.filter_desc_by_idf(description_column_name='description',
number_words_to_cut=number_to_remove)
# Create a CompanyGraph instance and test building the LSH forest
cg = CompanyGraph(cc)
cg.build_lsh_forest(company_name_column_name='domain')
self.assertTrue(cg.lsh_forest)
def test_save_graph(self):
# Create a CompanyCorpus instance, and initialize it with some data
test_corpus = [['company_1', 'Provider of software'], ['company_2', 'Provider of hardware technology'],
['company_3', 'Provider of Software technology'], ['company_4', 'Provider of software service']]
test_input = pd.DataFrame(test_corpus, columns=['domain', 'description'])
cc = CompanyCorpus(test_input)
number_to_remove = 1
cc.build_idf(description_column_name='description')
cc.filter_desc_by_idf(description_column_name='description',
number_words_to_cut=number_to_remove)
# Create a CompanyGraph instance
cg = CompanyGraph(cc)
cg.build_lsh_forest(company_name_column_name='domain')
cg.build_graph(sensitivity=3)
save_graph(cg, filename='graph.pickle')
def test_get_jaccard_similarity(self):
# Create a CompanyCorpus instance, and initialize it with some data
test_corpus = [['company_1', 'Provider of software'], ['company_2', 'Provider of hardware technology'],
['company_3', 'Provider of Software technology'], ['company_4', 'Provider of software service']]
test_input = pd.DataFrame(test_corpus, columns=['domain', 'description'])
cc = CompanyCorpus(test_input)
number_to_remove = 1
cc.build_idf(description_column_name='description')
cc.filter_desc_by_idf(description_column_name='description',
number_words_to_cut=number_to_remove)
# Create a CompanyGraph instance
cg = CompanyGraph(cc)
cg.build_lsh_forest(company_name_column_name='domain')
cg.build_graph(sensitivity=3)
# print(cg.graph.todense())
# Run test of function
jaccard_similarity = cg.get_jaccard_similarity('company_1', 'company_3')
self.assertNotEqual(0., jaccard_similarity)
def test_get_dot_product_score(self):
# Create a CompanyCorpus instance, and initialize it with some data
test_corpus = [['company_1', 'business software application'], ['company_2', 'hardware technology'],
['company_3', 'consumer software service'], ['company_4', 'consumer saas application']]
test_input = pd.DataFrame(test_corpus, columns=['domain', 'description'])
cc = CompanyCorpus(test_input)
number_to_remove = 0
cc.build_idf(description_column_name='description')
cc.filter_desc_by_idf(description_column_name='description',
number_words_to_cut=number_to_remove)
# Create a CompanyGraph instance
cg = CompanyGraph(cc)
cg.build_lsh_forest(company_name_column_name='domain')
cg.build_graph(sensitivity=3)
# Run test of function
dot_product_score = cg.get_dot_product_score('company_1', 'company_3')
self.assertNotEqual(0., dot_product_score)