def test_expert_finder_topic_modeling():
alpha = np.ones(k)
beta = np.ones(a)
gamma = np.ones((k, v))
omega = np.ones((k, p))
toy_expert_finder = ef.ExpertFinder(
K=k,
docs_meta=docs,
P=p,
A=a,
V=v,
alpha=alpha,
beta=beta,
gamma=gamma,
omega=omega,
)
ef.expert_finding_learning(toy_expert_finder, 3000)
def test_expert_finder_hierarchy_rocks():
alpha = np.ones(k) #may try different topic distribution
beta = np.ones(a) #authors - uniform
gamma = np.ones((k,v))
back_phrase_prob = 1.0 / p
background_topic_dist = [back_phrase_prob] * p
fpm_topic_dist = generate_phrase_topic_dist(PHRASE_DIST_PATH + 'frequent_pattern_mining', 27451, 0.4)
ds_topic_dist = generate_phrase_topic_dist(PHRASE_DIST_PATH + 'data_stream', 27451, 0.4)
phrase_dist = [background_topic_dist, fpm_topic_dist, ds_topic_dist]
toy_expert_finder = ef.ExpertFinder(
K=k,
docs_meta=docs,
P=p,
A=a,
V=v,
alpha=alpha,
beta=beta,
gamma=gamma,
dist_phrase=phrase_dist,
)
ef.expert_finding_learning(toy_expert_finder, 3000)
def test_expert_finder_hits():
alpha = np.ones(k) #may try different topic distribution
beta = np.ones(a) #authors - uniform
gamma = np.array([[1,1,1,1,1,1,1,1,1,1],
[1,100,50,30,1,1,1,1,1,1],
[1,1,1,1,1,1,100,70,1,1]])
back_phrase_prob = 1.0 / p
background_topic_dist = [back_phrase_prob] * p
fpm_topic_dist = generate_phrase_topic_dist(PHRASE_DIST_PATH + 'frequent_pattern_mining', 27451, 0.4)
ds_topic_dist = generate_phrase_topic_dist(PHRASE_DIST_PATH + 'data_stream', 27451, 0.4)
phrase_dist = [background_topic_dist, fpm_topic_dist, ds_topic_dist]
toy_expert_finder = ef.ExpertFinder(
K=k,
docs_meta=docs,
P=p,
A=a,
V=v,
alpha=alpha,
beta=beta,
gamma=gamma,
dist_phrase=phrase_dist,
)
ef.expert_finding_learning(toy_expert_finder, 200)
toy_hin = build_hin.HIN(docs_meta=docs)
toy_hits_1 = hits.HITS(toy_expert_finder, 1, toy_hin)
#toy_hits_2 = bibrank.HITS(toy_expert_finder, 2, toy_hin)
hits.propagate_with_hits(toy_hits_1, 300)
#bibrank.propagate_with_hits(toy_hits_2, 300)
print "auth papers"
print toy_hits_1.auth_papers
print "hub papers"
print toy_hits_1.hub_papers
print "auth authors"
print toy_hits_1.auth_authors
print "auth venues"
print toy_hits_1.auth_venues