def test_evaluate():
gp = GraphPattern((
(SOURCE_VAR, wikilink, TARGET_VAR),
(SOURCE_VAR, a, URIRef('http://dbpedia.org/ontology/PopulatedPlace')),
(TARGET_VAR, a, URIRef('http://schema.org/Country'))
))
res = evaluate(
sparql, timeout, gtp_scores, gp)
# (655, 0.4048, 0.4048, 0.0089, 7.5, 3, 3, 2, 0, 0.1936)
# (remains, score, gain, f_measure, avg_reslens, gt_matches,
# patlen, patvars, timeout, qtime)
update_individuals([gp], [res])
fitness = gp.fitness.values
matching_node_pairs = gp.matching_node_pairs
gtp_precisions = gp.gtp_precisions
gp.matching_node_pairs = matching_node_pairs
logger.info(gp.matching_node_pairs)
assert fitness.remains == len(ground_truth_pairs), 'remains wrong?'
assert fitness.gt_matches == 3, "didn't match 3 gt pairs?"
score = fitness.score
assert 0 < score < 0.5, 'score not correct?'
assert score == fitness.gain, 'score and gain should be the same here'
assert 0 < fitness.f_measure < 0.1, \
'f1 measure not correct?'
assert fitness.patlen == 3, 'pattern should have 3 triples'
assert fitness.patvars == 2, 'pattern should have 2 vars'
if not query_time_soft_exceeded(fitness.qtime, timeout):
assert 0 < fitness.avg_reslens < 10, \
'avg match count should be ~7.5'
assert fitness.timeout == 0, 'should not be a timeout'
else:
assert 0 < fitness.avg_reslens < 15, \
'avg match count out of bounds for timeout'
assert fitness.timeout > 0, 'should be a timeout'
assert isinstance(gtp_precisions, OrderedDict)
assert list(gtp_precisions) == matching_node_pairs
logger.info(gtp_precisions)
assert sum(gtp_precisions.values()) == fitness.gain, \
'sum of precisions should be gain in this case'
def test_evaluate():
gp = GraphPattern(
((SOURCE_VAR, wikilink, TARGET_VAR),
(SOURCE_VAR, a, URIRef('http://dbpedia.org/ontology/PopulatedPlace')),
(TARGET_VAR, a, URIRef('http://schema.org/Country'))))
res = evaluate(sparql, timeout, gtp_scores, gp)
# (655, 0.4048, 0.4048, 0.0089, 7.5, 3, 3, 2, 0, 0.1936)
# (remains, score, gain, f_measure, avg_reslens, gt_matches,
# patlen, patvars, timeout, qtime)
update_individuals([gp], [res])
fitness = gp.fitness.values
matching_node_pairs = gp.matching_node_pairs
gtp_precisions = gp.gtp_precisions
gp.matching_node_pairs = matching_node_pairs
logger.info(gp.matching_node_pairs)
assert fitness.remains == len(ground_truth_pairs), 'remains wrong?'
assert fitness.gt_matches == 3, "didn't match 3 gt pairs?"
score = fitness.score
assert 0 < score < 0.5, 'score not correct?'
assert score == fitness.gain, 'score and gain should be the same here'
assert 0 < fitness.f_measure < 0.1, \
'f1 measure not correct?'
assert fitness.patlen == 3, 'pattern should have 3 triples'
assert fitness.patvars == 2, 'pattern should have 2 vars'
if not query_time_soft_exceeded(fitness.qtime, timeout):
assert 0 < fitness.avg_reslens < 10, \
'avg match count should be ~7.5'
assert fitness.timeout == 0, 'should not be a timeout'
else:
assert 0 < fitness.avg_reslens < 15, \
'avg match count out of bounds for timeout'
assert fitness.timeout > 0, 'should be a timeout'
assert isinstance(gtp_precisions, OrderedDict)
assert list(gtp_precisions) == matching_node_pairs
logger.info(gtp_precisions)
assert sum(gtp_precisions.values()) == fitness.gain, \
'sum of precisions should be gain in this case'
def test_timeout_pattern():
u = URIRef('http://dbpedia.org/resource/Template:Reflist')
wpdisambig = URIRef('http://dbpedia.org/ontology/wikiPageDisambiguates')
gp = GraphPattern([
(SOURCE_VAR, Variable('v1'), u),
(SOURCE_VAR, Variable('v5'), u),
(TARGET_VAR, Variable('v0'), u),
(TARGET_VAR, Variable('v3'), u),
(TARGET_VAR, Variable('v6'), Variable('v2')),
(Variable('v4'), wpdisambig, TARGET_VAR),
])
res = evaluate(sparql, timeout, gtp_scores, gp)
update_individuals([gp], [res])
fitness = gp.fitness.values
matching_node_pairs = gp.matching_node_pairs
gp.matching_node_pairs = matching_node_pairs
logger.info(gp.matching_node_pairs)
assert query_time_soft_exceeded(fitness.qtime, timeout)
assert fitness.score == 0
if query_time_hard_exceeded(fitness.qtime, timeout):
assert fitness.f_measure == 0
else:
assert fitness.f_measure > 0
def test_timeout_pattern():
u = URIRef('http://dbpedia.org/resource/Template:Reflist')
wpdisambig = URIRef('http://dbpedia.org/ontology/wikiPageDisambiguates')
gp = GraphPattern([
(SOURCE_VAR, Variable('v1'), u),
(SOURCE_VAR, Variable('v5'), u),
(TARGET_VAR, Variable('v0'), u),
(TARGET_VAR, Variable('v3'), u),
(TARGET_VAR, Variable('v6'), Variable('v2')),
(Variable('v4'), wpdisambig, TARGET_VAR),
])
res = evaluate(
sparql, timeout, gtp_scores, gp)
update_individuals([gp], [res])
fitness = gp.fitness.values
matching_node_pairs = gp.matching_node_pairs
gp.matching_node_pairs = matching_node_pairs
logger.info(gp.matching_node_pairs)
assert query_time_soft_exceeded(fitness.qtime, timeout)
assert fitness.score == 0
if query_time_hard_exceeded(fitness.qtime, timeout):
assert fitness.f_measure == 0
else:
assert fitness.f_measure > 0