def comparePolicies(query):
results = {}
print('Working on {}'.format(query))
arg = getArg(query)
entities = crosswikis.getEntityDistribution(arg)
entities = entities[:10]
topEntities1 = getTopEntities(entities, minCprob=0, limit=3)
topEntities2 = getTopEntities(entities, minCprob=0.2, limit=None)
topEntities3 = getTopEntities(entities, minCprob=0.15, limit=5)
results = (topEntities1, topEntities2, topEntities3)
return results
def crosswikis():
"""Crosswikis page. Performs a search if given the arg 'arg', otherwise, just
shows a search box."""
if 'arg' in request.args:
arg = request.args['arg']
entities = cw.getEntityDistribution(app.config['CROSSWIKIS_DB_PATH'], arg)
return render_template(
'crosswikis.html',
title='Crosswikis lookup',
scripts=['jquery-1.9.0.min.js', 'crosswikis.js'],
entities=entities
)
else:
return render_template(
'crosswikis.html',
title='Crosswikis lookup',
scripts=['jqeury-1.9.0.min.js'],
entities=None
)
def linkStringToEntity(string,
cprobThreshold=0.9,
countThreshold=1000,
tupleThreshold=500):
"""Gets the entity most likely to be referred to by the given string.
Args: string: The string to get the entity link for.
cprobThreshold: The minimum probability of the entity given the string we
want.
countThreshold: The minimum count of the entity we want.
Returns: The most likely entity given the string, or None if no entity was
found with high enough threshold.
"""
entityDistribution = cw.getEntityDistribution(string,
table='crosswikis_subset')
entityDistribution = [(ent, num, denom, cprob)
for (ent, num, denom, cprob) in entityDistribution
if cprob > cprobThreshold and num > countThreshold]
return entityDistribution[0] if len(entityDistribution) > 0 else None
def linkStringToEntity(string, cprobThreshold=0.9, countThreshold=1000,
tupleThreshold=500):
"""Gets the entity most likely to be referred to by the given string.
Args: string: The string to get the entity link for.
cprobThreshold: The minimum probability of the entity given the string we
want.
countThreshold: The minimum count of the entity we want.
Returns: The most likely entity given the string, or None if no entity was
found with high enough threshold.
"""
entityDistribution = cw.getEntityDistribution(
string,
table='crosswikis_subset'
)
entityDistribution = [
(ent, num, denom, cprob) for (ent, num, denom, cprob) in entityDistribution
if cprob > cprobThreshold and num > countThreshold
]
return entityDistribution[0] if len(entityDistribution) > 0 else None
