def test_profile_evolution(self):
"""Test that a profile evolves with a normal context"""
categories = list(DmozCategory.get_for_query(query="study"))
other_cats = list(DmozCategory.get_for_query(query="Einstein"))
docs = list(DocumentSurrogate.objects.filter(category__in=categories).values_list('pk', flat=True))
#first pass: both are added
r = update_profile.apply(args=[self.profile, 'Einstein', docs], kwargs={'lang': 'en', 'terms': True})
p = self.profile.preferences.all()
before_boost = {}
[before_boost.update({e.category.pk: e.score}) for e in p if e.category in categories]
before_decay = {}
[before_decay.update({e.category.pk: e.score}) for e in p if e.category in other_cats]
#second pass: only the docs:
r = update_profile.apply(args=[self.profile, [], docs], kwargs={'lang': 'en', 'terms': True})
q = self.profile.preferences.all()
after_boost = {}
[after_boost.update({e.category.pk: e.score}) for e in q if e.category in categories]
after_decay = {}
[after_decay.update({e.category.pk: e.score}) for e in q if e.category in other_cats]
#the docs must've been boosted, the others, decayed:
#REMEMBER: if a preferences gets to 1, it can't go farther up!! (100% is the maximum score)
self.assertFalse([k for k,v in after_boost.items() if after_boost[k] < before_boost[k]]
+[k for k,v in after_decay.items() if after_decay[k] >= before_decay[k]])
def test_as_dict(self):
"""Test the dictionary option"""
results = DmozCategory.get_for_query(self.query)
dict_results = DmozCategory.get_for_query(self.query, as_dict=True)
result = dict_results.keys().sort()
expected = [e.pk for e in results].sort()
self.assertEqual(result, expected)
def test_profile_creation_query(self):
"""Test that a profile is created with a direct query and documents"""
categories = DmozCategory.get_for_query(query="study")
other_cats = DmozCategory.get_for_query(query="Einstein")
docs = list(DocumentSurrogate.objects.filter(category__in=categories).values_list('pk', flat=True))
r = update_profile.apply(args=[self.profile, 'Einstein', docs], kwargs={'lang': 'en', 'terms': True})
self.assert_(self._check_expected_profile(self.profile, itertools.chain(categories, other_cats)))
def test_profile_expansion(self):
"""Test that a profile behaves properly with new preferences"""
categories = DmozCategory.get_for_query(query="study")
other_cats = DmozCategory.get_for_query(query="Einstein")
docs = list(DocumentSurrogate.objects.filter(category__in=categories).values_list('pk', flat=True))
#first pass: both are added
r = update_profile.apply(args=[self.profile, 'Einstein'], kwargs={'lang': 'en', 'terms': True})
before = self.profile.preferences.count()
r = update_profile.apply(args=[self.profile, [], docs], kwargs={'lang': 'en', 'terms': True})
after = self.profile.preferences.count()
self.assert_(after > before)
def test_query(self):
"""Test that a query returns an iterable of categories"""
results = DmozCategory.get_for_query(self.query)
self.assert_((len(results)
and hasattr(results, '__iter__')
and isinstance(results[0], DmozCategory)))
def test_category_list(self):
"""Test that when given ambiguous queries the classifier returns a list of candidates"""
results = sorted([e.pk for e in DmozCategory.get_for_query(self.query)])
expected = DmozCategory.objects.filter(
title__in=self.query_candidates).order_by('pk').values_list('pk', flat=True)
self.assertEqual(results, list(expected))
def test_classify_terms_spanish(self):
"""Test that a profile is created with a direct query in spanish"""
categories = DmozCategory.get_for_query(query="Einstein", lang='es')
r = update_profile.apply(args=[self.profile, 'Einstein', []], kwargs={'lang': 'es', 'terms': True})
self.assert_(self._check_expected_profile(self.profile, categories))
def test_build_query_english(self):
"""Test that a profile is created given an english context"""
categories = DmozCategory.get_for_query(query="Einstein", lang='en')
ctx = """Einstein himself is well known for rejecting some of the claims of quantum mechanics."""
r = update_profile.apply(args=[self.profile, ctx, []], kwargs={'lang': 'en', 'terms': False})
self.assert_(self._check_expected_profile(self.profile, categories))
def test_classify_only_docs(self):
"""Test that a profile is created with only documents"""
categories = DmozCategory.get_for_query(query="study")
docs = list(DocumentSurrogate.objects.filter(category__in=categories).values_list('pk', flat=True))
r = update_profile.apply(args=[self.profile, [], docs])
self.assert_(self._check_expected_profile(self.profile, categories))
def test_build_query_spanish(self):
"""Test that a profile is created given a spanish context"""
categories = DmozCategory.get_for_query(query="Einstein", lang='es')
ctx = """El mismo Einstein es conocido por haber rechazado algunas de las demandas de la mecánica cuántica"""
r = update_profile.apply(args=[self.profile, ctx, []], kwargs={'lang': 'es', 'terms': False})
self.assert_(self._check_expected_profile(self.profile, categories))
def test_spanish_query(self):
"""Test that a query in spanish works"""
result = list(DmozCategory.get_for_query(self.exact_queries['es'], 'es'))
if len(result) != 1:
self.fail('Only expecting one category!')
resultC = result[0]
self.assertEqual(resultC, DmozCategory.objects.get(title=self.exact_candidate))
def test_english_query(self):
"""Test that a query for a category in english works
Must return the quantum physics category and only that one!
"""
result = list(DmozCategory.get_for_query(self.exact_queries['en'], 'en'))
if len(result) != 1:
self.fail('Only expecting one category!')
resultC = result[0]
self.assertEqual(resultC, DmozCategory.objects.get(title=self.exact_candidate))
def test_empty_query(self):
"""Test that no categories are returned for a non-existant word"""
self.assertFalse(DmozCategory.get_for_query('lorem ipsum'))
def test_empty_query_as_dict(self):
"""Test that the dict option returns an empty iterable if an empty query is ensued"""
self.assertFalse(DmozCategory.get_for_query('lorem ipsum', as_dict=True))
def test_max_results(self):
"""Check that the max results constraint is respected"""
self.assertEqual(len(DmozCategory.get_for_query(self.query, max_results=1)), 1)
def test_score(self):
"""Test that an important candidate receives a high score"""
results = DmozCategory.get_for_query(self.query)
self.assert_(results[0].relative_weight >= 0.98)
def test_score_range(self):
"""Test that scores are numbers in [0,100]"""
results = DmozCategory.get_for_query(self.query)
self.assertFalse([e for e in results if (e.relative_weight > 1 or e.relative_weight < 0)])
def test_classify_terms_incorrect_lang(self):
"""Test that a profile is created with the default language if a given one is not supported"""
categories = DmozCategory.get_for_query(query="Einstein", lang='en')
r = update_profile.apply(args=[self.profile, 'Einstein', []], kwargs={'lang': 'fr', 'terms': True})
self.assert_(self._check_expected_profile(self.profile, categories))
def test_invalid_lang_query(self):
"""Test that, given an invalid language code, the method defaults to the base lang (english)"""
results = [e.pk for e in DmozCategory.get_for_query(self.query, lang='fr')].sort()
expected = [e.pk for e in DmozCategory.get_for_query(self.query)].sort()
self.assertEqual(results, expected)
def test_weighting(self):
"""Test that the categories receive a score"""
results = DmozCategory.get_for_query(self.query)
self.assert_(hasattr(results[0], 'relative_weight'))
def update_profile(profile, context, docs, lang='en', terms=True, **kwargs):
"""Update a profile with a spreading activation algorithm: determine the concepts in which the user might be interested,
save the session and update the activation values, proceeding then to update the profile itself
Args:
profile: the client user
context: the last context terms or text
docs: the ids of the documents of our database the user found interesting
lang: the language of the user
terms: whether the context is already a string of index terms or a full text
"""
#build the context list:
#context = context + list(DocumentSurrogate.)
lang = lang if lang in [e[0] for e in settings.LANGUAGES] else 'en'
#STEP 0: build the concepts set and set their activation values:
CON = {}
if not terms and context:
context = build_query(context, language=lang)
if not hasattr(context, '__iter__'):
context = [context,]
#Populate the concepts list with a dictionary of the form {concept: similarity}
for d in context:
CON.update(DmozCategory.get_for_query(d, lang, as_dict=True))
for d in DocumentSurrogate.objects.filter(pk__in=docs).values_list('category', flat=True).iterator():
#TODO: should I compute the document's summary similarity to its alleged category?
CON.update({d:1.0})
#logging.debug("Concepts gathered %s" % CON)
#Spreading: add to the interest list the attenuated weight of it's ancestors:
for c in CON.keys():
curr_concept = c
parent = DmozCategory.objects.filter(pk=curr_concept).values_list('parent', flat=True)[0]
while parent:
#multiple children of a parent might be in CON, ensure that the maximum score is the one that survives
#by selecting the maximum each time
ch_weight= DmozCategory.objects.filter(pk=curr_concept).values_list('weight', flat=True)[0]
CON.update({parent: max(CON.get(parent, 0.0), CON[curr_concept] * ch_weight)})
curr_concept = parent
parent = DmozCategory.objects.filter(pk=curr_concept).values_list('parent', flat=True)[0]
#logging.debug("After propagation: %s" % CON)
#STEP 2: Evolve the profile
#Use linear combination to update
existing_preferences = []
#logging.debug("Profile before update: %s" % profile.preferences.all())
for preference in profile.preferences.iterator():
#if the preference is not in this session, decay
ctg = preference.category.pk
if not ctg in CON:
preference.score = DECAY*preference.score
else: #it is, augment:
preference.score = DECAY*preference.score + (1-DECAY)*CON[ctg]
preference.save()
#add the preference to the set of existing ones:
existing_preferences += [ctg,]
#determine which preferences to add to the profile:
to_add = set(CON.keys()) - set(existing_preferences)
for newcat in to_add:
#pref = ClientPreference(category=DmozCategory.objects.get(pk=c), score=CON[newcat], user=profile)
#DO NOT store zero weighted preferences:
if CON[newcat]:
new_pref = ClientPreference(category_id=newcat, score=CON[newcat], user=profile)
new_pref.save()
#logging.debug("Profile after update: %s" %profile.preferences.all())
return True