def proposed_similarity():
try:
request_weights = feature_weights.copy()
body_string = request.data.decode("utf-8")
if len(body_string) == 0:
return jsonify("Empty String")
if 'weights' in request.headers:
header = json.loads(request.headers['weights'])
print(header)
for weight_key in header:
if weight_key in request_weights:
request_weights[weight_key]['weight'] = float(header[weight_key])
# pdb.set_trace()
method_vector = util.vector_from_text(body_string)
nearest = ml.proposed_kNearest(method_vector, solution_vectors, lang_dict,
lang_model, k=kNearest, weights=request_weights)
# nearest contains score, solutions_index and intersections
# print(kNearest)
nearest_vectors = []
for element in nearest:
result = {}
result['score'] = element[0]
result['code'] = solution_vectors[element[1]]
result['scores'], result['matches'] = extract_info(element[2])
nearest_vectors.append(result)
return jsonify(nearest_vectors)
except Exception as ex:
print(ex)
return jsonify(str(ex))
def mix_similarity():
try:
request_weights = feature_weights.copy()
body_string = request.data.decode("utf-8")
if len(body_string) == 0:
return jsonify("Empty String")
if 'weights' in request.headers:
header = json.loads(request.headers['weights'])
for weight_key in header:
if weight_key in request_weights:
request_weights[weight_key]['weight'] = float(
header[weight_key])
method_vector = util.vector_from_text(body_string)
# import pdb; pdb.set_trace()
proposed_nearest = ml.proposed_kNearest(method_vector,
solution_vectors,
lang_dict,
lang_model,
k=5,
weights=request_weights)
cosine_nearest = ml.cosine_kNearest(method_vector,
solution_vectors,
baseline_dict,
baseline_model,
k=5)
proposed_nearest = [tup[1] for tup in proposed_nearest]
cosine_nearest = [tup[1] for tup in cosine_nearest]
results = []
for i in range(len(proposed_nearest)):
reco_index = proposed_nearest[i]
reco = {}
if reco_index in cosine_nearest:
reco['source'] = 2
reco['rank_1'] = i + 1
reco['rank_0'] = cosine_nearest.index(reco_index) + 1
cosine_nearest.remove(reco_index)
else:
reco['source'] = 1
reco['rank'] = i + 1
reco['text'] = solution_vectors[reco_index].raw_text
results.append(reco)
for i in range(len(cosine_nearest)):
reco_index = cosine_nearest[i]
reco = {}
reco['rank'] = i + 1
reco['source'] = 0
reco['text'] = solution_vectors[reco_index].raw_text
results.append(reco)
random.shuffle(results)
return jsonify(results)
except Exception as ex:
print(ex)
return jsonify(str(ex))
def concept_tag_kNearest():
try:
body_string = request.data.decode("utf-8")
if len(body_string) == 0:
return jsonify("Empty String")
method_vector = util.vector_from_text(body_string)
# first transform the given method_vector with our model
nearest = ml.concept_tag_kNearest(method_vector, solution_vectors, lang_dict,
lang_model, k=kNearest)
# nearest contains score, solutions_index and intersections
# print(kNearest)
nearest_vectors = []
for element in nearest:
result = {}
result['score'] = element[0]
result['code'] = solution_vectors[element[1]]
result['match'] = element[2]
nearest_vectors.append(result)
return jsonify(nearest_vectors)
except Exception as ex:
return jsonify(str(ex))