class CategoryTagger(object):
def __init__(self,
top_level_classifier_host="ml-services.production.indix.tv",
threshold=0.75):
self.threshold = threshold
self.classifier_query = ClassifierQuery("http",
top_level_classifier_host,
"/api/classify")
self.client = HttpClient()
def tag(self, title):
title = title.lower().strip()
q = self.classifier_query.getSearchQuery(title)
json_response = json.loads(self.client.query(q))
conf_scores = [
float(score) for score in json_response['confidence'].split(",")
]
result_dict = {}
result_dict['brands'] = []
result_dict['stores'] = []
result_dict['categories'] = []
if conf_scores[0] > self.threshold:
matches = {}
matches['matches'] = [json_response['category_id']]
matches['token'] = title
result_dict['categories'] = [matches]
return (result_dict, title)
class Q2Category(object):
def __init__(self, qas_host="qas01.staging.indix.tv:8080"):
self.qas_query = QASQuery("http", qas_host, "/api/annotate")
self.client = HttpClient()
def getCategories(self, search_term, threshold=0.025):
query = self.qas_query.getSearchQuery(search_term)
qas_response = self.client.query(query)
cat_conf_dict = qas_response["ids"][0]["classes"]
categories = []
for cat, conf in cat_conf_dict.items():
if conf >= threshold:
categories.append(int(cat))
return categories
