def findSimilarQuestions(questionBody):
try:
# Analyzing the question
entity_tags, topics, categories = analyzeQuestion(questionBody)
# Search query
query = {"$or": []}
# Generating the query (topic)
for topic in topics:
query["$or"].append({
"topics": {
"$elemMatch": {
"label": {
"$regex": topic["label"],
"$options": "i"
}
}
}
})
# No topic is found
if len(topics) == 0:
return None
# Results after checking topic similarity
questionsFromSimilarTopic = Question.findGeneric(query)
# Query returns None
if not questionsFromSimilarTopic:
return None
# Score algorithm
foundQuestions = []
for question in questionsFromSimilarTopic:
questionTopics = question['topics']
currentVector, questionVector = createTwoVectorsFromTopics(
topics, questionTopics)
cosSimilarity = round(
cosineSimilarity(currentVector, questionVector) * 100, 2)
foundQuestions.append({
'question': question,
'similarity_rate': cosSimilarity
})
# Sorting according to the topic score
foundQuestions = sorted(foundQuestions,
key=lambda x: x['similarity_rate'],
reverse=True)
for i in range(len(foundQuestions)):
# We do not need them in the response (probably)
del foundQuestions[i]['question']['entity_tags']
del foundQuestions[i]['question']['topics']
del foundQuestions[i]['question']['categories']
return foundQuestions
except Exception as e:
raise e
return []
