from collections import Counter from models.person import Person import json ids_json = open('ids_for_clustring.json').read() ids = json.loads(ids_json) person = Person.find(5) all_tweets = person.tweets.all() tweets = [t for t in all_tweets if t.tweet_id in ids] all_hashtags = set() clusters = {} clusters_tweets = {} for tweet in tweets: hashtags = tweet.hashtags() all_hashtags.update(hashtags) assigned_cluster = False for cluster in clusters: cluster_hashes = clusters.get(cluster) if set(hashtags) == set(cluster_hashes): clusters_tweets[cluster].append(tweet) assigned_cluster = True break