def addTeachedTweets(self, tweets):
""" добавляет информацию о вручную определённых твиттах в базу. обновляет индексы для слов """
for tweetId in tweets:
if self.tweetsColl.find_one({"_id": int(tweetId)}) is not None:
continue
tweet = TwitterStuff.getTweet(tweetId)
if tweet is None:
continue
words = Parser().parseTweet(tweet['text'])
words.append("@" + tweet['user']['screen_name'])
self.tweetsColl.save({"_id": int(tweetId), "happy": tweets[tweetId]})
for word in words:
if len(word) < 4:
continue
self.saveWord(word, tweets[tweetId])
def addTeachedTweets(self, tweets):
""" добавляет информацию о вручную определённых твиттах в базу. обновляет индексы для слов """
for tweetId in tweets:
if self.tweetsColl.find_one({"_id": int(tweetId)}) is not None:
continue
tweet = TwitterStuff.getTweet(tweetId)
if tweet is None:
continue
words = Parser().parseTweet(tweet['text'])
words.append("@" + tweet['user']['screen_name'])
self.tweetsColl.save({
"_id": int(tweetId),
"happy": tweets[tweetId]
})
for word in words:
if len(word) < 4:
continue
self.saveWord(word, tweets[tweetId])
def predict(self, text, user):
words = Parser().parseTweet(text)
words.append(user)
explains = []
for word in words:
explain = self.wordsColl.find_one({"_id": word})
if explain is None:
continue
explains.append(explain)
explains.sort(lambda x1, x2: x1["number"] - x2["number"])
explains.reverse()
res = dict({"happy": 0, "neutral": 0, "negative": 0, "spam": 0})
states = {"happy", "neutral", "negative", "spam"}
for explain in explains:
n = explain["number"] / float(explains[0]["number"])
for state in states:
res[state] += explain[state] * n
best = ""
bestRate = -1
for state in states:
if res[state] > bestRate:
bestRate = res[state]
best = state
return best
def predict(self, text, user):
words = Parser().parseTweet(text)
words.append(user)
explains = []
for word in words:
explain = self.wordsColl.find_one({"_id": word})
if explain is None:
continue
explains.append(explain)
explains.sort(lambda x1, x2: x1["number"] - x2["number"])
explains.reverse()
res = dict({"happy": 0, "neutral": 0, "negative": 0, "spam": 0})
states = {"happy", "neutral", "negative", "spam"}
for explain in explains:
n = explain["number"] / float(explains[0]["number"])
for state in states:
res[state] += explain[state] * n
best = ""
bestRate = -1
for state in states:
if res[state] > bestRate:
bestRate = res[state]
best = state
return best
