Credibility as Informativeness: Predicting Informative Tweets

Social media is an important source of facts at some stage or in any principal event, mainly natural disasters or trends. Since increasingly users take part in numerous social networking services, the extent of streaming information is substantially increasing. It is becoming a big problem to discover informative messages from a large amount of data. In this paper, we have proposed multi-methods to investigate the hassle of detecting informative messages with the data we gathered from Twitter and to classify a tweet into informative and non-informative based on some factors which we found at most important in the classification of informative messages. Also, we have proposed an effective technique to solve the classification of Tweets. Most of the data we gathered had a usual behavior which we termed as garbage data which were either a useless tweet, emojis, characters, short forms of the sentence, etc. Our work was to overcome these hassles and to calculate the meaningful features which we can describe as one of the most important influences in the classification of tweets in any size of data. This would allow us to normalize our list of features and the accuracy of our Work. Which can increase the probability of informative or non-informative