Online social networks (OSN) has become highly popular, where users are more and more lured to reveal their private information. To balance privacy and utility, many privacy preserving approaches have been proposed which does not meet well users personalized requirements. In this paper, we present a privacy preserved Hadoop environment (PPHE) which automatically detects sensitive attributes using data mining techniques. This work considers Twitter which contains private information such as email addresses, mobile numbers, physical addresses, and date of births. First, we authenticate each Twitter user using the integrated algorithm RSA and Elgamal algorithm. Second, we categorize the tweets into private and non-private attributes based on the type-2 fuzzy logic system. Third, we apply a data suppression technique for private tweets and finally share the user's content based on their similarity information. Content similarity has been evaluated using cosine similarity. Finally we evaluate the system performance in terms of accuracy, precision, recall, and F-measure.