This paper presents approach for mining and analysis of data from social media which is based on using Map Reduce model for processing big amounts of data and on using composite applications for performing more sophisticated analysis which are executed on environment for distributed computingbased cloud platform. We applied this system for creation characteristics of users who write about drugs and to estimate factors that can be used as part of model for prediction drug usage level in real world. We propose to use social media as an additional data source which complement official data sources for analysis and modeling illegal activities in society.