Twitter has involved lots of users to share and distribute most recent information, resulting in a large sizes of data produced every day. Many private and/or public organizations have been reported to create and monitor targeted Twitter streams to collect and know users opinions about the organizations. However the complexity and hybrid nature of the tweets are always challenging for the Information retrieval and natural language processing. Targeted Twitter stream is usually constructed by filtering and rending tweets with certain criteria with the help proposed framework. By dividing the tweet into number of parts Targeted tweet is then analyzed to the understand users opinions about the organizations. There is an promising need for early rending and categorize such tweet, and then it get preserved on dual format and used for downstream application. The proposed architecture shows that, by dividing the tweet into number of parts the standard phrases are separated and stored so the topic of this tweet can be better captured in the sub sequent processing of this tweet Our proposed system on largescale real tweets demonstrate the efficiency and effectiveness of our framework