A set of tweets are characterized manually using human clarification with their sentiments and contemplate as training data. Then an additional set of tweets that is live streaming, are composed hinge on the text mining on Twitter Streaming Application Programming Interface (API). The tweets are fetched and retain as a text data and later will be utilized as a testing set. Testing data will grasp from training data calculation to forecast sentiment significance. The scenario of this research is no Twitter dataset collection available in the Malay language with characterizing sentiment significance. Then, findings are filtered to search tweets in the Malay language that is from Malaysia. The crucial dispute about this research is to cumulate a characterize entity as a training set. Considering there is no characterize Twitter entity available in the Malay language, a database of sentences is manually characterized with sentiments utilizing human explanation and utilize tweet's geo-position to search for tweets display within Malaysia. The outcome of this research, Twitter entity utilizing Twitter Streaming API capable to be collected and tweets from Malaysia collected by utilizing tweet's geo-position capable to be acquired.