Lexicon based opinion mining on twitter data by using hadoop

In this thesis, we will highlight findings of the assumptions obtained by using the methodologies of machine learning with Hadoop by Virtual Machine. The practical setup was started to carry out the experiments to study and find tweets by specific words and these tweets are to be collected only within a specific domain and data is to be saved in Hadoop. Then, training data such as the pre-processing operations is to remove all things which are not necessary and extract the features. After that, the classification of tweets using machine learning algorithms (supervised and unsupervised) with the ability to analyse the texts of tweet microblog is to detect emotions by different types of the lexicon. Furthermore, the cluster in Mahout was used to collect data at same polar to know what is best service or product which was expressed positively. Finally, we prove the objectives which were collected from the achieved results based on accuracy of the classification.