Amrita-CEN-SentiDB 1: Improved Twitter Dataset for Sentimental Analysis and Application of Deep learning

Powerful weapon in today's world is ones emotion is social media. They have the power to make an individual trending overnight or even may pull down anyone reputation. In this paper, the sentiment analysis task has been performed by collecting the dataset from the publically available sources and by merging them together to form a new dataset for the sentiment analysis task that is positive or negative sentiment based on the context of the subject. A new reliable dataset is subjected to various pre-processing techniques and then the feature extraction techniques aftermath they are passed to the deep learning techniques out of which by using the text representation method, global vectors (glovec) with the long short-term memory (lstm) has the highest accuracy of 75%, which is the benchmark accuracy for this dataset. For the research purpose the dataset used in this paper is made available publically for research purpose.