ESTIMATING DEGREE OF EMOTIONAL POLARITY OF TEXTUAL SENTENCE USING SENTIMENTAL ANALYSIS

The Key Task of Textual Exploration is to Estimate the Polarity of the Sentence. In several cases the polarity of the text is varied by the improper usage of words and improper orientation of the sentences. In this work we propose a method to reconstruct the improperly oriented text using a Deep-Text Correction module built using Deep Learning Network this neural network not only corrects the text but also standardizes the Lexicon before we calculate sentimental value of each line, which is not seen in any existing algorithms and then we use SentiWordNet to calculate the sentimental score obtained based on the text orientation to find the polarity of each sentence. The polarity can be divided into 3 classes positive, negative and neutral polarities. This result can be further used for text regeneration like generating a more polarized text or can also be used to generate a more context related text or can also be used to train much better speech synthesis models