Judgemental Analysis of Data and Prediction Using ANN

Abstract

Hemanth N S N Sarisa, Avjeet Singh*, Govind Sharma and Harshit Mahajan

Everyday lot of posts are created in social media and several users make comments on the post. The comments can be positive or negative or moderate. Just by watching the post one cannot estimate whether the post is positive or not. Manual analysis of each comment of the post related document using Vader analysis takes a lot of time (approximately 5 to 6 hours- depending on number of comments). Also, every post comment contains lots of redundant data which should be removed to get the exact sentiment of the post. We need to analyze the sentiment of the post within fraction of seconds by eliminating the redundant data. By using Machine Learning ANN Algorithms and applying them to social media post related datasets which contains a large number of comments. Initially dataset to be imported and at first stage data pre-processing is to be performed so as to remove the useless redundant information or data. The Algorithm uses ANN Data Dictionary which contains list of positive words and negative words to classify each comment in the dataset into positive or negative or moderate category. Along with this it will also give the information about the sentiment of the post by visualizing the total positive, negative and moderate comments. Apart from this we are calculating the positive words and negative words number in each comment and also analyzing the sentiment of the post based on the keyword analysis. Graph and charts can be generated using Chart tool available in Visual Studio.

PDF