Soft Computing Approaches to Classification of Emails for Sentiment Analysis

Email is a fast and well-liked communication medium on the internet. Email users are rapidly increased due to easy availability of internet. Email is used for personal as well as official communications. It is also used for illegitimate activities such as phishing, spamming, abusing, and threatening. Email mining gives the better solution to this problem. The clustering and classification methods of data mining are used to classify the emails into different categories. The paper tries to extract effective features for investigating an email to identify the sentiment which is helpful for forensic people. Data mining approaches such as k-means clustering, fuzzy c-means clustering and neural network backpropagation algorithm were applied on extracted features for classification of emails as per the sentiments hidden inside them. Evidence can be generated from the Negative sentiments. The paper does the comparative analysis of various algorithms. For this problem, neural network backpropagation algorithm gives the best recognition rate.