Improving Email Response in an Email Management System Using Natural Language Processing Based Probabilistic Methods

Email based communication over the course of globalization in recent years has transformed into an all-encompassing form of interaction and requires automatic processes to control email correspondence in an environment of increasing email database. Relevance characteristics defining class of email in general includes the topic of thee mail and the sender of the email along with the body of email. Intelligent reply algorithms can be employed in which machine learning methods can accommodate email content using probabilistic methods to classify context and nature of email. This helps in correct selection of template for email reply. Still redundant information can cause errors in classifying an email. Natural Language Processing (NLP) possess potential in optimizing text classification due to its direct relation with language structure. An enhancement is presented in this research to address email management issues by incorporating optimized information extraction for email classification along with generating relevant dictionaries as emails vary in categories and increases in volume. The open hypothesis of this research is that the underlying concept to fan email is communicating a message in form of text. It is observed that NLP techniques improve performance of Intelligent Email Reply algorithm enhancing its ability to classify and generate email responses with minimal errors using probabilistic methods. Improved algorithm is functionally automated with machine learning techniques to assist email users who find it difficult to manage bulk variety of emails.