Web-based Instant Messaging Application with Functionality for Filtering Distracting Content

Abstract

The internet has provided us various modes for transmission of data from one device to another such as emails, SMS/MMS, instant messaging applications etc. Instant messaging applications are very widely used for their cost effectiveness and facility for quick communication. Within the meantime, several undesirable messages are received by users which can be classified as spam, distracting, redundant. Despite the fact that people have the flexibility to promptly acknowledge a message as undesirable, performing such a task could be a waste of time. A platform independent instant messaging web-based application is developed which can filter distracting messages in real time. The application uses socket.io library for bidirectional communication necessary for instant messaging and uses NodeJS, MongoDB for routing and storing data respectively. The real time filtering of messages is done by Support Vector Machine classifier and Stochastic Gradient Descent. The machine learning algorithms are integrated with the web part with the help of APIs.