A Comprehensive Artificial Intelligence Based User Intention Assessment Model from Online Reviews and Social Media

Abstract

ABSTRACT Predictive analytics is being increasingly used to predict various aspects of applications and users. It offers vast opportunities in the growth of the modern era’s business transformation by enabling automated decision-making processes. Being able to determine the intention of users in an automated way is one of the important factors in enabling automated decision-making for applications and businesses using such applications. In this paper, we utilize and build upon the existing works, and propose a comprehensive intention assessment model that detects different possible intents of users by analyzing their text-based reviews on online forums, retail market websites, or on social media. If the information about a product or service experience is present somewhere in a review or post, our technique can accurately segregate different possible purchase intention labels (i.e., positive, negative, and unknown). Our proposed comprehensive model for intention assessment includes extensive data pre-processing, extended feature selection model, utilization of artificial intelligence (machine learning and deep learning) techniques, and customized cost and loss functions. We built a comprehensive testbed and carried out evaluations and comparisons. Our solution demonstrates high accuracy, precision, and F1 score. The proposed solution helps in mining and gaining deeper insights into behavior of consumers and market tendencies and can help in making informed decisions.