Product Recommendation and Context Validation Based E-Commerce System With Currency Free Economy

Abstract

Due to the covid-9 situation, online shopping shows rapid growth among Sri Lanka and other countries. Meanwhile, with the visible downward trend of the Sri Lankan economy, people have been suffering due to inflation, leading to higher expenses of goods and services. A web-based solution called ‘Ceylon Barter Bay’ was developed as an e-bartering platform for Sri Lankans to get bartering experience and develop one-to-one trading. This paper comes with an appropriate business model for ‘Ceylon Barter Bay’ as a novice entrepreneur idea. This website was developed with enhanced abuse detectors and a related product recommendation system. Natural language processing and machine learning techniques are used in the process to get a better solution. Since the developed system is mainly based on advertising, a random forest algorithm-based machine learning model with 99% accuracy detects the context offensiveness. To detect violent behavior in feedback/comments, the logistic regression algorithm-based machine learning model was used with 88% accuracy. ‘Ceylon Barter Bay’ will recommend related items. Both collaborative and content-based recommendations have been performed using linear regression, respectively. In Sri Lanka, this has been recognized as an acceptable solution to break the monopoly of money via a web-based application