Context Aware Computational Trust Model for Robust and Accurate Recommender Systems Algorithms for Ecommerce Platforms

Abstract

Online shopping has become part and parcel of our lives and more so as aggravated by the emergence of COVID-19 pandemic which necessitated need for social distancing and also work from home. This has led to unprecedented rise in online shops and consequently a myriad of alternatives for shoppers to consider before committing to a purchase. The myriad of alternatives has put a tall order on users in terms of information overload during decision making and made some shoppers to just rely on guesswork, putting them at a danger of losing income or lives to unscrupulous vendors. It is prudent to have a way of evaluating the how trustworthy an online shop is beforehand in order to assist the buyers to make meaningful decisions in time. In this study, we create a scale to estimate how trustworthy an online service provider is. We carry out a survey and then use factor analysis to come up with a model for estimating trustworthiness of an ecommerce platform from the consumer perspective. 2104 valid responses were attained from a total of 3,244 responses received from Google form whose link was shared directly to participant by reaching to them physically. The trust scale was then taken through reliability and validity tests. Confirmatory factor analysis yielded four components, which are security, privacy, deception and reliability. Cronbach’s alpha is found to be 0.956. The model is tested empirically and is found to improve the robustness and prediction accuracy of collaborative recommendation algorithms significantly.