A Reputation Model for Aggregating Rating Levels Using Beta Distribution

With the speedy advancement in technology, modern electronic marketing has typically deployed the use of the World Wide Web for transactions online. The use of the web have significant challenges and decision making in retail for launching new products and services in an open environment of high risk. It is imperative to enhance and safeguard product's name and trust between the buyer and the seller to reduce such threats from the users' perceptive. This call for the development of rating clustering methods such as simple mean, naive, and normal distribution methods using the customers’ rating views to generate product or service reputation scores for administrative judgment. However, the existing models only considered rating level weights and average weights within the unbounded limit. This paper proposes a method to produce a more accurate and effective reputation score using the standard beta-distribution of a bounded interval. This method will take into account both sparse and dense datasets to generate a reputation score.