RBM Based Joke Recommendation System and Joke Reader Segmentation

Abstract

In the recent scenario, consumers are bared to a variety of information as well as commodities which leads to a variance in their choices. Recommender systems are a way to endure this challenge. An appropriate approach in recommending jokes on the basis of their preferences will be of substantial help in the future reference of the probable jokes. It is followed by segmentation of readers that has the potential to allow analysts to address each Joke-Reader in the most efficient way. This study aims at the development of a Joke Recommendation System based on Collaborative Filtering and Joke-Reader Segmentation based on the similarities in their preference patterns. A Bernoulli Restricted Boltzmann Machine (RBM) Model is implemented for constructing the Joke Recommender and k-means Clustering Model is deployed for achieving the Joke-Reader Segments. Considering the recommendation operation altogether, it is observed that the Joke-Reader Segmentation is firmly associated with the recommended ratings.