Recommendation System Of Product Sales Ideas For MSMEs Using Content-based Filtering and Collaborative Filtering Methods

Abstract

Looking for an idea to differentiate a product from other sellers is not easy. Sometimes sellers of MSME products need sales recommendations on what is trending among the public. A product recommendation can help users recommend a product that is interesting and needed by that user. Recommendation systems can help users come up with previously unknown or unthinkable information, which can directly aid user knowledge in their search results. In this research, a recommendation system will be built to search for product ideas. This study uses content-based filtering and collaborative filtering methods as well as the TF-IDF algorithm to assist users in recommending the products they are looking for to assist users in finding product-selling ideas they expect. Previous research has examined the recommendation system for Modern Musical Instrument Sales using the Simple Additive Weighing method but has the drawback that the weighting calculation must use fuzzy numbers. Therefore, the content-based and collaborative filtering methods are assisted by the TF-IDF algorithm used in this study to answer these problems. After implementation, we test accuracy by dividing the test data and training data differently. System testing is done by using a confusion matrix. The results that have been tested get an accuracy of 78%. Subsequent research suggests adding MSME product data in recommending product sales ideas to MSMEs so that recommendations are more optimal.