Fashion Product Recommendation Platform With Frequent Itemset Mining

Abstract

The thesis proposes a fashion product recommendation system presented simulated in the form of on a web application platform with many other features such as post clothing item to sell, to share style between users like a social media, to visualize how a product looks like when mixed with other items, to save that collection for a later occasion in Virtual Look section of the website. In this work, I focus on building up a social media website with a recommendation model run below the web application system by using Frequent itemset mining, focusing on the information of the product to combine and generate a clothing collection for creating an association rule. I describe an implementation of this framework on a JSON server, and user interface as a Web Application to demonstrate this clothing recommendation system. Lastly, I discuss the approaches to applying Frequent itemset in a dataset by trying different algorithms to find out which is the most suitable for the model, determine the proposed quality of the system with a Confidence value.