Adaptive Commodity Suggestion System - Match-BOT Shreyaan Kaushal School of Computer Science and Engineering Vellore Institute of Technology Email:shreyaan kaushal@live.com Taranveer Singh School of Computer Science and Engineering Vellore Institute of Technology Email:taranveersingh12@gmail.com Jatin Agrawal School of Computer Science and Engineering Vellore Institute of Technology Email:jatinagrawal31@gmail.com Manav Gulati School of Computer Science and Engineering Vellore Institute of Technology Email:manav.veergulati19@gmail.com Abstract—Potential customers of E-commerce garment businesses purchase clothes based on their personal choice as well as contemporary perception about fashion trends which is highly sublime in nature. This, coupled with the influential influx of cinema, social media and the Internet has led to rapid change in dressing preferences. E-commerce sites have a huge inventory categorized based on different attributes like colour, fabric, fit, product type and vendor. Hence, it is a tedious task for the user to find products coherent with the latest trend and appropriate according to their personal taste and global trends. The recommendation engines of these websites gives recommendations based only on a linear similarity of products, not comprehensively taking into account latest trends in fashion backed by fashion specialists, user preference or wardrobe collections. We propose an Adaptive Commodity Suggestion System called ”Match-BOT” which takes into account the trends obtained through insights provided by Google Trends, individual preferences, and similarity between user preferences. The user preferences are extracted by analyzing user purchase patterns and thus obtaining categorised product attribute preferences in the form of weights. This is an application of Content Based Learning The global trend obtained from Google Trends is then overlapped with the user preference and a user specific suggestion list is generated. The similarity between two user is calculated using Collaborative Filtering and then taken into account by including cross-user suggestions based on the degree of correlation. The output of this process would be in the form of list of suggestions of products which have been customized to society’s understanding of fashion and predilections of the user and constantly updated to reflect dynamic global trends. This would result in an unparalleled browsing experience for the user and an efficient sales and successive inventory management systems. Keywords—Adaptive Commodity Suggestion System, Global Trends, E-commerce Technology, User Preference, Collaborative Filtering, Content Based Learning I. I NTRODUCTION E-Commerce businesses are booming in this age of technology enabled commodity purchasing. As the industry rapidly moves from traditional brick and mortar businesses to E-Commerce businesses, it has become essential that the website interface interacts with the user and assists him/her to provide a shopping experience as closely emulated as possible to a what it would be like with a real salesman. This effect is realised through the recommendations and suggestions delivered to the user terminal during browsing time. Every user has a profile on E-Commerce websites, with associated preferences, purchase history and other relevant browsing information. But in its existing form, these suggestions are not sophisticated and are mostly a linear matching based on product purchase, purchase quantity distributions and query relevance analysis [4]. Through this paper, we hope to introduce a computationally inexpensive, sophisticated and accurate suggestion system which takes into account all the factors that contribute to popularity of a clothing article in a particular user’s perspective. We have assumed that if it is an existing user, all this information is condensed and available to the engine [8]. If that is not the case and the user in a new user, this information is to be availed via interactive forms by the E-Commerce site and the successive purchase and browsing history is built over time. For the purposes of this paper, we have focused on the Clothing Industry and all relevant information used to populate the data structures pertains to mainstream clothing articles available on popular E-Commerce websites [7]. As per Oxford Dictionary, the term ”f ashion” is defined as a popular style or practice, especially in clothing, footwear, accessories, makeup, body piercing, or furniture. It is the prevailing styles in behaviour and the newest creations of textile designers. The term ”trend” is defined as a general