Development Of Body-aware Fashion Recommendation Chatbot Using OpenAI Api For E-commerce

Abstract

The rapid growth of e-commerce has transformed the fashion retail landscape, offering convenience and variety but also introducing new challenges related to personalization and fit accuracy, and user confidence, especially in the absence of physical try-on experiences. Clothing choices are often influenced not only by style but also by psychological and body image factors, including body shape, height, weight, and skin tone. These complex variables are often overlooked in current fashion recommender systems, which tend to rely heavily on historical data or generalized filters, resulting in recommendations that lack relevance and inclusiveness. This research presents the development of a body-aware fashion recommendation chatbot utilizing the OpenAI API by implementing the GPT-3.5 Turbo model to provide personalized clothing suggestions through natural language conversations. The chatbot system integrates multiple components, including a fashion-focused prompt engineering strategy that enables the language model to consider physical attributes and user preferences during recommendation generation. The product catalogue data was gathered via web scraping from local e-commerce platforms using Selenium and Beautiful Soup, while the web-based chatbot interface was implemented using a combination of PHP, Python, HTML/CSS, and JavaScript. To manage multi-turn dialogues and maintain context, a linked keyword system structure was designed, enabling the system to track user preferences throughout the conversation. The recommendations are refined through SQL queries that filter products based on real-time inventory availability, user preferences, and inferred body-UHODWHGSDUDPHWHUV7KHFKDWERW¶V performance was evaluated through User Acceptance Testing (UAT) using the ResQue framework, covering eight indicators: recommendation quality, interaction adequacy, interface adequacy, interface sufficiency, ease of use, perceived usefulness, transparency, user attitude, and behavioral intent. The findings demonstrate that incorporating body-aware personalization into a conversational AI framework significantly enhances the relevance, usability, and WUXVWZRUWKLQHVVRIDIDVKLRQUHFRPPHQGDWLRQ8VHUVH[SUHVVHGVDWLVIDFWLRQZLWKWKHFKDWERW¶V ability to understand their needs, provide appropriate suggestions, and improve their confidence during the online shopping experience.