Honors Program
Honors Program: Embargoed Theses
Date of this Version
5-2025
Document Type
Thesis
Citation
Hsu, E., Chin, V., Henrriquez, K., & Damle, R. AI Shopping Assistant. Undergraduate Honors Thesis. University of Nebraska-Lincoln. 2025.
Abstract
Buckle, Inc. is a leading retailer of high-quality casual apparel, footwear, and accessories for fashion–conscious men and women. Headquartered in Kearney, Nebraska, the company currently operates over 440 stores in more than 40 states. Buckle prioritizes providing the best possible guest experience through individual customer services.
Through this project, Buckle aims to optimize the shopping experience by integrating our natural language search solution into their existing internal applications. This will allow internal users to help customers find the perfect product for any occasion.
The Design Studio team developed a model powered by Retrieval Augmented Generation (RAG) to enhance the search experience. This allows users to find products using natural language instead of a typical keyword search. Current solutions may show the user products that aren’t contextually relevant to their search, which is not effective. For example, a search for “winter vibes” won’t return scarves because “winter” isn’t a keyword in the product name for scarves.
We used Anthropic’s Claude Sonnet model to create lists of words (‘tags’) that describe each of the provided products’ features. We created these tags to enhance the model’s knowledge about the Buckle products and give it context so the model can make the best recommendation. This solution improves the clothing shopping experience by allowing for more flexibility when looking for the perfect product. Instead of needing to use rigid keywords to find a specific clothing item, a user can use natural language to find products that fit any occasion.
Comments
Copyright Ealynn Hsu, Victoria Chin, Katia Henrriquez, and Radhika Damle 2025