Li, Y. T. R., & Cheung, A., & Li, Y., & Hsieh, C., & Timorabadi,, H. S. (2023, June), Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project  Paper presented at 2023 ASEE Annual Conference & Exposition, Baltimore , Maryland. 10.18260/1-2--43555

\bibitem{asee_peer_43555}
  Li, Y. T. R., \& Cheung, A., \& Li, Y., \& Hsieh, C., \& Timorabadi,, H. S. (2023, June), \emph{Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project}
  Paper presented at 2023 ASEE Annual Conference \& Exposition, Baltimore , Maryland.
  10.18260/1-2--43555

Yu Tong (Rayni) Li, Abby Cheung, Yongjie Li, Carmen Hsieh, and Hamid S. Timorabadi, P.E..  "Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project".  2023 ASEE Annual Conference & Exposition, Baltimore , Maryland, 2023, June.  ASEE Conferences, 2023.  https://peer.asee.org/43555 Internet.  15 Dec, 2024

\bibitem{asee_peer_43555}
  Yu Tong (Rayni) Li, Abby Cheung, Yongjie Li, Carmen Hsieh, and Hamid S. Timorabadi, P.E..
  "Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project".
  \emph{2023 ASEE Annual Conference \& Exposition, Baltimore , Maryland, 2023, June}.
  ASEE Conferences, 2023.
  https://peer.asee.org/43555 Internet.  15 Dec, 2024

@INPROCEEDINGS{asee_peer_43555
author = "Yu Tong (Rayni) Li, Abby Cheung, Yongjie Li, Carmen Hsieh, and Hamid S. Timorabadi, P.E."
title = "Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project"
booktitle = "2023 ASEE Annual Conference \& Exposition"
year = "2023"
month = "June"
address = "Baltimore , Maryland"
publisher = "ASEE Conferences"
note = {https://peer.asee.org/43555}
number = {10.18260/1-2--43555}
}

TY  - CPAPER
AB  - The COVID-19 crisis and the accompanying period of isolation has undeniably impacted consumer behavior. In the last few years, the world has seen global e-commerce sales increase significantly, pushing retailers to expand their online catalogs. This shift towards online shopping is expected to last and continue for the years to come. Skincare, encompassing the categories of facial and body care, skin protection, and make-up removal products, is one industry that has expanded its online landscape and has become increasingly popular throughout the pandemic. However, with a saturated skincare industry, shoppers are often faced with too many choices, leaving them confused or frustrated. Companies have developed technology to customize recommendations but provide choices limited to their products. Currently, the industry lacks an all-inclusive application that generates user-customized recommendations to allow consumers to focus on suitable products. This capstone project aims to fill this gap by building a web application that helps consumers who are aware of their basic skin conditions and needs, including skin type and category of their product of interest, find skincare products tailored to their needs and skin conditions.

The web application will collect sufficient and relevant data on skincare products from multiple sources, such as the number of reviews, ratings, ingredients, and product descriptions. The data is then cleaned, stored in, and retrieved from a database, in order for it to be ranked according to user preferences. This ranking will be generated for the recommender system which leverages an existing open-source content-based recommendation API, TensorRec. In order to train and improve the accuracy of TensorRec’s API, sample input will be generated from pre-built user profiles and fed into the recommendation system. As a result, the user will receive a list of ranked product recommendations, with access to additional information such as product descriptions, ingredients, online reviews, and retailers. From the list of products, the interface will allow users to ‘like’ or ‘dislike’ a recommended product, which will be saved to their profile and used to tailor future recommendations. 

The user interface of the web application will be designed with usability in mind. A major objective of the application is user accessibility, specifically for individuals who are blind or visually impaired. Considering font size, color contrast, screen reader compatibility, and keyboard accessibility, these specific design factors allow for easier readability and accessibility. Furthermore, by restricting the number of input questions and constraining the maximum querying time, the application provides users with a comprehensive way of determining the best products for their skin in a timely manner.

Utilizing machine learning predictions with personalized user profiles allows for tailored recommendations that are suited to individual needs. The web application solution efficiently compiles relevant product information necessary for consumers to decide between products in a centralized location. The application is currently being tested and users’ surveys are being conducted. The surveys’ outcomes and feedback are used as metrics to measure the level of success of the application as well as to further improve the application.
AU  - Yu Tong (Rayni) Li
AU  - Abby Cheung
AU  - Yongjie Li
AU  - Carmen Hsieh
AU  - Hamid S. Timorabadi, P.E.
CY  - Baltimore , Maryland
DA  - 2023/06/25
PB  - ASEE Conferences
TI  - Matilda: A Machine Learning Software Application to Virtually Assist with Skincare for Visually Acute and Impaired—A Capstone Design Project
UR  - https://peer.asee.org/43555
DO  - 10.18260/1-2--43555
ER  -