Machine Learning and scraping to create and test a large curated dataset of liquidus temperature in oxide glasses

• Grant number: 25/01797-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: June 01, 2025
• End date: May 31, 2026
• Field of knowledge: Engineering - Materials and Metallurgical Engineering - Nonmetallic Materials
• Principal Investigator: Edgar Dutra Zanotto
• Grantee: Gustavo Laranja Thomaello
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil
• Associated research grant: 13/07793-6 - CEPIV - Center for Teaching, Research and Innovation in Glass, AP.CEPID

Abstract

This project is intended to: 1) Develop a computational code that harvests chemical composition vs. liquidus temperature (TL) of oxide glass formers combining several datasets (TL) already available on the internet (patents and SciGlass) to create a dataframe that is easy to access and handle this information. 2) Submit the generated dataframe to a battery of joint tests using descriptive techniques and predictive Machine Learning (ML) models combined with Genetic Algorithms (GAs) to obtain oxide glass compositions with desired values TL. 3) Combine the dataframe generated with other databases publicly available to infer statistically how the dataframe generated by our code is consistent and can increase the performance of ML models acting on these databases. The ultimate goal is to develop glasses with desired liquidus temperature.