Project Grant 2431503

This National Science Foundation (NSF) Computer and Information Science and Engineering (CISE) Federal Grant Program (CFDA 47.070) award to the University of Texas Rio Grande Valley (UTRGV) provides $246,872 over 3 years starting January 1, 2025 to develop a time series text-based question answering system.

The project aims to address three key challenges in leveraging large language models for time series data: 1) lack of high-quality text information aligned with time series data, 2) under-research on deep learning models to reason across time series and text modalities, and 3) lack of explainable AI tools to ensure model trustworthiness. The award will fund the development of an automated pipeline to annotate time series data, a cross-modality pre-trained model to extract knowledge from time series and text, and explanation methods to enhance model transparency. The system will be evaluated on extreme weather forecasting through a collaboration with Idaho National Laboratory. This project promotes UTRGV's research capacity in advanced AI topics and encourages Hispanic undergraduate students to pursue higher education.