Federal Project Grant Award Summary

The National Science Foundation's Division of Computing and Communication Foundations awarded $989,917 to the University of Pennsylvania (Award Date: July 15, 2026; Ultimate Completion Date: June 30, 2030) under the Computer and Information Science and Engineering program (CFDA 47.070) to develop an integrated query optimization framework for hybrid data systems. The project will create both offline and online optimization components that enable efficient searching, combining, and verification of information scattered across structured databases, public websites, academic papers, and unstructured web content. The system will integrate large language models (LLMs) with traditional data sources, incorporating performance semantic models for machine learning operators to estimate computational costs, route queries among alternatives while maintaining accuracy, and provide explanatory feedback for plan modifications. Key deliverables include validation mechanisms, incremental pruning techniques, dynamic replanning capabilities, and a background data curation agent designed to improve data extraction from unstructured sources.

Beyond core technical development, the project includes validation across multiple domains including life sciences, with outcomes such as software tools, performance benchmarks, and datasets contributed to the broader research community. The award explicitly supports workforce development through student training in artificial intelligence systems and modern data science practices. These outputs are designed to accelerate scientific discovery, improve data-driven decision making, and reduce the operational costs associated with deploying contemporary artificial intelligence tools for evidence gathering and synthesis across dispersed, heterogeneous information sources.