Reproduce “TQL: Towards Type-Driven Data Discovery” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Andrew Kang, Sainyam Galhotra
TQL introduces a type-driven query language for tabular data that integrates data discovery and transformation, enabling users to articulate complex search criteria informed by downstream processing needs. The system, built on Type-Collected Relational Algebra, overcomes limitations of existing query languages by unifying operations for finding and manipulating datasets.
Key insights
- A language-driven design, leveraging principles from programming language research, offers greater flexibility and expressive power for data discovery compared to system-driven or model-driven approaches.
- Unifying data search and transformation within a single query language allows users to express not only the characteristics of desired data but also how that data might be composed or transformed to meet end-task requirements.
- Operating on collections of datasets, each typed by a corresponding predicate, provides a powerful abstraction for discovery, moving beyond querying explicitly known tables to finding suitable data sources.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.