Rhyme: A Data-Centric Multi-paradigm Query Language Based on Functional Logic Metaprogramming

We present Rhyme, a declarative multi-paradigm query language designed for querying and transforming nested structures such as JSON, tensors, and beyond. Rhyme is designed to be multi-paradigm from ground-up allowing it to seamlessly accommodate typical data processing operations–ranging from aggregations and group-bys to joins–while also having the versatility to express diverse computations like tensor expressions (à la einops) and declaratively express visualizations (e.g., visualizing query outputs with tables, charts, and so on). Rhyme generates optimized JavaScript code for queries by constructing an intermediate representation that implicitly captures the program structure via dependencies. This paper presents a system description of Rhyme implementation while highlighting key design decisions and various use cases covered by Rhyme.