expti

Expti is an open-source framework and data-standard initiative designed to facilitate the design, execution, and sharing of computational experiments across scientific disciplines. The project originated in 2019 through a collaboration among researchers in data science, software engineering, and information science who sought to improve reproducibility in computational research. Expti provides a modular runtime, a machine-readable schema for experiment definitions, and libraries that connect code, parameters, computational environments, and data provenance.

The architecture treats an experiment as a self-contained unit that includes an experiment plan, input datasets, and a run configuration. The runtime supports execution across local machines and cloud platforms via adapters, and emphasizes provenance, versioning, and auditability. Each run records software versions, hardware characteristics, data lineage, and parameter values, enabling precise replication of results. The framework uses a declarative configuration language and offers adapters for common programming languages, notably Python and R, to integrate existing analysis pipelines.

In practice, Expti is used in education to illustrate experimental design and in research to organize replication studies, sensitivity analyses, and benchmark experiments. It also serves as a bridge for journals and repositories to archive executable experiments with linked datasets and code. Critics point to the need for disciplined metadata practices and note potential overhead for small projects, but proponents argue that standardized experimentation workflows improve transparency and collaboration.

See also: reproducibility, data provenance, open science, workflow management systems.