Data validation consists of a number of automatic software routines to check that data are valid according to a set of rules (as opposed to reviewing, which is an inspection and commenting by physical persons). Validation rules can cover consistency with format requirements, consistency and plausibility of data points within datasets, and consistency and plausibility between datasets within the database.
The purpose of data validation is to reduce the reliance on physical persons for review and manual editing. The ecoinvent database, for example, currently applies more than 100 validation rules for checking the quality of new data before they enter the review procedure. This saves reviewers from a lot of unnecessary and repetitive work, and catches simple errors that might go unnoticed in a manual review procedure.
This Working Group will initially make a structured collection of validation rules used in existing footprint databases. The Working Group will complement these validation rules based on the experiences with manual reviews and the resulting error corrections typically made in existing footprint databases.
The Working Group is responsible for the development and maintenance of tools and procedures for machine-assisted data exploration and learning, to allow system and domain experts to continuously improve the consistency and plausibility rules and the validation procedures of the BONSAI database. All validation rules and procedures will be maintained as open source.
The Working Group is also responsible for ensuring that validation results are presented to data providers, reviewers and users in an optimal way for their use in further data processing, improvement and documentation.
This Working Group is expected to be composed of developers from the existing LCA/footprint databases and softwares.