Technical Report: CSVM Ecosystem

TL;DR

The CSVM format extends CSV to include metadata, facilitating data management, exchange, and integration in scientific fields, with implementations and tools supporting its use for open and long-term data preservation.

Contribution

This paper introduces the CSVM format, a flexible extension of CSV with metadata, and demonstrates its application across multiple laboratories over ten years.

Findings

Facilitates data management without databases

Enhances data exchange and integration

Supports long-term RAW data preservation

Abstract

The CSVM format is derived from CSV format and allows the storage of tabular like data with a limited but extensible amount of metadata. This approach could help computer scientists because all information needed to uses subsequently the data is included in the CSVM file and is particularly well suited for handling RAW data in a lot of scientific fields and to be used as a canonical format. The use of CSVM has shown that it greatly facilitates: the data management independently of using databases; the data exchange; the integration of RAW data in dataflows or calculation pipes; the search for best practices in RAW data management. The efficiency of this format is closely related to its plasticity: a generic frame is given for all kind of data and the CSVM parsers don't make any interpretation of data types. This task is done by the application layer, so it is possible to use same format and same parser codes for a lot of purposes. In this document some implementation of CSVM format for ten years and in different laboratories are presented. Some programming examples are also shown: a Python toolkit for using the format, manipulating and querying is available. A first specification of this format (CSVM-1) is now defined, as well as some derivatives such as CSVM dictionaries used for data interchange. CSVM is an Open Format and could be used as a support for Open Data and long term conservation of RAW or unpublished data.