A provenance-based approach to manage long term preservation of scientific data

Long term preservation of scientific data goes beyond the data, and extends to metadata preservation and curation. While several researchers emphasize curation processes, our work is geared towards assessing the quality of scientific (meta)data. The rationale behind this strategy is that scientific data are often accessible via metadata — and thus ensuring metadata quality is a means to provide long term accessibility. This paper discusses our quality assessment architecture, presenting a case study on animal sound recording metadata. Our case study is an example of the importance of periodically assessing (meta)data quality, since knowledge about the world may evolve, and quality decrease with time, hampering long term preservation.