Curation means adding value to the data during its life cycle, i.e., making sure that has meaning and context and can be re-used outside the creator’s environment. Curation infrastructure includes repositories, access procedures, policies, processes and institutional support. The unit of curation and preservation is a file. It’s important to maintain files integrity by ensuring their fixity, duplicate storage and format validation.
To preserve files, they need to be in proper formats, i.e., durable (transparent, documented, used widely, renderable) and supported with standards (syntactic and semantic). Syntactic standards don’t have context (e.g., in CSV we don’t know how columns were created and what they mean). Semantic standards are better.
In preservation a good practice is to use a master file for preservation (highest quality and fidelity) and derivative files for active use and delivery. For example, high-resolution TIFF for preservation and lossy JPEG for viewing
Source: DIKW: Data, Information, Knowledge, Wisdom