Creation of Derivatives (CPP-028)

Definition and scope

The process of Creation of Derivatives refers to generating new Representations of a digital Object (based on the original Representation of the Object) to address specific needs of the Designated Community. Some of these needs are:

• Provide easy access to the information for a broader community (e.g. conversion from TIFF to JPEG in order to allow viewing in a web browser)
• Provide access for a specific user base (e.g. converting CAD Files to a specific format that is more commonly in use by a group of researchers)
• Provide different forms of access to the data (e.g. combining a tree of HTML-pages, images and other resources into a WARC File; unpacking a ZIP File; compiling data into a pivot table or chart)
• Provide access to the data in formats that address accessibility challenges (e.g. creating DAISY derivatives of PDFs for visually impaired consumers)
• Restricting or protecting the access to the data (e.g. creating a PDF from an e-book containing only the first few pages; lowering the resolution and watermarking; thumbnails)
• Enriching the information (e.g. adding AI-generated transcriptions as subtitles to a video; overlaying transparent OCR text over a scanned image in a PDF; adding AI-generated Descriptive metadata)

Creation of Derivatives differs from other processes that create new Files or Representations:

• Unlike File Migration (CPP-014), Creation of Derivatives reproduces only the information and significant properties of the original that is useful to address specific needs of the Designated Community. Thus, the output of Creation of Derivatives may lack part of the information that is not required to satisfy said needs.
• File Normalisation (CPP-026) is performed during ingest and aims at reducing the number of formats by converting Files or Representations to preservation formats defined by the institutional formats policy.
• File Migration (CPP-014) is performed after ingest to address a specific preservation risk.

Derivative copies can be short-lived and there can be more than one derivative copy for a given original. For example, derivative copies for a video master File could include multiple resolutions, different levels of compression and different encoding schemes (e.g. suited for previewing rather than reuse). The derivative videos may be generated to support contemporary web delivery standards such as HTML5 and as a result may be short lived compared to the copies created as a result of migration and/or normalisation.

It is up to the TDA to decide how far they take this and if they support any of these use cases, if at all. The generation of the derivative copies can be performed during delivery, during or before the ingestion or any time during the Object’s life cycle in the TDA. Performance, Object size, availability of conversion tools and workflows, and user requirements will be some of the elements that could play a role in this decision.

Depending on whether the derivative copy maintains the significant properties of the original Object, the derivative copy may then (a) be added to the same AIP as the original (e.g. as a new Representation), (b) result in a new AIP that is stored in addition to the original and is linked to it, or (c) may be held in a DIP which in turn may be stored temporarily or permanently, and is linked to the original AIP. In PREMIS terms, the derivative copies can be stored as a new Representation if the derivative copy is a complete and meaningful rendition of the Intellectual Entity. However, if the derivative copies change the meaning and of the original (e.g. significant properties are not maintained) then a new AIP or DIP is recommended.