Adaptive metadata is a model for deciding what metadata attributes are applied to assets based on multiple distinct definable characteristics. For example, if the asset is time based (e.g. audio or video) then a duration attribute would be present, whereas it would not be for a static asset such as a photo. The theoretical basis of Adaptive Metadata is that assets have multiple characteristics, they have a hierarchical relationship and they might cross-refer to other associated entities, for example, user permissions or another taxonomy unrelated to the one the asset itself exists within. Assets can inherit multiple sets of characteristics also to create complex metadata models that can be modified (adapted) to enable DAM solutions to model a wide range of assets from many different sources.
DAM solutions which fully implement adaptive metadata do not just hide or suppress fields but create a specific associative relationship between the field and the asset based on one or more of the characteristics. Adaptive metadata can also be referred to using a more technical term such as 'class oriented metadata' to highlight the relationship with object oriented software engineering principles (e.g. inheritance or polymorphism) which underpins the implementation methods typically employed to realise the concept.
Adaptive Metadata as a term was first used by vendor, Picturepark. There are other DAM solutions which use similar methods, but the developers involved have chosen to focus on the ability of their metadata model to support key asset attribute such as asset type or permissions. See http://picturepark.com/dam-blog/picturepark-tips/what-is-adaptive-metadata/
Click version numbers to view differences.