How useful is informationWithheld

[peterdesmet]

@tucotuco how useful is informationWithheld for sampled down datasets, if the only information in that field for all records is see metadata?

[tucotuco]

I would argue that "see metadata" is far less useful in an aggregated setting than it is when looking at a data set in isolation. If the information is with every record, it will not be lost when a result contains data from lots of data sets. In other words, it helps the end user.

For sampled down data sets though, I think what you want is dataGeneralizations, unless something is simply being omitted as opposed to being made less specific.

[peterdesmet]

Ok, let me clarify with an example

tracking dataset

• dataGeneralizations: subsampled by hour: first of 2 record(s)
• informationWithheld: see metadata could refer to the fact that e.g. acceleration records are not included that might be available in source dataset

camera trap dataset

• dataGeneralizations: (no example yet)
• informationWithheld: see metadata could refer to the fact that e.g. observations of humans and non-animal objects are excluded from this dataset, but maybe that is a dataGeneralization?

So for a generic mapping procedure (i.e. a function), I think I would argue that dataGeneralizations is really useful to set when the function samples down, but that informationWithheld: see metadata depends a bit on dataset per dataset and is less useful to set by the function?

[peterdesmet]

Conclusion: avoid informationWithheld: see metadata and only use the term if the script can provide something more meaningful here, like informationWithheld: camera settings.