Metadata is data about data. Consequently, it allows for very structured documentation, with descriptors facilitating cataloguing data and data discovery. Metadata is intended for machine-reading, and thanks to metadata you can browse sites like Zenodo using filters like Access Right and File Type.
Watch this web lecture to learn about the different types of metadata and how metadata can help make your research data better findable. You are pointed to useful sources for metadata standards.
There is generally no need to generate machine-readable metadata yourself. Metadata often follows a certain schema, which is nothing but a predefined set of elements. There are also metadata standards, which in some cases are domain-specific. An example of such a standard is the DublinCore schema, which is a set of 15 metadata elements applicable to a wide range of datatypes
Different disciplines like biology, earth sciences, physical sciences and social sciences and humanities have their own standards. By choosing a well-supported standard, you will maximise the chance that your data can be (re)used and understood by other researchers.
You are advised to use a ‘controlled’ vocabulary when you collaborate with others or when you regularly do the same type of research, A ‘controlled’ vocabulary uses predefined and authorised terms. This avoids that the same concept is given different names and ensures consistency. Hence it makes it easier to find and understand data.
Without you knowing, you may already use a generally accepted controlled vocabulary, for instance:
If there is no vocabulary available, you can consider making a custom list within your research team.
Units are important, and metadata field names need explanations in for instance a codebook or another document explaining names. If metadata field names are from a standard, these are explained in the standard. This saves you time and you can simply refer to the standard in your data.
Metadata might include information like the source code, location, contributors, licence, version, identifier, references and how to cite the software (Jiménez 2017).