How metadata helps the research process
Summary of the basic metadata concepts and how they can help
This page is still in the works. The guide (#51) could cover such topics as:
Metadata is to help find, reuse, understand everything. How-to-fair guide on the topic provides an intro, the FAIR cookbookis also not bad.
Possible metadata that are key to know about:
Metadata that helps findability - ideally all objects of a research process should have persistent identifiers (or PIDs) so that these can be easily found and be citable: In price statistics, some already exist/are possible, and others are not yet set up:
Researchers in the discipline can sign up to create an ORCID. This helps you be found and get fair recognition for your work.
Datasets published to data repositories like zenodo help mint DOIs. TBC how to handle proprietary datasets though (i.e. #46)
Papers in official journals have DOIs that the journal creates as part of the publication process. Ideally papers published as part of conference proceedings could also have DOIs (as many disciplines now do), however this isn’t yet done in price stats.
Code (i.e. the research compendium) is published in a way that mints a persistent identifier. Note that GitHub doesn’t mint a DOI but that may be okay for interim code and published code could be pushed to zenodo (which does).
Metadata that helps interoperability:
The descriptive and structural metadata (i.e. info about each dataset) is outlined in the catalogue – hence we aim to help solve some of this with the catalogue.
While not exclusive, we are trying to follow the basic dublin core
The way we define various things is as standard as possible so that its easy to use
The idea is that researchers (and their programs) can more easily understand open datasets they use for their research, understand them, etc.
Accessiblity:
- Ways to get the data is as simple as possible - say using download_zenodo (in R) to automate the downloading of data via its DOI
Metadata that helps reusability:
Knowing how datasets or code is licensed so that you know when and how to research it. We document this in the limitations section of the dataset record
Provenance is clear. Say a dataset is made available on zenodo - where it came from and how it was created/modified is clear.