This is a demonstration project inspired by the BagIt standard for storage and network transfer of arbitrary digital content, including research data.
With research datasets represented as "Bags" with metadata in
the bag-info.txt file, and the pointers to the actual data
files on a download server listed in the fetch.txt file,
writing a repository server becomes relatively simple. This project
is an exploration into such a research data repository server.
At this moment it only includes a few examples with the BagIt metadata format; most of the examples here use various metadata schemas from other repositories. In the future I can imagine that tools could be implemented to convert the repository specific metadata (and manifest file list) to the BagIt specification.
The project demonstrates a research data repository by reusing some public datasets from the following research data repositories:
The web server implemented in this project hosts the metadata of the datasets (or "research data collections"). It consists of an overview page that lists all datasets, and an individual landing page for each dataset.
If this were a proper repository publishing the original datasets, then the DOI of the datasets shoudl directo to the corresponding landing page here.
Besides the metadata, the landing page contains (or should contain) links to the actual data files to download. The files themselves are not hosted on the same web server but could be on a FTP server, a WEBDAV server, an S3 server, etc.
At this moment there is no mechanism implemented nor procedure documented regarding the construction of datasets. Also the minting of DOIs is not part of the current efforts. It is assumed that a file with metadata is provided in JSON, YAML, TSV, CSV, or XML format, and that the actual files have been made available on a download server.
This project is implemented using Jekyll, a static website generator. To run it on your own computer, you should install Ruby and Gem, and run
gem install bundler jekyll
git clone https://github.com/robertoostenveld/bird.git
cd bird
bundle install
Subsequently, you can convert the markdown into html documents with
bundle exec jekyll serve --livereload --incremental
Since I am also running this site on Github pages, you may need to
edit the baseurl field in the _config.yml file. For a local
deployment it should be empty (""), for deployment on Github
it should correspond to the repository name ("bird") .