Skip to content

IDEA: Your list of files is actually a network of data to cooperate through, which the UI should reflect …

Notifications You must be signed in to change notification settings

steltenpower/FileFAIRy

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

72 Commits
 
 

Repository files navigation

FileFAIRy (or in Dutch: bestandenfee)

Long version of how I got to this idea(s): https://raw.githubusercontent.com/steltenpower/PowerGists/main/LCRDMsheets.pdf

First think of any list of files, e.g. Windows File Explorer, or what you get in DropBox, Google Drive, Microsoft Teams, OwnCloud (now used in SURF Research Drive), NextCloud (some day used in SURF Research Drive?), Figshare, Dataverse, DANS EASY and actually the file menu of many applications. Requested for functionality at Search Research Drive: https://servicedesk.surf.nl/jira/plugins/servlet/desk/portal/13/SD-85229

Then think of what step every character of FAIR makes possible:

  • Findable makes the bridge from search terms to addresses
  • Accessible makes the bridge from addresses to files
  • Interoperable makes the bridge from files to datasets you can technically and semantically work with
  • Re-usable makes the bridge from datasets you CAN work with to what is allowed with these datasets

See also Dutch table as too-easy explanation of FAIR data

Now look at the quick sketch here and below I'll explain how and why they differ more than you think. alt text

YES!, Rick has a minimal working example (and video) at https://github.com/rickdkk/fair-file-columns

DESIGN: The columns and menus (not all drawn yet)

  • OPEN SPEC: Is the file type of a file a so-called open specification? I used a green O for open (e.g. HTML), a dashed O for 'open' (e.g. OOXML, open by license, but seems not designed to make interoperability easy) and a red C for closed formats.

  • PREF: Is it a preferred/allowed format by parties (journals, research archives, anybody) you find important? I used the direction and color of thumbs to visualize. Clicking it pops up settings for (more) validation: spelling, linters, accessibility (a11y), link checkers, etc., etc.

  • VALID: What's the result of automatic validation of the file against the specification and schema? See goodtables.io, validator.w3.org and others.

  • ICON: for file types, not for the associated application. To help data-centric thinking of datacentricmanifesto.org, especially now filename extensions are often hidden by default nowadays. Clicking it pops up specific help options (colleagues supporting, open communities, convertors, etc.) This part was inspired by a researcher saying "What do you mean with file type? I use app X." He was really disappointed when I had to explain him vendor lock-in.

  • NAME: clicking it pops up all adresses, with protocols (DOI/HTTPS/webDAV/etc), this fileviewer is aware of that the file can be reached. Not drawn yet, to be added: a left arrow that reveals something about what links here and a right arrow for what this links to. Arrows are faded if there's nothing to reveal.

  • CONTEXT-menu (on the name): list of supporting applications (installed on machine, available as webapp, in institute appstore, or open source on the web)

  • USERS: files are usually shared over teams, roles, institutes, and our digital assistents. Who's in control, or who's it shared with and how?

  • DATE:

  • SIZE:

  • COMMITED: everything has a commit/version. We all GIT, right?

  • (not in drawing yet): MORE METADATA (DC and other vocabulaires) and based on all of the above:

  • (not in drawing yet): RELATED/SUGGESTIONS: Find related stuff and people to cooperate with (or compete against?).

  • add something like “phase”={active, published} and “usage licence”? Somehow linked to Users.

  • Also some file names give trouble with Windows, or OneDrive, or ...

  • Hints on what files are allowed to be shared where, for NDA, privacy, whatever reasons perhaps? Use SoLiD pods?

ON IMPLEMENTATION:

Seems DataVerse is at least somewhat interested: github issue (and also https://groups.google.com/g/dataverse-community/c/jMt-COccvo0 )

Put the suggestion towards ownCloud too: https://central.owncloud.org/t/how-to-add-new-defined-columns-to-lists-of-files/37834

Where to try to get information about file types:

On Windows File Explorer it could be realized by:

possible companion of webber

Was suggested for 4TU's repository here: 4TUResearchData/djehuty#6

Combinations with https://github.com/ITC-CRIB/fairly perhaps? That is a Python library / unix shell tool / Jupyter extension to communicate with data repositories and archives in a git-like manner.

Manager of SURF's researchDrive devteam liked the idea and mentioned a problem at a university that would likely would have been prevented by it. Said he would discuss it with the team, but warned that as they don't use a fork, but the general release of ownCloud, it might not be so likely to happen (there). Update: won't happen there. The part of ownCloud that this idea is aimed at: https://github.com/owncloud/core

Research Cloud actually makes a better fit SURF told me

The guy behind https://www.sen-labs.org/ and https://twitter.com/labs_sen said this idea is actually quite similar to what he's working on.

JupyterLab is interested (see Github issues)

complexity of filetype detection (extension is not enough): https://youtu.be/PBbld8xB2Bo

https://www.nationalarchives.gov.uk/pronom/BasicSearch/proBasicSearch.aspx?status=new

About

IDEA: Your list of files is actually a network of data to cooperate through, which the UI should reflect …

Topics

Resources

Stars

Watchers

Forks