Files and folders

The design of research datasets should be carefully considered at the outset of the project.

Dataset design varies by discipline, types of data and variables, medium of data, units of analysis, methodology, relationship between data elements, and whether or not the dataset is part of a series.

However, clear and consistent naming for folders, files, and variables, should be a strategy across all disciplines.

It also helps making research data findable, accessible, interoperable and re-usable (FAIR principles).

File naming

There are two essential starting points for your file naming strategy:

  • A file name is a principal identifier of a file
  • File naming strategy should be consistent in time and among people in a research team
Elements in a file name

According to the CESSDA Data Management Expert Guide, you should consider the following elements when developing a file naming strategy:

  • Version number;
  • Date of creation (date format should be YYYY-MM-DD);
  • Name of creator;
  • Description of content;
  • Name of research team/department associated with the data;
  • Publication date;
  • Project number.
Best practice

The CESSDA Data Management Expert Guide suggests also that best practice in naming files is to:

  • Create meaningful but brief names;
  • Use file names to classify types of files;
  • Avoid using spaces, dots and special characters (& or ? or !);
  • Use hyphens (-) or underscores (_) to separate elements in a file name;
  • Avoid very long file names;
  • Reserve the 3-letter file extension for application-specific codes of file format (e.g. .doc, .xls, .mov, .tif);
  • Include versioning of file names where appropriate.

Folder structure

The folder structure of research datasets (eg. hierarchical / horizontal) should be considered early in the project.

Qualitative datasets containing text, interviews, images &c. may require individual files for every element.

Versioning systems can be used to track changes to data andcode, eg. the GIT (global information tracker) version control resource.[^81]

The software format of the dataset should facilitate flexible use of the data. Scholars using one format during a research project, may consider a different format for preservation - taking into consideration open source accessibility. To avoid future file obsolescence, it is best to avoid proprietary formats, where possible. For example .rtf instead of .doc/.docx; .tif instead of .jpg; .flac instead of .mp3 &c. Details of how to submit dataset outputs to the EUI research repository, Cadmus, are in Section 6 below.