3. Describing research data (documentation, metadata)

At the core of data description, we have two basic terms, i.e. documentation and metadata. These terms may not necessarily have precise definitions on the everyday level and may be used somewhat differently in various guidelines. However, the purpose of both terms is to describe the origin, structure, contents, and treatment of the data. 

Data documentation is typically understood as explaining the research process during the research in such detail that it would be possible to repeat the research afterwards. It can be done, for example, by using a formal research laboratory book or an Excel sheet, where essential information of the data processing stages will be recorded. 

Metadata, on the other hand, is connected to, for example, administrative or descriptive information about the data.  Metadata tell where the data is stored, how the data can be used, whether there are any special conditions on the use of the data, or who can provide additional information, etc. 

In the context of open science, The Helsinki Term Bank for the Arts and Sciences defines metadata as describing data in summarised form (e.g., its contents, structure, context etc.) and states that metadata are particularly needed for the retrieval and use of data. 

In the context of data management, metadata often refers to standardized and structured information about data that explains, e.g. the origin, purpose, time reference, geographic location, creator, access conditions, and terms of use of a dataset. Metadata come in many forms, such as information of folder structure that keep all of research data organised and in a logical order. Different stages of the research may require different forms of metadata. In the experimental stage of research, information of the software or equipment used in data collection will be needed. In the publishing phase of the data, metadata related to sharing and access is needed, including, for example, information about licenses. 

Metadata of research data can be structured according to international standards, i.e. according to uniform and machine-readable ways of describing research data. A machine-readable document is a document whose content can be readily processed by computers. Using standards will make data more understandable, facilitate reuse, and enable combining datasets. The easiest way to apply metadata standards is to fill up a structured e-form that consists of different fields. 

There are several types of metadata standards available, such as the generic and commonly used Dublin Core (DC), a standard for geographic information (ISO standard 19115 ), and a standard covering the worldwide exchange and communication of date and time-related data (ISO standard 8601). Metadata standards also vary across research fields and disciplines, and researchers can find more information about disciplinary metadata standards from the website hosted by Digital Curation Centre or Research Data Alliance. 

Metadata can be stored in a free-form text file, i.e. as a README-type file in connection with the dataset. The README file can include, for example, basic information about the data, how the data has been collected and for what purpose, how and where the data is stored, or how the research has been conducted. 

The general information may include the following: 

• name of the data(set) 
• research field / discipline 
• principal investigator/author or person responsible for collecting the data (name, institution, contact information) 
• associate or co-investigators or people involved with collecting, processing, analysing and/or submitting the data (name, institution) 
• funding sources. 

The description of the research implementation may include the following information: 

• original purpose (information on the research project for which the data has been collected) 
• date of data collection (project start and end date) 
• description of the collected data and methods for data collection or generation 
• the size of datasets, and files that belong to the dataset (name, format, content) 
• a list of publications in which the data has been used and/or described 
• the modification of dataset: how the dataset has been modified (e.g. anonymisation, changes and edits have that been made to the data or data variables). 

Some examples of README-files with templates: 

• Harvard Biomedical Data Management
• Cornell Data Services

Standardized metadata for all kinds of research datasets regardless of the repository can be provided by using the Fairdata Qvain tool. Qvain allows you to publish the dataset descriptions, after which they are available through the Fairdata Etsin service.
The Qvain User Guide contains information e.g. on the minimum requirements for data descriptions. You can also watch Qvain instruction video by CSC.

The Etsin search service enables researchers to find metadata of research datasets that have been created with the use of Qvain but also datasets that have been distributed by other services. In other words, Etsin harvests (copies) metadata from external sources such as Finnish Social Science Data Archive or The Language Bank of Finland. Furthermore, the metadata from Etsin can be transferred to other services, such as UEF eRepository (eRepo). Therefore, it is important that all data produced in UEF is described in Etsin. 

Qvain and Etsin are part of the Finnish Fairdata.fi services that are introduced in chapter 1 of this learning material.  

High quality data are well-organised, structured, named, and versioned. Well-organised file names and folder structures make it easier to find and keep track of data files. 

Already at the beginning of your research project, you should design a folder structure that makes it easier to find information and come up with a consistent and informative way of naming files. You should use the naming system consistently throughout the research project. Write down the system you use and save it in a place where it is easy to find. This also contributes to the comprehensibility of the data and is an essential part of the description of your data. 

It is recommended to document the following information for each file: 

• file name 
• project abbreviation (e.g., acronyms) 
• file location (file path) 
• file size 
• file format 
• software used to create the file 
• date of creation 
• file creator 
• file version 
• access rights for the file. 

Tips for file naming: 

• create meaningful but brief names 
• use file names to classify broad types of files 
• avoid using spaces and special characters 
• if using a date in the file name, always state the date ‘back to front’: year, month, day (YYMMDD (e.g. 20200901) or YY-MM-DD (e.g. 2020-09-01 ) and/or hours, minutes, seconds (HHMMSS or HH:MM:SS) 
• the version number should be indicated in its file name (“V” for version followed by the version number, e.g. V02). 

Data description and metadata. Finnish Social Science Data Archive.

Siiri Fuchs, & Mari Elisa Kuusniemi. (2018). Making a research project understandable – Guide for data documentation (1.2). Zenodo.