Think like a Digital Curator
- 1. Tips for good research data management (RDM) Think like a Digital Curator Digital Curation Officers UCT Libraries, DLS University of Cape Town Erika Mias & Kayleigh Roos
- 2. Tips for good research data management (RDM) File management • File management practices help you identify, locate and use your data effectively • Good file management helps others to understand, collaborate and/or reuse your data effectively • Well managed files are: ○ distinguishable ○ easy to locate and browse ○ easy to collaborate with ○ easy to work with (open formats) Image Credit: Cliparts
- 3. File naming • Three criteria to assist with naming files: ○ Consistency ○ Organisation ○ Context • Elements to consider when naming files: ○ version numbers ○ creation / publication date ○ creator’s name / group name ○ content description ○ project number • Always consider scalability when naming files ○ e.g. 001 vs 01 • Don’t ○ use punctuation, or capital letters ○ use special characters or spaces • Do ○ replace full-stops with underscores ○ replace spaces with dashes ○ keep to YYYY-MM-DD date format ○ keep file names relevant and as short as possible Image credit: XKCD http://xkcd.com/1179/ Tips for good research data management (RDM)
- 4. Always record changes to your data files, even if it seems unnecessary! ● Don’t use the word “final” - instead, number or date versions ● Avoid using labels - eg. ‘draft’, ‘test’, ‘final’, ‘rev’, ‘corrected’, etc ● Indicate major version changes with: ○ YYYY-MM-DD_Title_Author_V1 ○ YYYY-MM-DD_Title_Author_V2 ● Indicate minor version changes with: ○ YYYY-MM-DD_Title_Author_V1-1 ○ YYYY-MM-DD_Title_Author_V1-2 REMEMBER… ● Avoid full-stops, so: ○ V1-2 instead of V1.2 File versioning Tips for good research data management (RDM) credit: PHD Comics 1) “A story told in filenames:” http://phdcomics.com/comics.php?f=1323 2) “Final”.doc http://phdcomics.com/comics.php?f=1531
- 5. ● File formats encode information about a file that enable it to be recognised by a computer program or application ● File formats are indicated in the filename by an extension that follows a full-stop ○ jpg, docx, pdf ● Proprietary vs Open file formats ○ Proprietary file formats can only be opened by the software used to create the file ○ Open file formats are openly available and can be recognised by a number of applications ○ Best to use open formats wherever possible, or convert to open formats and save alongside proprietary files. File formats Tips for good research data management (RDM)
- 6. ● File format obsolescence ○ Changes in technology ○ Updates of software ● Migration vs Normalisation ○ both involve converting files from one format to another (typically preservation-friendly, open formats) ○ Migration refers to the conversion of files when the file format is at risk of obsolescence ○ Normalisation is the practice of converting file formats upon acquisition or creation for long-term preservation ○ Always practice normalisation to ensure preservation of your data and avoid emergency migration! Resources: ● DPC - File formats and standards ● Stanford University - Best Practice for file formats ● Archivematica - Format Policy Registry ● DCC - Open source software and open standards ● Open Data Handbook - File formats File formats continued... Tips for good research data management (RDM)
- 7. ● Involves changing the actual data (not file format) ○ sensitive/personal/private data ■ de-identification, anonymisation ○ converting qualitative data into quantitative data ○ visualising quantitative data ■ numerical data to bar graphs/pie charts ● Data transformation enables further analysis of the data collected ● Data transformation also enables sharing and analysis of data that might not otherwise be ethical to share Resources: ● DataFirst Data transformation Tips for good research data management (RDM)
- 8. Tips for good research data management (RDM) Metadata Definitions... “A set of data that describes and gives information about other data” (Oxford Dictionaries). “Metadata is structured information that describes, explains, locates, or otherwise makes it easier to retrieve, use, or manage an information resource. Metadata is often called data about data or information about information” (NISO, 2004). Metadata is ‘Information about data’ that helps us to find, access, understand and use (or reuse) data Three types: ○ Descriptive ○ Administrative ○ Structural Resources: ● DCC list of disciplinary metadata ● UCT metadata entry guidelines NISO, 2017
- 9. ● Readme Files ○ Advantages of creating well structured Readme files: (See Daniel Beck’s excellent Readme checklist and presentation) ○ Helps the reader to identify, evaluate, use and engage with your project. ○ Increasing requirement of data repositories and funders to submit a readme file when depositing in a data repository. ● Codebooks and Laboratory Notebooks ○ Raw data such as these also need to be managed effectively and preserved where possible- NB for reproducibility. (Researcher interview: Shaun Bevan) Document your data while you’re creating it so that it is easy to understand and use later on... Documentation Tips for good research data management (RDM)
- 10. ● Storage ○ From the outset, think about how much storage you require ○ Think about who needs to access your data and how that affects your storage location ○ Include costs of storage in your DMP and funding applications ○ Network drives highly recommended for storing and accessing master copies ■ UCT eResearch data storage services ● Backup ○ Find out about your network provider’s backup services ○ Set up your own backup workflow… ■ daily / weekly / monthly ■ 2 - 3 copies, different locations ■ Incremental vs. Full ■ Cloud vs. Local ● Security ○ Who needs access? How will you control access? ○ Sensitive data and encryption researcher interview on file management and security: Natalia Calanzani Storing and backing up your data Tips for good research data management (RDM)
- 11. ● Research Data Management Tutorials ○ Mantra ○ Leeds University ○ Research Data Management and Sharing (Coursera) ● DMPonline tutorials ○ EUDAT presentation on writing a DMP ● Metadata schemas and disciplinary metadata ○ Overview of metadata types ○ Disciplinary metadata ○ Metadata standards ● Open Research ○ SPARC ○ Why Open Research ● Researcher Interviews ○ Odum Institute interviews researchers on why RDM is important. (On a less serious note : “A Data Sharing and Management Snafu in 3 Parts”) Other Resources Tips for good research data management (RDM)
- 12. ● RDM makes it easier to understand, preserve and work with your data ● RDM assists you with planning your research project in order to satisfy funder requirements. ● RDM increases the likelihood of reproducing your results and validating your research and eases the transition into a research project for new members or collaborators. ● Data transformation can lead to unforeseen uses in other research disciplines: new use cases = more citations and greater reach / recognition of your work Extra work!? Tips for good research data management (RDM) Source: Cliparts
- 13. ● Do I have to keep my data and where? ● Isn’t just backing up my files to an external hard drive enough? ● What will it cost to store my data? ● Who can advise me on meeting the requirements of my grant funder’s data policy? ● What research data services are we are currently offering at UCT? ● What are my responsibilities as a researcher to manage my data? ● What is the best and safest way to share data with my research collaborators during the research project? ● We have more than one funder for a collaborative research project, what are the implications of this in terms of policies around data storage and copyright? ● Does UCT have a data repository? ● Can you track how many times my data are downloaded? ● Can I get credit for publishing my data? FAQs Tips for good research data management (RDM)
- 14. Thank You Contact us: Erika Mias: erika.mias@uct.ac.za Kayleigh Lino: kayleigh.roos@uct.ac.za Follow us: Twitter: Digital@UCT Facebook: DigitalLibraryServices Tips for good research data management (RDM)