1. Powerful access to qualitative data: What’s behind the UK QualiBank
Darren Bell
Data & Services Developer
UK Data Archive
IASSIST, Toronto June 2014

2. QualiBank project: rationale & aims
Provide enhanced access to key qualitative data via online data browsing and exploration: UK QualiBank
Based on existing metadata schemas and known technologies
Offer a mechanism for reliably citing data located in the system
Project includes large-scale digitisation of precious and undigitized materials
Maximise the impact from existing research and resource investments – demonstrate re-use

3. UK Data Service and its own needs
We have one of the largest qualitative data collections– over 350 data collections
A proportion of these have been digitised from older paper sources
Currently users find and download these from our website
Not so easy to find, but study documentation good
No searching within collections
No file manifest shown until download
It can be a bit of guess work!
Have Datacite DOIs; cannot reliably cite parts of data

4. Finding & accessing qualitative data
Search for “health” in our data catalogue, Discover
Retrieve catalogue record, e.g. SN 6124: Being a Doctor: a Sociological Analysis,
DDI 2.5 very limited for describing file content
View limited user guide
Web download as RTF bundle (46 transcripts)

5. Data listing

6. Download Zip of data and doc

7. Complex data collections
SN 5801: Concepts of Healthy Eating Food Research: Phases I and II,
293 interview transcripts; 73 diaries; 6 observation field notes
Not represented well at all in a DDI 2.X catalogue

8. Metadata demands for UK QualiBank
Explore data through a data journey
Find relevant extract, examine in context, cite
Link data to still and moving images, and other related research outputs
Some collections completely open
Demands highly structured and consistently marked-up data
Qualitative data requires object (file-level) descriptive metadata, e.g. interviews, audio-visual files, images
Use of common metadata elements enable federated catalogues across providers and borders

9. Description below the collection
DDI 2.5 for catalogue metadata
QuDEx schema for file level description: allows detailed identification of data objects:
Interview transcript or audio recording etc.
Descriptive categories at the object level, e.g. mime type, interview characteristics, interview setting
Relationship to another data object or part of data
Capacity to capture rich annotation of parts of data (e.g an extract)
Based on published QuDEx model in use (Schema at:
Object-level description = a lot of manual work!
Limited use of TEI schema for mark-up of textual data items

10. User expectations Search/browse for data Browse Search:
Search /faceted browse of data - text; image/PDF, audio
Browse
Faceted browse by categories:
Collection level, title, date and openess
Collection object: data type, interview characteristics, location
Search:
Display no. hits and minimal item metadata
Word in paragraph; thumbnail image/pdf; AV link
Context: other related objects,within system or external
Access full object
View data, key metadata and all related files and links
Get citation for part of data

11. System assumptions
BaseX for metadata storage; Java loading; Solr search
Data must be fully prepared on loading/publishing to the system. Data not ‘managed’ within the system
Mark-up, metadata, relationships all pre-defined
Pre-defined GUIDs to be used for citation (DOI + drilldown)
Cannot search audio-visual data content
Simple QuDEx metadata data entry tool created using SharePoint
Technologies for user interface use existing in-house systems, .NET
No download of data collection/subset - route to the UK Data Service
Citation of selected extract of text; user-annotation possible

12. UK QualiBank Dataflow

13. Digitisation of key data sources
Selectively digitize paper-based materials:
Original survey questionnaires
Open ended questions
Transcribed interviews
Handwritten field notes, essays
Diagrams
Photographs
Destination formats:
All text files treated as XML
Image files (photos and text) as PDF
Audio as mp3

14. QuDEx collection level metadata

15. Objects in collection metadata

16. Object relationships
Rich set of verbs available to define relationships between all objects
Converse verbs generated automatically:

17. QuDEx Category Schemes

18. Use of Text Encoding Initiative (TEI)
Minimal use of TEI tags, of massive profile
To denote structural mark-up
Headers, turn takers, paragraphs
Corrections, errors
Use of unique GUIDs to identify all QuDEx IDs:
Collection, Files, Paragraphs

19. School Leavers on the Isle of Sheppey

20. TEI XML: School Leavers on the Isle of Sheppey

21. Search interface - hits

22. Target page for an interview

23. Target references

24. Audio file target page

25. Citation mechanism
System allows extract/quotation level citation; 1 or more consecutive paragraphs
Citation object and citation format created on the fly – using GUIDS and system URI
URI resolves directly to the data extract
Some more sensitive collections are closed, so cannot resolve to data without login
Is related to our collection-level DOIs e.g /UKDA-SN

26. Contact details Darren Bell dbell@essex.ac.uk
Louise Corti
Agustina Martinez