DL.org All WGs Meetings, Rome, 26-28 May 2010 Quality Interoperability Approaches, case studies and open issues DL.org Quality Working Group Rome, 28 th May 2010
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality InteroperabilityApproaches, case studies and open issues
DL.org Quality Working Group
Rome, 28th May 2010
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality comprehensive models
Fuhr et al., 2001
Digital Libraries: A Generic Classification and Evaluation Scheme
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality comprehensive modelsGoncalves et al., 2006What is a good digital library? A quality model for digital libraries
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality comprehensive models
Zhang, 2010
Holistic DL evaluation
model
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality comprehensive modelsCandela et al., The DELOS RM Quality concept map, 2008Annotations by the DL.org Quality WG
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality
Quality is subjectiveQuality is dynamic
Quality is vagueQuality needs policies
…
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality WG meeting results
• Our motivating scenario: consider that representatives of two (or more) DLs have a round table to negotiate a service level agreement (SLA) defining their interoperability requirements and for this establish a quality threshold that each individual DL has to meet or exceed; “Quality” would provide transparent qualitative or quantitative parameters for defining the threshold
• Our approach is practical: Quality Interoperability Survey, Quality scenarios
• The Cookbook TOC• The interoperability scenarios 1&2
DL.org All WGs Meetings, Rome, 26-28 May 2010
Quality Interoperability Survey
• Survey Pilot questionnaire + results• Simplification and improvement• Disambiguation (Glossary)• Collection strategy• Data analysis and interpretation• Results expected by June 2010• Best practices
DL.org All WGs Meetings, Rome, 26-28 May 2010
The Cookbook
• Definitions of quality• Context: how the Quality WG faced the
investigation challenges• Consistent terminology
Introduction
DL.org All WGs Meetings, Rome, 26-28 May 2010
The CookbookLooking at Web solutions at a
technical level
FRONT END: Web Interfaces (usability, accessibility), eg. W3C
BACK END: Web services ontologies (QoS), eg. FIPA, WS-QoS, MOQ, DAML-QoS
DL.org All WGs Meetings, Rome, 26-28 May 2010
The CookbookQuality insurance procedures
for specific DL systemsSemantic and organisational
levels• Document repositories (eg. DINI, DRIVER)• Research data archives (eg. Data Seal of
Approval)• Preservation systems (eg. TRAC,
DRAMBORA)↓
Template for comparison within each class
DL.org All WGs Meetings, Rome, 26-28 May 2010
The Cookbook – Case studiesTemplate
Aspect DINI Certificate
DRIVER Guidelines
Explicit quality policy for protocol and metadata implementation Yes Yes
Explicit policy for operations (personell, support etc.) Yes No
Personal quality check (questionaire, on-site review) Yes No
Intellectual quality check (remote) Yes Yes
Automatic self validation No Yes
Organized through sustainable Organisation DINI COAR
Explicit branding when checked Yes No
Translation in English, Spanish, Portuguese, Japanese No Yes
Green and Gold Yes No
Strictly full-text oriented Yes Yes
DL.org All WGs Meetings, Rome, 26-28 May 2010
The CookbookBest practices from the
professional community
The results of the Survey will be included as best practices from the professional community. We are aware that quality is subjective, that we are dealing also with two “primitive” interoperability challenges
1. researchers vs professionals 2. different disciplines involvedBut we want to know from DLs people!
DL.org All WGs Meetings, Rome, 26-28 May 2010
The CookbookThe Quality WG checklist
Light-weight document – based on the Quality Core Model parameters - with practical
recommendations from the DL.org Quality WG based on the QCM parameters and the
Quality Interoperability Survey results
DL.org All WGs Meetings, Rome, 26-28 May 2010
Interoperability ScenariosScenario no. 1 – Quality and
Functionality issues Metadata Evaluation Formats quality Geographic referencing (Content) and related pinpointing
issues (Functionality) Controlled vocabulary + related Functionalities issues
(search, browsing, etc) Multilingualism Quality of the annotations (User profiling, User
authentication) Sources evaluation, i.e. Provenance
DL.org All WGs Meetings, Rome, 26-28 May 2010
Interoperability ScenariosScenario no. 1 – Policy issues
Copyright issues Reuse issues Content policy issues
Possible solutions: change the copyright policy, use creative commons license
Suggestions: review the scenario and make it consistent (logical incoherencies on map reproduction, restored video, 3D Model regeneration), less chaotic and more realistic
DL.org All WGs Meetings, Rome, 26-28 May 2010
Interoperability ScenariosScenario no. 2 – Quality issues
• Merged query results• Machine translations for supporting two languages• Articles quality (one has editorial reviews the other has not)• Mark-up of content• Metadata evaluation: Accuracy and Completeness• System performance• Technical correctness
DL.org All WGs Meetings, Rome, 26-28 May 2010
Interoperability ScenariosScenario no. 2 - Approaches
• Two localizations, one for each country, each supporting one language for content from both systems (very difficult!)• Functionalities of both source DLs (i.e. union of), including: access control,
search, results views • Metadata quality: solved selecting targets, customizing
search screen based on targets e.g. which common metadata fields they support• Search QoS: solution in the user
interface design to indicate how far along the search is and cluster the results
• Ownership of images: search results must contain author, photographer, agency and rights metadata so the user can see if they need to pay, who they pay, and what rights they get
DL.org All WGs Meetings, Rome, 26-28 May 2010
Some conclusive thoughts
• Quality: dynamic, subjective, not completely definable
• Provenance = the resource story = how to establish quality
• DL systems and their implementation teams will always struggle with quality, but users will have the last word