Information quality in the context of CRIS and CERIF Maximilian Stempfhuber GESIS-IZ Social Science...
-
Upload
kelly-tucker -
Category
Documents
-
view
215 -
download
0
Transcript of Information quality in the context of CRIS and CERIF Maximilian Stempfhuber GESIS-IZ Social Science...
Information quality in the context of CRIS and CERIF
Maximilian StempfhuberGESIS-IZ Social Science Information CentreBonn, Germany
CRIS 2008, June 5-7, Maribor
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
205.06.2008
Agenda
• Information quality (IQ) and CRIS: Why bother?
• IQ in the context of (euro)CRIS– Code of Good Practice (CGP)– IQ coverage at CRIS 2002 – CRIS 2008
• IQ research: An overview• IQ and CRIS: Towards better integration• Conclusions
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
305.06.2008
Information quality (IQ) and CRIS: Why bother?
• CRIS vs. library catalogues, repositories, websites etc.Any difference concerning authority, completeness, correctness?
• Are CRISs meant to be of quality? Which?• Does the quality of CRIS contents influence its
use?• Are all CRISs the same (concerning quality)?• Networked CRISs: Just add up individual
quality?• Would CRISs and the CRIS community benefit
of a more explicit, comparable model of quality?
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
405.06.2008
IQ in the context of (euro)CRIS – Code of Good Practice
The CGP view to (information) quality: “fit for purpose”
“To ensure the continued use of a CRIS, it is necessary to provide additional value or benefits to both users and contributors to the system. This may be achieved by adhering to a quality plan which defines the accuracy, timeliness, data completeness, presentation of data to the end user, and the functionality offered by the search software.”CGP V3.0, page 14
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
505.06.2008
IQ in the context of (euro)CRIS – Code of Good PracticeRESPONSIBILITY PROCESS
RIS proposal
RISDesign
InformationProcessing
Reject
Reject
Reject
Accept
Accept
Accept
RTD-PromoterSteering Group
Development FunctionPublishing Function
RTD-PromoterSteering Group
Collection FunctionProduction FunctionMarketing Function
RTD-Promoter
ConceptAnyone
Project Manager
Reject
Accept
RTD-PromoterSteering Group
Publishing/Distribution
Maintenance
CONTROLS
RIS Proposal - Definition of Purpose - Identification of Users - Definition of Content
RIS Design Plan - Database Specification - Structure and Presentation - Classification and Indexing - Search and Navigation
RIS Information Processing Plan - Data Collection Plan - Collection Guidelines - Quality Control Plan - Acceptance Test Plan
Distribution PlanMarketing Plan (revisited)Economoic model (implementation)
Maintenance Plan /Acceptance
Concept Acceptance
RIS Proposal Acceptance
RIS Development Acceptance
Information processing acceptanceStructuredOutput
Ong
oing
Proc
ess
Rev
iew
Marketing Plan- Market Analysis- Cost benefit analysis/ economic model
Currently waterfall-like approach (one „big“ cycle)
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
605.06.2008
IQ in the context of (euro)CRIS – Code of Good PracticeQuality control Essential Important Nice to have
Ensure consistent structure of information
Ensure content is fit for purpose
Availability of usage statistics
Define quality tolerances with provider
Define traffic rules with provider (timescales for receipt and publication)
Ensure consistent presentation of information
Define consequences of not complying versus benefits of complying
Define standard terms, organisation names
Define mandatory fields/kernel information
Establish error correction procedures
Define sampling plan
Establish version control procedures
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
705.06.2008
IQ in the context of (euro)CRIS – Code of Good Practice / CERIF
Discussion• Waterfall model might not be adequate for
complex scenarios in or for which CRIS are designed
• “Fit for purpose” only one (specific) view to quality
• General problem: Abstract models (CPG) are hard to translate into actual systems in a deterministic way
• General problem: Detailed specifications (CERIF) do not guarantee systems meeting users’ demands
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
805.06.2008
Alternative models: Spiral model
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1005.06.2008
Alternative models: Model Driven Architecture (MDA)
Basic Idea: Separation of concerns(specification from implementation)
Provide a platform-independent model (PIM) for CRISsandtransformations to generateplatform-specific models (PSM)
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1105.06.2008
IQ in the context of (euro)CRIS – CRIS 2002 to 2008 topics
Summing up 16 of 69 papers from CRIS proceedings:
• Information quality: To improve on aspects like correctness, authoritative registers, controlled vocabularies, persistent identifiers, automatic checking of values and structure are used, and through intellectual processes carried out by experts the data is enriched to make it more useful or trustworthy. Semantic Web technologies are suggested to improve completeness of data (also across individual CRISs).
• Data integration: This becomes an issue as soon as data is exchanged or individual CRIS are networked. Methods employed are the certification of information systems, checking of data structures and values against formal requirements, mapping between vocabularies, and automatic and intellectual de-duplication.
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1205.06.2008
IQ in the context of (euro)CRIS – CRIS 2002 to 2008 topicsSumming up 16 of 69 papers from CRIS proceedings (cont.):
• Quality as a process: Checking data towards quality criteria as soon as it is created, using existing data to verify new data, and enabling feedback loops from users of data to incrementally improve overall data quality.
• Personalization: Better matching CRIS features (e.g. amount and level of detail of data, presentation of information, availability of features) to the specific demand of individual users or well defined user groups.
For a community of practice, there is not much concerning IQ that can be shared (practices, tools etc.)
Individual results are not generalized; hard (how?) to apply IQ has not the same coverage as data structures, data
exchange etc.
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1305.06.2008
What is (information) quality?
• Degree to which a set of inherent characteristics fulfills requirements (ISO 9000)
• Conformance to requirements (Philip B. Crosby)
• "Fitness for use". Fitness is defined by the customer. (Joseph M. Juran)
• The quality has two dimensions: "must-be quality" and "attractive quality“ (Noriaki Kano)
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1405.06.2008
What is (information) quality? (cont.)
IQ or data quality denotes the degree of relevance of information in relation to a specific context and information need:
• Requirements may be user specific or very general
• Total of all requirements towards information or information products ([information] process oriented view)
• Information that is fit for use by information consumers (user oriented view)
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1505.06.2008
IQ research: An overview
Alternative views to IQ• fit for purpose• exceptional view (quality as something special)• perfection (quality as a consistent or flawless
outcome)• value for money (quality in terms of return on
investment)• transformation (quality in terms of change from
one state to another) Harvey 1995
Question: Which views could contribute / support our approach to build and promote (the quality of) CRIS?
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1605.06.2008
IQ research: A framework for IQ assessment
Ge&Helfert 2007
Spelling errors, incorrect values, outdated dataViolation of domain constraints, company or government regulations
Inaccessible or insecure information,difficult to aggregate / transform
Information not based on facts,impartial view, hard to understand
Accurracy,
Timeliness,Completeness
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1705.06.2008
IQ research: PSP/IQ as an example for an IQ model
Product and Service Performance model for Information Quality (PSP/IQ)Kahn et al. 2002
Information meets standards of accuracy, completeness, and free-from-error
Indicates a process by which information consumers regularly receive information in a timely manner
Information product must be useful and relevant to the user’s needs
Information consumers can easily obtain and manipulate information that adds value to their task
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1805.06.2008
IQ and CRIS: Towards better integration
IQ and CRIS – what is missing?• A model (see PSP/IQ) expressing a common
approach to IQ, shared by the (euro)CRIS community
• Sets of well defined IQ dimensions, matched to the model and user / CRIS providers’ needs (there are over 180 already defined)
• Common IQ metrics, connected to IQ dimensions and applicable to “real” CRIS
• A shared understanding on how IQ dimensions influence each other
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
1905.06.2008
IQ and CRIS: Towards better integration
IQ and CRIS – what is missing? (cont.)• A set of transformation how to transfer the
CRIS IQ model to an individual CRIS (see MDA)
• Standardized ways for assessing IQ (measuring and creating IQ metrics)
• Tested methods for improving IQ • Formal ways for expressing IQ dimensions
at the record / attribute level and on the CRIS level
• Agreed on ways how to use information abot IQ
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
2005.06.2008
IQ and CRIS: What can we gain?
• Better and comprehensible service to users• Procedures for improving a CRIS during its
lifetime• Promotion of CRIS to external users for
critical tasks (research evaluation, strategic planning etc.)
• IQ as an incentive for researchers (providing information), sponsors and users of CRIS
• Higher IQ for networked CRISs
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
2105.06.2008
Conclusions
• IQ research offers formal models of assessing and improving IQ
• IQ in the context of (euro)CRIS currently has not the same role as data structures and semantics (CERIF)
• Formalizing IQ in the context of CRIS is a precondition of making CRIS a reliable source of information
• The quality of networked CRISs at the ERA level is depending on assessing, preserving and improving IQ
Information quality in the context of CRIS and CERIF
Maximilian Stempfhuber
2205.06.2008
Thank You!
Dr. Maximilian StempfhuberGESIS-IZ Social Science Information CentreLennéstr. 30, 53113 Bonn, [email protected]