Research Data Management

Transcription

1 Introduction to Research Data Management Stefan Kühne, University Computing Center, Research and Development

2 Introduction to Research Data Management Stefan Kühne, University Computing Center, Research and Development

3 AGENDA Relevance/ Motivation Research Data Policy RDM Examples Systematic Data Management Research Data RD.. Research Data RDM.. Research Data Management 3

4 ABOUT ME Background in Computer Science Diploma in Computer Science Scientific assistant at Department Business Information Systems PhD thesis: Transformation Space for modeldriven service-oriented Integration Head of Research and Development at UCC since 2014 Research projects on big data, semantic data, linked open data Scientific Computing Research Data Management [Kühn09] 4

5 WHAT IS SO SPECIAL ABOUT RESEARCH DATA MANAGEMENT? It is about long-term preservation of data (10 years). Re-use of data in new contexts But it is expensive! It is about data publishing. Open Access But maybe we want to use our data (again) exclusively in the future. It is about reproducible research. Avoid scientific misbehaviors But it is time-consuming. 5

6 DIGITALISATION increased amount of data: digitisation of objects (books, maps, events, ), data streams, social networks, open data repositories, world wide interconnectedness of resources (data resources, computing resources, services, ) mobile computing machine learning techniques digitalisation digitisation 6

7 DOCUMENTS Bach digital BACH RESEARCH Passacaglia BWV 582 (transcript from J.S. Bach s brother) some research questions Which conclusions to the original sources can be derived from existing transcripts? What was where and from whom transcripted? By which work and/or compositions of J.S. Bach were his sons and other later musicians influenced?

8 STREAM DATA ANOMALY DETECTION IN WIND POWER PLANTS some research questions How unexpected behavioural changes can be detected? How anomalies can be classified into known failure categories? Which parts of the plants must be intensively monitored? Which preventive measures should be performed to avoid failures?

9 BIG DATA THE COMMON CRAWL ARCHIVE, open repository of web crawl data web pages are available as raw data, meta data, text data) 2.8 billion web pages (09/2018) 242 TByte uncompressed content common crawl web graph represents the linking structure of the web as graph data 886 m. nodes (07/2018) 5.4 bn. edges (07/2018) Some research questions prevalence of Web advertising etymologies of words characteristics of the Web (e.g. degree of cross-linking) link patterns (e.g. communities) long term development of subnets 9

10 from DATA SCIENCE to RESEARCH DATA MANAGEMENT

11 DATA SCIENCE traditional research approaches scientific findings based on observations scientific findings based on model development and theory building new options enabled by IT, e.g. HPC-based simulation of complex phenomena data-driven approaches exploration and analysis of huge amounts of data from different sources selection and limitation of data is affected by the analysis approach generation of new hypothesis stimulation of interdisciplinary collaborations 12

12 INTERDISCIPLINARY RESEARCH research question: How to treat obesity in a global context? Civilisation Diseases Examples: structured data of lifestyles, diseases, states of health Sources: questionnaires, examinations, images, audios, videos research question: Do limited biodiversity forsters infectious diseases? Globalisation Biodiversity Examples: data of global streams (assets, financial capital, cultural goods), data resources to cultural artifacts (e.g. digital humanities) Quellen: questionnaires, archives, text corpora, images, audios, videos research question: What is the impact of local conflicts to biodiversity? Examples: data of zur diversity of species, genetic diversity, ecosystems, process data (such as nutrient cycles, evolutions, carbon dioxide cylce, nitrogen cycles, biochemical time series of proteins) Sources: repositories of genetic data, databases of occurrences of species, microarrays, images, audios, videos 13

13 REQUIREMENTS OF A SYSTEMATIC MANAGEMENT trust in data security and safety reliable infrastructures data workflows (transformation, storage, communication) How can an integrated processing and analyzing of data from different sources be achieved? explicit and reproducible processes sufficient information to interpretate and reuse of data Independence of data producer and data user 14

14 VARIABILITIES IN REALISATION Integrated processing and analyzing of data from different sources organizational governance structures infrastructure resources human resources, new roles and job profiles Realisation is context-specific: interchange formats meta data data publishing services data curation services incentives 15

15 a closer look at RESEARCH DATA

16 WHAT IS (RESEARCH) DATA? {"id":"005a ","data":{"topleveldomai n":"com","url":"com.regulargeek","secondleveldomain":"reg ulargeek"},"meta":{"label":"host","graphs":["5aaae58dd8a2 821b48439b6c"]}}{"id":" ","data ":{"topleveldomain":"com","url":"com.casetech","secondlev eldomain":"casetech"},"meta":{"label":"host","graphs":["5a aae58dd8a2821b48439b6c"]}}{"id":"00d ","data":{"topLevelDomain":"com","url":"com.redmarkt echnologies","secondleveldomain":"redmarktechnologies"}, "meta":{"label":"host","graphs":["5aaae58dd8a2821b48439 b6c"]}}{"id":" ","data":{"topleve ldomain":"de","url":"de.henningschuerig","secondleveldom ain":"henningschuerig"},"meta":{"label":"host","graphs":["5 aaae58dd8a2821b48439b6c"]}}{"id":"01a34a ","data":{"hostLevelDomain":"gentoo","topLevelDomai n":"ru","url":"ru.theserverside.gentoo","secondleveldomain ":"theserverside"},"meta":{"label":"host","graphs":["5aaae5 8dd8a2821b48439b6c"]}}{"id":"01ec ","data":{"topleveldomain":"com","url":"com.stcinnovations ","secondleveldomain":"stcinnovations"},"meta":{"label":"h OST","graphs":["5aaae58dd8a2821b48439b6c"]}}{"id":" ","data":{"hostLevelDomain":"tecno metro","topleveldomain":"com","url":"com.blogspot.tecno metro","secondleveldomain":"blogspot"},"meta":{"label":"h OST","graphs":["5aaae58dd8a2821b48439b6c"]}}{"id":"0338 7a ","data":{"hostLevelDomain":"curac ao-akelijk","topleveldomain":"nl", "url":"nl.startpagina. cura 17

17 MEGAMODEL Technical Space (XML, DB, OWL, CSV, ) has long hair, has bushy tail, intension concept represents L0 L1 L2 Collie linguistic instance-of type Class linguistic instance-of extension is a Meta Class instance element represents Lassie linguistic instance-of Object linguistic instance-of element represents extension Language intension 18

18 RESEARCH DATA IS A REPRESENTATIVE FOR SOMETHING TO DESCRIBE OR A FACT FROM REALITY TO EXPLAIN SOMETHING research process (inspired by critical rationalism) generation of a research hypothesis evaluation of the hypothesis in reality for this purpose we use data as marks/traces of reality Data may be collected under certain research questions or even generated. Some data is used from a purely observational perspective, e.g., the use of process-generated data. Some data are "re-used" after being collected for analytical purposes other than scientific research (e.g. clinical data) Existing data collected under other research aspects are re-used in new contexts (secondary analysis). 19

19 DEFINITION Research data refers to all information (independent of form or presentation) needed to support or validate the development, results, observations or findings of a research project, including contextual information. Research data include all materials which are created in the course of academic work, including digitisation, records, source research, experiments, measurements, surveys and interviews. This includes software and code. Research data can take on several forms: during the lifespan of a research project, data can exist as gradations of raw data, processed data (including negative and inconclusive results), shared data, published data and Open Access published data, and with varying levels of access, including open data, restricted data and closed data. [Lear18a] 20

20 development and definition of RESEARCH DATA MANAGEMENT

21 POTENTIAL DISADVANTAGES OF PUBLISHING RESEARCH DATA Time and money required for preservation and publishing data Disadvantages in scientific competition (research data may be evaluated by others) Reputation-damaging detection of methodical inadequacies of the data analysis disadvantages in the economic use of research data legal problems due to the publication of data (violation of data protection or property rights, e.g. in the case of protected measuring methods) Misinterpretation by third parties 22

22 SYSTEMATIC HANDLING OF RESEARCH DATA local storage within institutes without specifications, guidelines vs. data-intensive sciences (e.g. climate research, meteorology) continuous increase in the amount of data to be stored and processed required new information systems for long-term preservation (mechanism and services for storage, backup, registration, deposit and retention) to support current and future access development was driven by economic considerations handling of the dynamic research data lifecycle (long-term interpretability) 23

23 INITIATIVES, RECOMMENDATIONS, GUIDELINES 1998/2013 German Research Foundation (DFG) Safeguarding Good Scientific Practice 2009 German Research Foundation Recommendations for secure storage and provision of digital research primary data 2010 Alliance of Science Organizations in Germany Principles for handling research data 2014 German Rectors' Conference (HRK) Management of research data - a key strategic challenge for university management 2015 German Research Foundation Guidelines for handling research data Recommendation 7 Protection and Storage of Primary Data: Primary data as the basis for publications should be kept on durable and secure media in the institution where they were created for ten years. 24

24 INITIATIVES, RECOMMENDATIONS, GUIDELINES 1998/2013 German Research Foundation (DFG) Safeguarding Good Scientific Practice 2009 German Research Foundation Recommendations for secure storage and provision of digital research primary data 2010 Alliance of Science Organizations in Germany Principles for handling research data 2014 German Rectors' Conference (HRK) Management of research data - a key strategic challenge for university management 2015 German Research Foundation Guidelines for handling research data 1. Subject-specific definition of primary data, e.g. level of aggregation 2. Subject-specific organisational concept 3. Standards 4. Storage of data is related to researcher 5. Open Access 6. Meta-data 7. Data quality 25

25 INITIATIVES, RECOMMENDATIONS, GUIDELINES 1998/2013 German Research Foundation (DFG) Safeguarding Good Scientific Practice 2009 German Research Foundation Recommendations for secure storage and provision of digital research primary data 2010 Alliance of Science Organizations in Germany Principles for handling research data 2014 German Rectors' Conference (HRK) Management of research data - a key strategic challenge for university management 2015 German Research Foundation Guidelines for handling research data Concretization of recommendations in the context of DFG funding programs + Research Data Management Plan + Provision/Publishing of RD + Long-term archiving 26

26 INITIATIVES, RECOMMENDATIONS, GUIDELINES 1998/2013 German Research Foundation (DFG) Safeguarding Good Scientific Practice 2009 German Research Foundation Recommendations for secure storage and provision of digital research primary data 2010 Alliance of Science Organizations in Germany Principles for handling research data 2014 German Rectors' Conference (HRK) Management of research data - a key strategic challenge for university management 2015 German Research Foundation Guidelines for handling research data + Scientific reputation + Education and qualification + Development of infrastructures 27

27 INITIATIVES, RECOMMENDATIONS, GUIDELINES 1998/2013 German Research Foundation (DFG) Safeguarding Good Scientific Practice 2009 German Research Foundation Recommendations for secure storage and provision of digital research primary data 2010 Alliance of Science Organizations in Germany Principles for handling research data 2014 German Rectors' Conference (HRK) Management of research data - a key strategic challenge for university management 2015 German Research Foundation Guidelines for handling research data Establishing RDM at Universities + Development guidelines + Establishing collaborations across organizations + Strengthening the information competence with training programs + Extension of institutional infrastructures 28

28 ADVANTAGES OF RESEARCH DATA MANAGEMENT supports the scientific progress getting access to financial resources (e.g. grants, funding) direct scientific reputation gain by publishing high quality research data increased visibility of the own research work recognition in the scientific community (fulfillment of social norms) opening up new opportunities for cooperation and publication (e.g. co-authoring with data users) scientific suggestions through collaborative work with the same data 29

29 RESEARCH DATA MANAGEMENT Research data management concerns the organisation of data, from its entry to the research cycle through to the dissemination and archiving of valuable results. Research Data Management is part of the research process, and aims to make the research process as efficient as possible, and meet expectations and requirements of the university, research funders, and legislation. [Leic18] describes the process including all methods and procedures of handling RD covers the complete lifecycle: includes planning, collecting/generating, processing, preservation and documenting of research data ensures controlled access, reuse, reproducibility and quality assurance of RD enables reproducibility and the reusability of scientific results requires the establishment of governance structures creates a requirement for interdisciplinary work and future project applications 30

30 KEY PERFORMANCE INDICATORS PREPARATION OF RDM Institutional policy Steering committee dedicated to RDM RDM Services Dedicated staff to RDM Job profiles dedicated to RDM Training sessions on RDM IMPLEMENTATION OF RDM Monitoring activities Review of the policy Engagement of disciplines within the research institution data sets stored data sets published persistent identifiers for published RD [Lear18b]] 31

31 the starting point RESEARCH DATA POLICY

32 RESEARCH DATA POLICY defines the principles of handling research data in a specific context context: interdisciplinary vs. disciplinary, institutional, journal within an institution a research data policy represents a starting point for RDM components of a research data policy preample definitions (researchers, research data, research data management, research data lifecycle) jurisdiction intellectual property rights principles of handling of research data responsibilities, rights, duties for researchers and for the research institution validity 33

33 A RESEARCH DATA POLICY DEFINES RESPONSIBILITIES different roles IT service provider scientific library research support legal services researcher storage systems for long term preservation legal compliance, legal and ethical issues handling of sensitive data creation of data management plans quality assurance 34

34 CONCLUSION OUTLOOK

35 CONCLUSION benefits and risks of Research Data Management for scientists offers potential for excellent, reproducible and sustainable research defines requirements for scientific work increases effort benefits and risks of Research Data Management for research institutes offers potential for increased visibility represents a requirement for interdisciplinary research (proposals in the future) requires new information services, infrastructure resources and consulting services 36

36 OUTLOOK Research Data Policy Systematic Data Management Relevance/ Motivation RDM Research Data Examples Publishing of Research Data Stephan Wünsche Thomas Riechert Research Data Lifecycle Marc Schönwiesner Toralf Kirsten Data Repositories, Research Infrastructures Gerhard Heyer Data Scientists Chris Armbruster Other aspects Panel discussion Data Management Plan Barbara Weiner 37

37 THANK YOU FOR YOUR ATTENTION! Dr. Stefan Kühne Universitätsrechenzentrum Augustusplatz 10, Leipzig T F kuehne@uni-leipzig.de

38 LITERATURE [AdWO10] Allianz der deutschen Wissenschaftsorganisationen: Grundsätze zum Umgang mit Forschungsdaten [BüHM10] Stephan Büttner, Hans-Christoph Hobohm, Lars Müller (ed.): Handbuch Forschungsdatenmanagement, BOCK + HERCHEN Verlag, Bad Honnef, 2011, ISBN [DFG09] Deutsche Forschungsgemeinschaft: Empfehlungen zur gesicherten Aufbewahrung und Bereitstellung digitaler Forschungsprimärdaten [DFG13] Deutsche Forschungsgemeinschaft: Sicherung guter wissenschaftlicher Praxis. ergänzte Auflage. 2013, [DFG15] Deutsche Forschungsgemeinschaft: Leitlinien zum Umgang mit Forschungsdaten [Helb17] Kerstin Helbig: Einführung in das Forschungsdatenmanagement, RDA-DE-Trainings-Workshop A1, 8. Juni 2017 [HRK14] Hochschulrektorenkonferenz: Management von Forschungsdaten eine zentrale strategische Herausforderung für Hochschulleitungen [Kühn09] Stefan Kühne: Ein Transformationsraum zur prozessmodellgetriebenen serviceorientierten Integration. Disseration, Universität Leipzig, [Lear18a] learn: Model Policy for Research Data Management (RDM) at Research Institutions/Institutes [Lear18b] learn: Introduction to LEARN KPIs [Leic18] University of Leicester: What is Research Data Management. 39