Handling the complexity of relevant data requires new techniques with regard to data access, visualization, perception, and interaction for innovative and successful strategies. As a response to increased graphics performance in computing technologies and Information Visualization, Card et al. developed the Information Visualization Reference Model. Due to further developments in Information Systems as well as Data Analysis and Knowledge Management Systems in recent years, this model has to be adapted for addressing the recent advancements. Thus, the hybridly re ned and extended IVIS4BigData Reference Model was derived from the original model to cover the new conditions of the represent situations with advanced visual interfaces providing opportunities for perceiving, managing, and interpreting Big Data analysis results to support insight and emerging knowledge generation. After deriving and qualitatively evaluating the conceptual IVIS4BigData model as well as deriving the competences with focus on supporting management functions, as key consumers of Big Data analysis in Business Intelligence scenarios, this research will address the modeling of end user empowerment to support distributed Big Data analysis in VREs to support insight and emerging knowledge generation, based on a functional system description for end users, domain experts, as well as for software architects.
The availability of data has changed dramatically over the past ten years. The wide distribution of web-enabled mobile devices and the evolution of web 2.0 technologies are contributing to a large amount of data (so-called Big Data) [15]. Due to the fact that, \we live in the Information Age" [29], cognitive e cient perception and interpretation of knowledge and information to uncover hidden patterns, unknown correlations, and other useful information within the huge amount of data (of a variety of types) [24] is a big challenge. \The process of optimally controlling ows of large-volume, high-velocity, heterogeneous and uncertain data" [20], for creating value from data (emergent knowledge) [20], will become one of the key factors in competition, underpinning new waves of productivity growth, innovation, and consumer surplus [22]. \The revolutionary potential" of the bene ts of Big Data technologies [27] and the use of scienti c methods in business / operational data analysis and problem solving for, e.g., managing scienti c or industrial enterprise operations in order to support staying innovative and competitive and being able to provide advanced customer-centric service delivery, has also been recognized by Industry [21]. 1.1
Distributed Systems, like, e.g., Hadoop [
The overall goal of this research is to develop a reference model that can support advanced visual user interfaces for distributed Big Data analysis in virtual labs to be used in e-Science, industrial research, and data science education. By enabling the dynamic ad-hoc de nition of new interdisciplinary research projects within advanced visual user interfaces supporting cognitive e ciency as well as user empowerment, the surrounding infrastructure will support the life cycle of VREs.
In this way, this research will create a visual user interface tool suite supporting a VRE platform infrastructure that can host Big Data analysis and corresponding research activities sharing distributed research resources (i.e., data, tools, and services) by adopting common existing open standards for access, analysis and visualization, realizing an ubiquitous collaborative workspace for researchers which is able to facilitate the research process and its Big Data analysis applications. 1 Education for Data Intensive Science to Open New Science Frontiers
For addressing and archiving the aforementioned goals, research questions were developed. In the following, gaps in current research are highlighted and for each gap an addressing research question is de ned. By working on those research questions, the gaps in research in the highlighted elds will be addressed in this research.
Research Question A: As a response to increased graphics performance in
computing technologies and information visualization, Card et al. [
Modern cloud technologies and distributed computing are leading to almost unlimited storage and computing performance. Moreover, usable access to complex and large amounts of data over several data sources require new techniques for accessing and visualizing data with innovative and successful strategies, at the border between automated data analysis and enterprise decision making [14]. Not in alignment in these new required techniques, the original IVIS Reference Model transforms data from a single data source on the left directly to a visual representation for the end user on the right, without a direct view and interaction possibility in the single process stages. Due to the aforementioned aspects, the original model by Card et al. does not cover the new conditions of the present situation. In order to close this gap in research with regard to the IVIS model, the following research question is used.
) How can an adaptation of the IVIS reference model look like, to cover the new conditions of the present situation with advanced visual interface opportunities for perceiving, managing, and interpreting distributed Big Data Analysis results to support insight in virtual labs to be used in e-Science, industrial research, and Data Science education?
Research Question B: For archiving an usable and sustainable implementation of the reference model in practice, a Service-Oriented Architecture should be designed. However, this SOA must ensure easy operability as well as a certain exibility for special accommodations by their customers. In addition, due to limited nancial scope of medium-sized enterprises, the operating costs of this SOA must be considered as well. Therefore, the SOA should be implemented with open source solutions. For an implementation of the adapted IVIS reference model generated by answering Research Question A, the following research question is used.
) How can a bookable and exible visual user interface SOA look like, for demonstrations and hands-on exercises for the identi ed eScience user stereotypes with special attention to the overall user experience to meet the users (students, graduates, as well as scholars, and practitioners) expectation and way-ofworking [11]?
Research Question C: \Users are increasingly willing and, indeed,
determined to shape a software they use to tailor it to their own needs" [
) How can a conceptual framework for the visual user interface, developed by answering Research Question B, look like, in which the users of the system become co-designers to give domain workers more independence from computer specialists?
Research Question D: Big Data Analysis is based on di erent perspectives and intentions. Deduced from this perspectives and intentions, there are di erent use cases and related user stereotypes that can be identi ed for performing Big Data analysis collaboratively within an organization, which are important to tailor the reference model as well as the visual user interface to their demands. In order to evaluate the reference model and the corresponding visual user interface generated by answering Research Questions A, B and C with real use cases scenarios and user stereotypes, the following research question is used.
) Which use cases scenarios and user stereotypes for applying Big Data analysis in Business Intelligence scenarios in industry can be identi ed?
Research Question E: Only an organizations' technical ability for Big Data analysis does not automatically result in human resources therefore. Within this research, based on the results of the research questions above, the competences of the new profession Data Scientist in designing, establishing, operating and customizing Big Data analysis infrastructures as well as extracting meaningful value from the Big Data analysis results, supposed to be derived and integrated into the EDISON Project.
) Which competences with focus on supporting management functions, as key consumers of Big Data analysis in Business Intelligence scenarios, are required to apply the theoretical model and its implementation developed in Research Questions A, B, and C successful in practice? 2
As illustrated in gure 2, Big Data Analysis supporting emerging knowledge
generation and innovation-oriented decision making [
As outlined in gure 3, in addition to the two competence elds technical understanding (Technical Competences ) and professional understanding of their customers contextual levels and perspectives (Management Functions ), Social Skills is an important third competence eld required in industry.
In this illustration, the disposal of these three elds of Data Scientist
competences clari es the importance of the Social Skills. Whereas the Technical
Competences as well as the Management Functions, derived from Lothar Gulicks
POSDCoRB 2 [16] management function model, are located at the outsides,
the Social Skills are the central element in this illustration, because social skills
like creativity, exibility, or eloquence are essential to combine the technical
understanding and the understanding of their customers contextual levels and
intentions.
2 Planning, Organizing, Sta ng, Directing, Coordinating, Reporting, and Budgeting
As a response to increased graphics performance in computing technologies and
information visualization, Card et al. [
In IVIS4BigData, the IVIS pipeline is segmented in a series of data transformations. Furthermore, due to the direct manipulative interaction between di erent user stereotypes within the single process stages and their adjustments and con gurations of the respective transformations by means of user-operated controls, each segment in the IVIS4BigData pipeline needs to support an interactive user empowerment work ow allowing to con gure the transformations and visualizations in the di erent phases. 2.3
For enabling end-users \to articulate incrementally the task at hand" [13], \the information provided in response to their problem-solving activities based on partial speci cations and constructions must assist users to re ne the de nition of their problem" [13]. To realize this interaction, represented with functional arrows at each process stage between the cross-functional Knowledge-Based Support layer and the corresponding layers above, Fischer's and Nakakoji's [13] multifaceted architecture, illustrated in gure 5, is utilized to derive a functional use case framework for the resulting user empowered con guration work ow use cases of the di erent IVIS4BigData process stages.
In this architecture model, ve central elements (speci cation, construction, argumentation base, catalog base, and semantics base) can be identi ed, \that assist end users to re ne the de nition of their problem in their problem-solving activities based on partial speci cations and constructions" [13].
Derived from an end users con guration perspective, a domain independent problem solving, i.e., user interface con guration process can be divided in three layers. The Design Creation layer contains the construction and speci cation components, that represent the interactive part of this process utilizing the three static components argumentation base, catalog base, and semantics base within the lowest Domain Knowledge layer. Moreover, the Feedback layer in the middle of this architecture represents the interactive user actions (critics, casebased reasoning, and simulation), that are initiated during the speci cation or construction process. In addition to this architectural illustration and to emphasize the importance of the construction and speci cation elements, Fischer and Nakakoji de ned a process based illustration of the whole design process, that is outlined in gure 6.
In this process, \starting with a vague design goal, designers go back and forth between the components in the environment" [13]. Thus, \a designer and the system cooperatively evolve a speci cation and a construction incrementally by utilizing the available information in an argumentation component and a catalog and feedback from a simulation component" [13]. As a result, a matching pair of speci cation and construction is the outcome.
Based on the described principles, a generic use case framework is de ned that covers all activities within a IVIS4BigData process stage. Outlined in gure 7, this framework will be the context for the corresponding IVIS4BigData use cases.
From a vertical perspective, the use case framework consists of three layers according to Fischer and Nakakoji [13]. Although it contains all elements of their multifaceted architecture, the layers di er from the constitutive model. Whereas the middle Domain Knowledge layer accord to the original model, Fischer and Nakakoji de ned a separate layer especially for feedback. Not in alignment to the original model, this framework utilizes the Application layer, that represents Fischer's and Nakakoji's Design Creation layer, also for feedback and adaptation. Moreover, to cover the new situation of distributed processing, an additional Persistency layer is de ned underneath both layers.
In more detail, the Application Layer, where all design speci cation and construction functions of the IVIS4BigData process stage work ow are located as well as the the functions to run prede ned work ows, is separated in three areas. Whereas the Semantic Representation and Knowledge Management area of this layer contains all activities to con gure the data, analysis, or visualization work ow models depending on the IVIS4BigData process stage, the central Integration and Analysis area includes all functions to execute the resulting interactive work ow of each stage. This central area also includes the central transformation of each IVIS4BigData process stage (Data Integration, Data Transformation, Visual Mapping, or View Transformation) within the use cases. Finally, the Visualization, Adaptation, and Simulation area includes all activities to support the construction and speci cation process. In combination with the End User Stereotype and the Domain Expert, where both of them interacting with the activities within these three areas, the Work ow layer represents the central element of this framework. In this way, this central layer combines all Design Creation and Feedback activities of Fischer's and Nakakoji's Multifaceted Architecture in a more detailed and e ective way.
The Domain Knowledge layer in the middle of this framework contains all data, analysis, or visualization catalogs and other types of domain knowledge that can be utilized within the IVIS4BigData process stages. In this way, this layer, based on its elements, will support the activities of the central Appl. layer with regard to the respective IVIS4BigData transformation of each process stage.
Finally, the Persistency layer contains the data sources as well as the phases of the IVIS4BigData pipeline and is responsible for the data persistence during the interactive transformation between two consecutive IVIS4BigData phases. As a result of the ability to utilize distributed architectures and cloud services for data storage as well as for data processing, this layer emphasizes the importance to manage and control the data during and after an IVIS4BigData process stage.
To support researchers and organizations maintaining research resources, a proofof-concept reference implementation of our IVIS4BigData model will be implemented, by developing an interoperable, cognitive e cient and user empowering VRE infrastructure. As illustrated in Figure 8, the infrastructure is based on providing and managing access though open standards and is materialized though existing open components procured from successful research projects dealing with resources at scale, and supported by their owners as project partners.
It is building on the concept of semantic integration and mediation and
corresponding mediator architectures to support information integration across
borders of scienti c knowledge domains. In this way digital research resources from
di erent scienti c disciplines can be mediated by means of semantic
integration of domain models on the level of the mediator and by means of domain
adaptation on the level of the corresponding wrappers. This means, information
integration concept is independent of media types and persistency platforms.
Moreover, the infrastructure builds on the concepts and service model of
Infrastructure as a Service (IaaS) Clouds exploiting its scalability, elasticity,
and accounting capabilities as de ned by NIST [23]. Finally, it will establish a
scienti c collaboration model within its VREs that is building on Computer
Supported Collaborative Work (CSCW) concepts of con guring, using,
and dissolving such VREs in a very controlled way that is at the same time very
intuitive, i.e. user-centered and focusing on cognitive e ciencies and attractive
user experiences. Thus, it lowers barriers of adoption by harmonizing these o
erings with a cloud infrastructure for service delivery and provides services for the
secure collaborative management of interdisciplinary research projects in VREs
with a comprehensive set of standard-based services, interfaces, and tools that
support their complete life cycle in a domain agnostic fashion.
4 Evaluation
After deriving the IVIS4BigData reference model which covers the new
conditions of the present situation with identifying advanced visual user interface
opportunities for perceiving, managing, and interpreting distributed Big Data
analysis results, the necessity to evaluate the reference model still existed. In
this context, a full day workshop on Road Mapping Infrastructures for
Advanced Visual Interfaces Supporting Big Data Applications in Virtual
Research Environments [
In addition to the qualitative evaluation user study, the resulting IVIS4BigData
proof of concept implementation has to be evaluated from a end user as well as
from an applicability perspective. Thus, as utilized in [18] and [28], a usability
study, where di erent types of end users rst interacting with the resulting proof
of concept implementation and then where interviewed about their experiences
is planned as well as an application coverage study, where the applicability of
di erent real world scenarios is evaluated.
5 Discussion and Outlook
After presenting [
However, what is still missing and can be considered as a remaining
challenge for this research for achieving an usable and sustainable implementation
of IVIS4BigData and its functional system description, is the implementation of
a conceptual Service-Oriented Architecture (SOA) that must ensure easy
operability as well as a certain exibility for special accommodations by their
customers. Moreover, after designing the architecture and to evaluate the generic
process structure framework and use case scenarios, the integration of certain
presented Advanced Big Data Applications of the road mapping activity in 2016
[
This research has been produced in context of the EDISON project [11], which has received funding from the EU Horizon 2020 research and innovation programme under grant agreement No 675419. However, this paper re ects only the author's view and the European Commission is not responsible for any use that may be made of the information it contains. 10. Carusi, A., Reimer, T.: Virtual research environment collaborative landscape study.
JISC, Bristol p. 106 (2010) 11. European Commission: Education for data intensive science to open new science frontiers. H2020-INFRASUPP-2015-1, Proposal number: 675419, Proposal acronym: EDISON (2015) 12. Fischer, G.: In defense of demassi cation: Empowering individuals. Human
Computer Interaction 9(1), 66{70 (1994) 13. Fischer, G., Nakakoji, K.: Beyond the macho approach of arti cial intelligence: empower human designers - do not replace them. Knowledge-Based Systems 5(1), 15 { 30 (1992) 14. Fraunhofer IGD: Visual business analytics (2015), http://www:igd:fraunhofer:de/ en/Institut/Abteilungen/Informationsvisualisierung-und-Visual-Analytics/ Visual-Business-Analytics, accessed: 2015-12-02 15. Freiknecht, J.: Big Data in der Praxis. Carl Hanser Verlag GmbH & Co. KG,
Munchen, Deutschland (2014) 16. Gulick, L.: Notes on the theory of organization. Classics of organization theory 3, 87{95 (1937) 17. Harris, H., Murphy, S., Vaisman, M.: Analyzing the Analyzers: An Introspective
Survey of Data Scientists and Their Work. O'Reilly Media, Inc. (2013) 18. Heer, J., Card, S.K., Landay, J.A.: Prefuse: a toolkit for interactive information visualization. In: Proceedings of the SIGCHI conference on Human factors in computing systems. pp. 421{430. ACM (2005) 19. Hemmje, M.L., Brocks, H., Becker, J. (eds.): Demand Of Data Science Skills &
Competences (Expert Roundtable) (Nov 2015) 20. Kaufmann, M.: Towards a reference model for big data management (2016), https: //ub-deposit:fernuni-hagen:de/receive/mir mods 00000583 21. Manieri, A., Demchenko, Y., Wiktorski, T., Brewer, S., Hemmje, M., Ferrari, T., Riestra, R., Frey, J.: Data science professional uncovered - how the edison project will contribute to a widely accepted pro le for data scientists (2015) 22. Manyika, J., Chui, M., Brown, B., Bughin, J., Dobbs, R., Roxburgh, C., Hung Byers, A.: Big data: The next frontier for innovation, competition, and productivity.
McKinsey Global Institute (May 2011) 23. NIST: The nist de nition of cloud computing. Recommendations of the National Institute of Standards and Technology (2011), http://csrc:nist:gov/publications/ nistpubs/800-145/SP800-145:pdf, accessed: 2016-01-05 24. Prajapati, V.: Big Data Analytics with R and Hadoop. Packt Publishing (2013) 25. Ryza, S., Laserson, U., Owen, S., Wills, J.: Advanced Analytics with Spark, vol. 1.
O'Reilly Media, Inc., Sebastopol, CA, USA (3 2015) 26. Upadhyay, S., Grant, R.: 5 data scientists who became ceos and are leading thriving companies, http://venturebeat:com/2013/12/03/5-data-scientists-whobecame-ceos-and-are-leading-thriving-companies/, accessed: 2015-10-30 27. Wang, W.: Big data, big challenges. In: Semantic Computing (ICSC), 2014 IEEE
International Conference on Semantic Computing. pp. 6{6 (June 2014) 28. Wongsuphasawat, K., Moritz, D., Anand, A., Mackinlay, J., Howe, B., Heer, J.: Voyager: Exploratory analysis via faceted browsing of visualization recommendations. IEEE transactions on visualization and comp. graphics (1), 649{658 (2016) 29. Wood, J., Andersson, T., Bachem, A., Best, C., Genova, F., Lopez, D.R., Los, W., Marinucci, M., Romary, L., Van de Sompel, H., Vigen, J., Wittenburg, P., Giaretta, D., Hudson, R.L.: Riding the wave: How europe can gain from the rising tide of scienti c data. nal report of the high level expert group on scienti c data; a submission to the european commission (2011)