Modeling and Qualitative Evaluation of a Management Canvas

for Big Data Applications

Michael Kaufmann

, Tobias Eljasik-Swoboda

, Christian Nawroth

, Kevin Berwind

Marco Bornschlegl

and Matthias Hemmje

Lucerne School of Information Technology, Rotkreuz, Switzerland

Faculty of Mathematics and Computer Science, University of Hagen, Germany

Keywords: Data Management, Big Data, Reference Model, Project Management, Case Study, Pilot Application.

Abstract: A reference model for big data management is proposed, together with a methodology for business

enterprises to bootstrap big data projects. Similar to the business model canvas for marketing management,

the big data management (BDM) canvas is a template for developing new (or mapping existing) big data

applications, strategies and projects. It subdivides this task into meaningful fields of action. The BDM

canvas provides a visual chart that can be used in workshops iteratively to develop strategies for generating

value from data. It can also be used for project planning and project progress reporting. The canvas

instantiates a big data reference meta-model, the BDM cube, which provides its meta-structure. In addition

to developing and theorizing the proposed data management model, two case studies on pilot applications in

companies in Switzerland and Austria provide a qualitative evaluation of our approach. Using the insights

from expert feedback, we provide an outlook for further research.

1 INTRODUCTION

The digital age has fostered the data explosion in

which the global information capacity doubles every

3 years (Hilbert and López, 2011). With this speed

of growth, a data intelligence gap is created: The big

data available to an organization is growing

exponentially, while the percentage of the data that

an organization can process and actually use

declines as rapidly (Zikopoulos and Eaton, 2011).

Ultimately, this percentage could become

infinitesimally small.

It is said that big data is the oil of the 21

century

and, thus, those individuals, companies, and even

nations who possess the skills to turn raw data into

something valuable will have major competitive

advantages. Therefore, there is pressure for

enterprises to adapt and implement a big data

management (BDM) strategy, even for companies

that are not experienced in this field. However, often

the question is not how to implement scalable

architectures, but how to get started with big data

management in the first place – especially in non-

technical companies. Therefore, our research

question guiding the investigation presented in this

paper is the following: How can the development of

new big data applications (BDA) be facilitated for

non-technical decision makers? (RQ1)

Bootstrapping new big data projects from scratch

is a formidable task. Reference models can help to

analyze and subdivide this process to reduce its

complexity and to provide a frame of reference and

guidance. To provide a possible answer to the

research question, the authors propose a new

framework for the management of big data projects

entitled “Big Data Management Canvas”. The

proposed framework extends the existing NIST Big

Data Interoperability Framework to make it more

actionable by providing a frame of reference for

extracting value from big data, called “data

effectuation”. This is accomplished by a knowledge-

based embedding of big data management in a frame

called “data intelligence” and by aligning technical

aspects of big data with business aspects.

The working hypothesis is that this model

accurately modularizes BDM and that it is useful

and valuable for companies for developing new big

data strategies. Our research methodology follows

design-oriented information systems research

Kaufmann, M., Eljasik-Swoboda, T., Nawroth, C., Berwind, K., Bornschlegl, M. and Hemmje, M.

Modeling and Qualitative Evaluation of a Management Canvas for Big Data Applications.

DOI: 10.5220/0006397101490156

In Proceedings of the 6th International Conference on Data Science, Technology and Applications (DATA 2017), pages 149-156

ISBN: 978-989-758-255-4

149

(Österle et al., 2010). In this paper, we present a

completed research cycle of problem analysis,

artifact design, evaluation, and diffusion. We present

our findings of evaluating our hypothesis

empirically and qualitatively in the context of two

pilot applications with large companies in

Switzerland and Austria.

2 STATE OF THE ART

2.1 Big Data Reference Models

Three new dimensions of data management became

apparent in the beginning of the 21

century:

“Volume, Velocity and Variety” (Laney, 2001). For

this, Gartner later coined the term “big data”

(Gartner, 2012). Since then, there have been

countless alternative definitions of the concept. In

2015, the definition of big data was standardized by

the NIST Big Data Public Working Group: “Big

Data consists of extensive datasets primarily in the

characteristics of volume, variety, velocity, and/or

variability that require a scalable architecture for

efficient storage, manipulation, and analysis”

(Chang, 2015a, p. 5). The NIST has analyzed the

existing state of the art in big data architectures and

models (Chang, 2015b). This analysis refers to big

data reference models from several organizations,

including ET Strategies, Microsoft, University of

Amsterdam, IBM, Oracle, EMC/Pivotal, SAP,

9Sight Consulting, and Lexis Nexis. Based on this

analysis, the NIST defined the NIST Big Data

Interoperability Framework (Chang, 2015c), a

standardized reference model for big data

applications (BDA). This reference model provides

five layers of activities for big data management. By

following these activities, value is generated from

data. These activities are listed in ascending value in

the information value chain: (1) data collection, (2)

data preparation, (3) data analytics, (4) data

visualization, and (5) access for data consumers.

2.2 Project Management

The reference model discussed in the previous

section is very descriptive, but not actionable

enough to be directly applied. To provide actionable

reference models, these can be linked to business

and project management methods. For example, the

business model canvas (BMC) (Osterwalder and

Pigneur, 2010) is a method to generate, and optimize

business models by dividing them into nine central

areas of interest. These areas are customer segments,

customer relationships, channels, value proposition,

revenue streams, key activities, key resources, key

partners, and cost structure. These fields interact

with each other, having the strongest influence on

directly adjacent areas. Osterwalder and Pigneur

provide example questions for every area of interest

within the BMC. Examples include “for what type of

market (e.g., mass market, niche market, segmented

market, diversified market, business customers,

private customers…) do we create value?”, “What

service are our customers ready to pay for?” etc.

Scrum (Schwaber, 2004) is an agile project

management method intended for software

development that accommodates shifting

requirements. One of Scrum’s core concepts is the

user story that encapsulates the properties of a

software product in sentences following the

following pattern: “As a <end user role>, I want

<the desire> so that <the rationale>”. The desire

coded into a user story describes a specific business

value.

In our approach, we have linked our reference

model with these two management methods to

enhance its actionability.

2.3 Big Data Management

In the past, data management (Mosley, 2008), has

been understood as an administrative information

technology (IT) task. However, in the age of

digitalization, big data management—as we

understand it—concerns a completely different level

of organization. To create value from big data, both

IT and business aspects need to be considered.

Therefore, it is important to shift data management

conceptually and culturally from mere

administration and governance within the IT

department to the overall valuation and effectuation

of big data on the executive level in accordance with

business goals.

With a v for value, the 5v model of big data by

Demchenko et al., (2013), in contrast to many other

definitions, poses a value question for big data

theory. Managing big data is not an end for itself; it

is more than an update of what was called “data

management,” with more volume, velocity, and

variety. Successful big data management creates

value in the real world, based on the ubiquitous,

omnipresent and ever-growing ocean of data in the

digital universe. A reference model for big data

management should facilitate the generation of value

from available data. Therefore, data is processed to

generate intelligence that supports data-driven

decision-making (Provost and Fawcett, 2013).

DATA 2017 - 6th International Conference on Data Science, Technology and Applications

150

However, the other direction is at least as important:

The application of intelligence to generate new big

data applications.

The term management is defined in the Oxford

dictionary as “the process of dealing with or

controlling things or people”. Based on that, in

combination with the 5v model of big data, we

define BDM as the process of controlling flows of

large-volume, high-velocity, heterogeneous, and/or

uncertain data to create value.

2.4 Contribution

The aim of the model proposed in this paper is to

provide an actionable frame of reference for creating

value from big data. The existing NIST reference

model discussed in Section 2.1 lacks three important

aspects presented in Sections 2.2 and 2.3. First, a big

data management reference model needs to be linked

to management methods to make it actionable

(Argyris, 1996) and to enable the operationalization

of big data in practice. Second, it should specifically

set the generation of value (Davenport, 2013) as the

primary goal of big data applications. This does not

necessarily mean monetary value; however, any big

data application should provide value to anyone or

else they become ends in themselves. Third, it

should address the bi-directional process of

intelligence (Floridi, 2012) as an important aspect of

data application: the knowledge and skills needed

for big data operationalization, as well as the

knowledge and skills generated by it.

3 A REFERENCE MODEL FOR

BIG DATA MANAGEMENT

The model proposed in this section analyzes BDM,

the process of creating value from big data, into

smaller fields of action to handle its complexity.

Using a constructivist epistemological approach to

business intelligence as a cognitive system,

(Kaufmann, 2016) identified, as a hypothesis, six

general aspects (or layers) of BDM, namely

datafication, data integration, data analytics, data

interaction, and data effectuation, as well as the

successful management and engineering of the

emergent knowledge in this process, which can be

called data intelligence. To create value, iterative

cycles from datafication to effectuation are

performed with a closed feedback loop and

intelligent human control.

This division of BDM into six layers, as shown

in Figure 1, is a meta-model, where more specific

BDM models represent instances implementing

certain aspects of the six layers. The purpose of this

meta-model is twofold: It can be used for classifying

and extending existing specific BDM and it can be

an orientation to derive new BDM models for big

data projects. Therefore, the model shown in Figure

1 is entitled “BDM cube”, which stands for Big Data

Management Meta Model and, hence, the M cube,

the third power of M, in the name.

Figure 1: The proposed big data management cube

provides layers of abstraction as a meta-model.

3.1 Big Data Management Canvas

Big data processing information systems should be

aligned toward the generation of knowledge and

value. From expert feedback, we know that business

/ IT alignment (Luftman and Brier, 1999) is most

important for successful BDM projects. Therefore,

Figure 2 describes business aspects, as well as

information technology (IT) aspects, for the

implementation and application of each layer in the

management model.

Analogous to the business model canvas

(Osterwalder and Pigneur, 2010), this model can be

plotted onto an actual canvas and used in

management workshops to develop big data

strategies and applications. Therefore, it is entitled

Big Data Management Canvas. The fields on the

canvas are addressed by its BDM cube layer

(numbers 1-5) and by positioning it as a business or

IT question (letters A-B). On top, data intelligence

supports the whole process. Each field contains a

title and a question that didactically guides the

Modeling and Qualitative Evaluation of a Management Canvas for Big Data Applications

151

Figure 2: The proposed big data management (BDM) canvas provides fields of action for planning big data applications.

canvas users toward productive thinking within that

field of action. In addition, for bootstrapping ideas,

every field provides an example from the Migros (a

Swiss supermarket) big data application project case

study published by Gügi and Zimmermann (2016).

By applying the canvas in workshops, big data

applications can be planned and documented by

pinning or sticking requirements, visions, plans,

tasks, and other relevant information written on

cards to the corresponding fields on the canvas. This

method can be applied in project management for

requirements engineering and status reporting. The

application direction is shown with arrows in Figure

2. Planning new big data applications should start

with their intended business value and applying

processes before going into technical details,

following the arrows in a counterclockwise direction

and filling it with current versus target state. The

following paragraphs define the canvas fields in

detail.

Data intelligence refers to the competence of an

organization to acquire and apply knowledge and

skills for big data management. This can be

understood as the management and engineering of

intelligence for all steps of the data-to-knowledge

pipeline (Abadi et al., 2016). Data intelligence is a

knowledge-driven, cross-platform function that

ensures that data assets can be optimally deployed,

distributed, and used over all layers of big data

management. This includes the proper establishment

of necessary basic conditions, as well as setting up

and developing technological infrastructures, know-

how, and resources.

1. Datafication is the capture of real-world

signals (1A) in the form of data sensors (1B). In the

case that relevant analytic data is not yet available,

new data can be generated by datafying physical

metrics, as well as user input.

2. Data integration is the combination of existing

analytic data (2A) from different business

applications into a single platform with consistent

access. Interfaces to data sources, big data

processing systems—as well as database

management systems—form an integrated database

(2B) for analytics. Here, special care must be taken

for scalability regarding the big data characteristics

of volume, velocity, and variety.

3. Data analytics is the transformation of raw

data into usable information. (OECD, 2017). In this

step, analytic processes apply data science (3A)

methods to the integrated database. This is defined

by NIST (2015a, p. 7) as “the extraction of

actionable knowledge directly from data through a

process of discovery, or hypothesis formulation and

hypothesis testing”. With respect to big data, a

scalable analytic platform (3B) is implemented to

DATA 2017 - 6th International Conference on Data Science, Technology and Applications

152

deliver statistical and machine learning tools that

operate on a parallel computing infrastructure.

4. Data interaction consists of mutual

interferences of data analytics and applying

processes (4A) that use the information resulting

from data analysis. At this point, user interfaces

(4B) enable the interaction of data analytics results

and the socio-technical organization.

5. Data effectuation means the utilization of the

data analysis results for value creation (5A) in the

products, services, and operations of the

organization. Analytics generate predictive signals

that enable data feedforward (5B) that, in contrast to

feedback, helps prevent disturbances and increases

system stability proactively in advance. This is

achieved by loading data analytics results into the

productive process of the organization.

4 PILOT APPLICATIONS

4.1 Case Study: BDM Canvas for Big

Data Management Strategy

EKZ is the power utility company of Zurich. It is a

public institution that serves one million customers.

EKZ employs 1,400 people and has a balance sheet

of 2 billion euros. Because of the possible

liberalization of the energy market in Switzerland,

EKZ needs to adapt to new retail strategies to gain

new customers. To do so, EKZ is designing a digital

strategy that entails the processing and utilization of

big data. To bootstrap a big data strategy, a

workshop was held to evaluate the current state of

BDM at EKZ and to sketch a vision of BDM goals.

The BDM canvas was utilized as a pilot

application in a big data strategy workshop. Three IT

managers, two product managers, and two marketing

specialists were involved. In the beginning, the

participants were introduced to the BDM cube meta-

model and to the BDM canvas. Each field was

explained in detail. Then, red and yellow Post-it

Notes were distributed. The workshop participants

were asked to write down the actual states of BDM

at their enterprise on yellow Post-its and target states

of BDM on red Post-its. During this creative phase,

a participant asked for specific examples because the

names of the BDM canvas fields are rather abstract.

Here, the Migros case study (Gügi and

Zimmermann, 2016) was used as an example to

explain the fields to the participants while they were

writing down their ideas. After 15 minutes, the

participants were asked to share their ideas and to

stick the Post-its on the canvas to the corresponding

field. One by one, each participant went to the

canvas and explained his or her thoughts. So, the

canvas was filled with insights and plans. Then, the

group gathered around the canvas and each item was

reviewed. Some items had to be moved to a better

corresponding field. In the big data strategy

workshop, the following insights were generated.

Data Intelligence. There were many inputs in the

workshop concerning knowledge gaps about basic

conditions that need to be clarified to enable

successful big data applications. Clear business

goals regarding big data management are needed,

with legal and compliance aspects of data protection

to be clarified first. The data culture must change

from departmental silos to enterprise-wide data

integration. The long-term goal is to enable a closed

loop between the target group definition and the

campaign outcome for continuous optimization.

1. Datafication. As real-world signals (1A), the

behavior of the customers on the company’s website

and the customers’ energy consumption over time

were identified as most relevant. The company has a

diversified business next to the core business that

can also be interesting. For the data sensors (1B),

the smart meter is an energy usage sensor that

delivers the energy consumption data of individual

customers in fixed intervals of 15 minutes to the

power utility company. The goal is to provide real-

time data from sensors in the customers’ homes

based on internet of things (IoT). That allows to

create a timeline of energy consumption for every

customer as a base for analytic applications.

2. Integration. Analytic data (2A) on cost

centers’ credit-worthiness is available from the

enterprise resource planning (ERP) application.

There are several customer relationship management

(CRM) applications with different customer

numbers that need to be harmonized. The goal is to

make geographical data, Google Analytics data, all

ERP data, user data from the website, and IoT

power-usage data available for analytics. It has been

established in the workshop that there is a need for

an integrated database (2B). Analytic data in the

company at the time of the workshop were not

optimally integrated. There is a data warehouse, but

it integrates only a part of the data needed for

analytics; CRM systems and customer information

needs to be consolidated and integrated. All relevant

data from all online transactional processing (OLTP)

and online analytical processing (OLAP) platforms

need to be combined and integrated for analytics.

3. Analytics.

The main goals for data science

(3A) at EKZ are to analyze the smart meter time

series to predict energy consumption and to develop

Modeling and Qualitative Evaluation of a Management Canvas for Big Data Applications

153

new products; to predict credit-worthiness and

customer value; and to predict cross-selling

potential. However, there are no data scientists

inside the company. The internal resources need to

be built up. Concerning the analytics platform (3B),

all IT is outsourced to a service provider company.

There is no know-how regarding analytics tools and

platforms. This know-how needs to be established,

especially regarding customer segmentation and

campaigns.

4. Interaction. Two applying processes (4A) that

interact with analytics results were identified: energy

network coordination and personalized marketing.

Regarding the user interfaces necessary to interact

with the analytics results, no inputs were given at the

workshop.

5. Effectuation. The intended big data application

at EKZ should provide value creation (5A) by

reducing the cost per order; by improving cross-

selling; by preventing credit default and losses; and

by accurately meeting energy demands. As data

feedforward (5B), analytics results could be loaded

into operational systems to enable dynamic pricing

based on predictions to balance energy loads in the

network.

The application of the BDM canvas at EKZ

helped to capture the current state of BDM in the

organization; it also helped to organize the target

state of BDM to bootstrap big data applications. The

canvas, as a frame of reference, primed the

discussion in the workshop in a productive direction.

Many of EKZ’s requirements were about basic

conditions concerning data intelligence, such as

building up data science know-how, business goals,

and compliance. Therefore, the field for data

intelligence was very important, especially in the

beginning, but the fields were not well understood

by the end users. The model was too abstract to be

applied without coaching by a human expert.

4.2 Case Study: BDM Canvas for

Project Management

“Silberkredit” Bank is a major financial service

provider in Austria. To provide anonymity, the name

of the company has been changed by the authors.

Silberkredit employs approximately 1,500 people,

amounting to a balance sheet total of about 30

billion euro. The bank started a proof-of-concept

project to evaluate possible applications of big data.

Within this proof of concept, Silberkredit applied the

BDM canvas as a big data project planning and

reporting methodology. To do so, it was combined

with the Business Model Canvas (BMC)

(Osterwalder and Pigneur, 2010) and Scrum

(Schwaber, 2004). Initially, Scrum user stories were

positioned on the canvas to generate a product

backlog with ideas to optimize the business value of

dats, for example: “As a product manager, I want to

know more about the segmentation of my market so

that I can propose better offerings” or “As account

manager, I want to know what my customers are

really ready to pay for to optimize my individual

value proposition”. Other user stories centered on

key activities, such as cost savings in generating the

necessary reports for the financial market authority.

These requirements were elaborated upon in a

brainstorming fashion among an interdisciplinary

group of professionals so that no idea was

prematurely eliminated.

Once the requirements were collected, they were

filtered to select the relevant ones. The user stories

were assessed and selectively removed from the

product backlog. A possible reason to remove a user

story was if it did not directly address data

management issues; for example, adding a non-data-

based functionality into the online banking portal.

Another reason to remove a user story was if it

regarded a small benefit / cost ratio. The remaining

user stories were positioned on the BDM canvas:

The user stories’ desires, representing target

business value, were pinned to the field value

creation (5A). The corresponding end user roles

were placed in the field applying processes (4A).

The BDM canvas was then filled with

corresponding tasks in a counterclockwise direction.

First, consumers of the analytics results for the field

applying processes were derived from the value

creation entries. Second, the abstract analytical

methods necessary to generate the required insights

were developed from the existing entries to generate

tasks in the field of data science (3A). This was

followed up by an assessment of the data required to

apply these analytical methods in the field analytic

data (2A). If the required data would not have been

available within existing data sources, this data

could have been captured in the field real-world

signals (1A); however, that was not the case. These

steps were made with a strict business perspective in

a sense of what results were pursued and which data

content was used. This was followed up by a

corresponding decuction of the technical

implementation. First, no need for data sensors (1B)

was identified. In the next step, the field integrated

database (2B) was analyzed in three ways: Existing

data systems containing the data listed in field 2A

were identified; the Hadoop distributed File System

(HDFS) was identified as the central data storage

DATA 2017 - 6th International Conference on Data Science, Technology and Applications

154

technology and for data integration; and interfaces to

link analytic data from field 2A with this central

integrated database (2B) were selected. In the next

step, software, tools, and algorithms for the field

analytics platform (3B) were selected by assessing

the needs identified in field 3A and compatibility

with the data storage technology chosen in field 2B.

This information was used in the field user

interfaces (4B) to identify the interfaces for the

interaction of consumers with the data analytics

results. The last step was the identification of the

interfaces to feed forward the data analytics results

into the productive process of the organization.

These results were listed in the field data

feedforward (5B) of the BDM canvas.

This planning was manually performed by using

a large paper canvas and Post-its. While collecting

the canvas entries, the number identifying the

corresponding user story was written on a Post-it.

The results were collected in an Excel spreadsheet.

In analogy to the Scrum story points, the entries

were named BDM points and numbered for future

identification. This way, every user story consisted

of multiple BDM points. In the next step, the

required effort for the implementation of every

BDM point was assessed by using the scrum

methodology, and the overall estimate for every user

story was identified. Based on that, the Scrum

methodology of release- and sprint-planning was

applied to plot the technical implementation. Status

reports were generated using Scrum “burndown

charts” and “traffic light reports” for every field of

the canvas.

The application of the BDM canvas at

Silberkredit supported the engineering requirements

for big data applications by aligning Scrum user

stories with specific fields of BDM. Also, it

supported a frame for progress reporting for the big

data project. The canvas fields gave Silberkredit

information about how to subdivide the giant task of

building new big data applications into clearly

manageable areas, thus reducing entropy in the

bootstrapping phase. The datafication aspect was not

used at Silberkredit because it turned out that the

required data was already largely available in

existing data sources. In the Silberkredit case study,

the direction of application started with the business

goals of the big data application and deduced the

technical implementation counterclockwise step by

step. It was important for Silberkredit to align big

data management with a clear understanding of how

value will be generated and to go into technicalities

only in a second step. However, the BDM canvas

method was applied manually. There is a lot of

potential for business process automation using the

BDM canvas.

5 DISCUSSION

To support non-technical decision-makers to

implement big data applications, we have proposed a

reference model for big data management that

extends current theorizing in the NIST Big Data

Reference Architecture by adding the aspects of

actionability, effectuation, and intelligence. The

existing NIST model was extended by the following

points with the intention of making it more

actionable: our proposed model aligns the

dimensions of business and IT in BDM; it introduces

the explicit management of data intelligence, (i.e.,

the ability to apply as well as acquire knowledge

and skills by and about BDM); it introduces the

explicit management of effectuation of data

analytics results to create value; it links BDM to the

business model canvas method to provide a

procedure model; and it links BDM to project

planning and reporting using the Scrum method.

We have evaluated our model in two pilot

applications. Thus, we qualitatively establish the

following discussion: The value of our method

consists of providing information to structure big

data application design from scratch to decision

makers. The model seems effective, especially for

recognizing and analyzing phases by reducing

entropy about possible starting points for BDM. The

correspondence between business and IT aspects is

very interesting. The two case studies are also

interesting and show many valuable findings.

However, the proposed method is only a high-level

business informatics framework for big data

management with very little technical details. To

make the BDM canvas effective for practitioners,

support and guidance for the users are needed (e.g.,

moderation support for managers to assign their

inputs to the right fields and guidance support to

choose technical options within the fields of action).

6 OUTLOOK

As a next step, software support will be developed

for using the BDM canvas to document current

states, to plan new applications, and to track

progress in big data projects. We intend to develop a

collaborative BDM documenting, planning, and

configuration and reporting software platform based

on a virtual BDM canvas. This software should

Modeling and Qualitative Evaluation of a Management Canvas for Big Data Applications

155

support the method and process of filling the canvas

by the users without the need for expert coaching by

providing step-by-step user guidance. This software

should provide, for each field of action, a menu of

specific technical options and provide technical

depth to the point that the choices can later be

applied for a semi-automated cloud platform

configuration for new big data projects. To

empirically support the choice of technical options

for each canvas field, a meta-analysis of several

existing big data application case studies (e.g., in

Davenport and Dyché, 2013) should be performed.

The BDM canvas in its electronic form can, in turn,

support process automation in BDM by partially

automating the configuration of new cloud big data

applications. Therefore, a second step is to develop a

configuration tool structured by the BDM canvas to

automate building infrastructures for data storage,

analytics, and visualization. This automated big data

cloud computing environment will ask specific

parameters in each of the fields of action of the

BDM canvas, and use this input for the automated

configuration of (1) the analytics platform using the

CRISP4BigData method (Berwind et al., 2016) and

(2) the interactive visualization using the

IVIS4BigData method of Bornschlegel et al., (2016).

The aim is to provide a platform that automates the

task of requirements engineering and configuration

for cloud big data applications as far as possible.

REFERENCES

Abadi, D., Agrawal, R., Ailamaki, A., Balazinska, M.,

Bernstein, P. A., Carey, M. J., et al. (2016). The

Beckman Report on Database Research. Commun.

ACM, 59(2), 92–99.

Argyris, C. (1996). Actionable Knowledge: Design

Causality in the Service of Consequential Theory. The

Journal of Applied Behavioral Science, 32(4), 390–

406.

Berwind, K., Bornschlegl, M., Hemmje, M., & Kaufmann,

M. (2016). Towards a Cross Industry Standard Process

to support Big Data Applications in Virtual Research

Environments. Proceedings of Collaborative

European Research Conference CERC2016. Cork

Institute of Technology – Cork, Ireland.

Bornschlegl, M. X., Berwind, K., Kaufmann, M., Engel, F.

C., Walsh, P., Hemmje, M. L., et al. (2016).

IVIS4BigData: A Reference Model for Advanced

Visual Interfaces Supporting Big Data Analysis in

Virtual Research Environments. Advanced Visual

Interfaces. Supporting Big Data Applications (pp. 1–

18). Springer, Cham.

Chang, W. L. (2015a). NIST Big Data Interoperability

Framework: Volume 1, Definitions. NIST Special

Publication, NIST Big Data Public Working Group.

Chang, W. L. (2015b). NIST Big Data Interoperability

Framework: Volume 5, Architectures White Paper

Survey. Text.

Chang, W. L. (2015c). NIST Big Data Interoperability

Framework: Volume 6, Reference Architecture. Text, .

Davenport, T. H. (2013). Analytics 3.0. Harvard Business

Review, 91(12), 65–72.

Davenport, T. H., & Dyché, J. (2013). Big Data in Big

Companies. Portland, Oregon: International Institute

for Analytics.

Demchenko, Y., Grosso, P., Laat, C. de, & Membrey, P.

(2013). Addressing big data issues in Scientific Data

Infrastructure. 2013 International Conference on

Collaboration Technologies and Systems (CTS) (pp.

48–55).

Floridi, L. (2012). Big Data and Their Epistemological

Challenge. Philosophy & Technology, 25(4), 435–437.

Gartner. (2012, May 25). What Is Big Data? - Gartner IT

Glossary - Big Data. Gartner IT Glossary.

Gügi, C., & Zimmermann, W. (2016).

Betriebswirtschaftliche Auswirkungen bei der

Nutzung von Hadoop innerhalb des Migros-

Genossenschafts-Bund. In D. Fasel & A. Meier (Eds.),

Big Data, Edition HMD (pp. 301–317). Springer

Fachmedien Wiesbaden.

Hilbert, M., & López, P. (2011). The World’s

Technological Capacity to Store, Communicate, and

Compute Information. Science

, 332(6025), 60–65.

Kaufmann, M. (2016). Towards a Reference Model for

Big Data Management. Research Report, University

of Hagen, Faculty of Mathematics and Computer

Science.

Laney, D. (2001). 3D Data Management: Controlling

Data Volume, Velocity, and Variety. Application

Delivery Strategies. Report, Stamford: META Group.

Luftman, J., & Brier, T. (1999). Achieving and Sustaining

Business-IT Alignment. California Management

Review, 42(1), 109–122.

Mosley, M. (2008). DAMA-DMBOK Functional

Framework. DAMA International.

OECD. (2017). Data analysis definition. OECD Glossary

of Statistical Terms -, .

Österle, H., Becker, J., Frank, U., Hess, T., Karagiannis,

D., Krcmar, H., et al. (2010). Memorandum on design-

oriented information systems research. European

Journal of Information Systems, 20(1), 7–10.

Osterwalder, A., & Pigneur, Y. (2010). Business Model

Generation: A Handbook for Visionaries, Game

Changers, and Challengers (1st ed.). Hoboken, NJ:

John Wiley & Sons.

Provost, F., & Fawcett, T. (2013). Data Science and its

Relationship to Big Data and Data-Driven Decision

Making. Big Data, 1(1), 51–59.

Schwaber, K. (2004). Agile Project Management with

Scrum. Microsoft Press.

Zikopoulos, P., & Eaton, C. (2011). Understanding Big

Data: Analytics for Enterprise Class Hadoop and

Streaming Data (1st ed.). McGraw-Hill Osborne

Media.

DATA 2017 - 6th International Conference on Data Science, Technology and Applications

156