Bridging the Gap between Multidimensional Business Problem

Formulation and the Implementation of Multidimensional

Data Models

Franca Piazza and Friedrich Röhrs

Chair of Management Information-Systems, Saarland University, Saarbrücken, Germany

Keywords: Data Warehouse Development, Multidimensional Data Model, Knowledge Base, ADAPTed UML, Human

Resource Management.

Abstract: In data warehouse implementations the comprehensive identification of user requirements is a critical

success factor. The translation of business problems into conceptual data models is very challenging and

currently there are no information systems that assist in these tasks. In this research, we therefore elaborate

on the challenges in that specific area and develop an information system to gather information needs and

automatically transform them into conceptual multidimensional data models. We then demonstrate the use

of the information system within the domain of Human Resource Management.

1 INTRODUCTION

One of the most critical success factors of data

warehouse (DW) implementation is the stringent

user involvement and complete definition and

consideration of user requirements (e.g., Olbrich, et

al., 2012); (Yeoh, 2011); (Yeoh and Koronios,

2010); (Xu and Hwang, 2008). DW store a huge

amount of business data from heterogeneous source

systems and provide users with relevant business

information. The decision whether a piece of

information is relevant or not has to be determined

by users according to the business problems to be

solved. Within the DW context, the business

problems addressed show certain characteristics

such as multidimensionality or the consideration of

hierarchies etc., which are specified in detail in

section 2. Nevertheless, the DW system has to be

business-driven in order to contribute to the solution

of business problems (Olbrich et al., 2012); (Yeoh

2011); (Kimball, 1996). Therefore, the user

requirements (in terms of their information needs)

regarding certain business problems have to be

comprehensively considered within DW projects.

They serve as a basis for the business and technical

concept of DW. One possibility to develop DW is to

use the Model Driven Architecture (MDA)

approach, which has been adapted for DW

development (Mazón and Trujillo, 2008). Following

this approach the user requirements have to be

transformed into a conceptual model understandable

by IT professionals but still independent from a

specific DW platform.

Once the user requirements are known and

designed as a conceptual model using for example

UML (Mazón and Trujillo, 2008) system support is

given in order to transform this model into

multidimensional platform specific models and even

into multidimensional code to implement DW

systems (Mazón and Trujillo, 2008); (Gluchowski et

al., 2009).

Concerning the user requirements methodical

support is available to ascertain the specific

information needs in the DW context (Winter and

Strauch, 2003).

System support to transform conceptual

multidimensional data models into multidimensional

coding and methodical support to analyze

information needs in the DW context are available.

However, an integrated solution to support the

ascertainment of information needs and to

automatically transform these user requirements into

a conceptual multidimensional model is still lacking.

Additionally there is a gap between the

information needs articulated by business

professionals using domain specific terms and the

conceptual multidimensional model modeled in a

language, such as UML for example, understandable

Piazza F. and Röhrs F..

Bridging the Gap between Multidimensional Business Problem Formulation and the Implementation of Multidimensional Data Models.

DOI: 10.5220/0004394700790087

In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 79-87

ISBN: 978-989-8565-59-4

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

by IT professionals. Given that the comprehensive

consideration of user requirements is still a critical

success factor, the support of these two phases, the

ascertainment of user requirements and the

derivation of a conceptual model, as well as bridging

the gap between business professionals and IT

professionals within the DW implementation is

important.

Therefore, our aim is to develop an integrated

solution to ascertain information needs and to

automatically transform the user requirements

concerning multidimensional business problems into

multidimensional conceptual model so as to improve

and support this step of DW system implementation.

Following the process model for design science

research (Peffers et al., 2007) the paper is structured

as follows:

The research problem is elaborated by depicting

the state of the art concerning the MDA approach to

develop data warehouse systems and by deriving the

specific challenges to ascertain information needs

concerning multidimensional business problems. In

section three we derive the objectives for a solution

to solve the research problem. Afterwards, the

design and prototypical implementation of a

software artifact, that solves the problem, is

described. In section five the application of the

prototype is demonstrated within the Human

Resource Management (HRM) context and finally

conclusions are drawn.

2 FROM USER REQUIREMENTS

TO DW IMPLEMENTATION

2.1 The Model Driven Architecture

Approach for Data Warehouse

Development

The MDA serves as a standard framework for

software development and allows the development

of models by using a standard notation (OMG

Group, 2012). The MDA separates the specification

of functionality in a computer independent model

(CIM) and a platform independent model (PIM)

from the specification that is based on a certain

technology in a platform specific model (PSM).

Mazón and Trujillo (2008) and Mazón et al. (2005)

developed an MDA approach especially for the

development of data warehouses (see figure 1). In a

first step all the user requirements have to be

ascertained in a computer independent model (CIM).

These user requirements serve as a basis for the

modeling of every data warehouse layer by

corresponding platform independent models (PIM).

These models are specified using several UML

profiles. The PIM can be automatically transformed

into several platform specific models (PSM) which

depend on the required target technology. Finally,

from the PSM the necessary SQL code can be

derived to create the multidimensional data

structures for the Data Warehouse in a relational

platform (Mazón and Trujillo 2008); (Gluchowski et

al., 2009); (Kurze and Gluchowski 2008).

Figure 1: MDA approach for data warehouse modeling

(Mazón and Trujillo, 2008).

According to MDA, the ascertainment of the

information needs representing the multidimensional

business problems is an essential step. As soon as a

multidimensional platform independent model (MD

PIM) exists an automatic transformation into

multidimensional platform specific model (MD

PSM) and even the generation of code is possible

(Mazón and Trujillo 2008); (Gluchowski et al.,

2009). This automatic transformation ensures that

the specifications are comprehensively considered

and mistakes due to manual transformations are

eliminated. The automation therefore leads to time

savings concerning the data warehouse

implementation. Sophisticated methods to collect

information needs in DW projects with clearly

defined steps have been developed (Winter and

Strauch, 2003), but they lack the support to

transform CIM into MD PIM and require the manual

modeling of MD PIM.

2.2 Formulating Multidimensional

Business Problems

Data warehouses aim at providing OLAP

functionalities to users and allow the interactive

analysis of measures from multiple perspectives, the

so called dimensions. Multidimensional business

problems reflect the business need to analyze

measures interactively from different and combined

views, for example the analysis of the measure

“headcount” at a certain time for a specific

department detailed by qualification with the

possibility to additionally view the age distribution.

The main stakeholder in this context is the business

(CIM) user

requirements

MD PIM MD PSM MD Code

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

professional who is responsible for formulating the

multidimensional business problems. Business

professionals comprise managers or controllers of

certain business domains such as Human Resource

Management (HRM), sales or financial management

for example. Their information needs, based on their

multidimensional business problems, reflect the user

requirements to be ascertained in a CIM and which

serve as a basis for the MD PIM.

First of all a clear definition and description of

the measure has to be found. In order to build a

homogeneous information basis with the data

warehouse that provides information from

heterogeneous source systems, the identification of

synonyms and homonyms and a clear and precise

definition of the relevant measure are indispensible

(Winter and Strauch, 2003). Furthermore, there are

not only simple measures, such as “headcount”, but

also derived measures which are calculated using

simple measures, such as “level of overtime” for

instance, which is calculated using the “actual

working time” and the “target working time”. A

clear definition and in the case of derived measures a

valid calculation rule ensures a common

understanding of the measures which is essential in

the data warehouse context since the data warehouse

should serve as a general information base for the

company. Multidimensional business problems aim

at the analysis of measures from different

dimensions. These dimensions (for example “time”

or “region”) reflect the analysis facilities of certain

measures. Beyond the multidimensional aspect there

is also a hierarchical aspect within multidimensional

business problems. It is important to perform

analyses interactively on different aggregation levels

along dimensions. This means for example, that the

measure “headcount” should be analyzed within the

dimension “time” on the level “day” as well as

“week”, “month” and “year” while an interactive

change between these different aggregation levels is

possible.

The formulation of multidimensional business

problems by defining measures or calculating

derived measures, specifying dimensions and their

hierarchical structure is very time intensive and

hence costly. Deep domain knowledge is necessary

to achieve a consistent multidimensional business

problem formulation. The information needs have to

be ascertained manually within a company. Usually,

the requirement analysis is done with the business

professionals by interviews and by analyzing

existing reports and documents (Winter and Strauch,

2003). It is obvious that the formulation of a

multidimensional business problem has to be based

on business concepts related to a specific business

domain such as sales, controlling or HRM. At the

same time it needs to be understood by IT

professionals in order to derive relevant

multidimensional data models. The manual

translation of the information needs into a MD PIM

understandable by IT professional includes media

breaks and hence bears the risk of

misunderstandings. To meet these challenges we

propose to develop an information system with

business content stored in a knowledge base which

supports the ascertainment of multidimensional

information needs as well as the derivation of

technically and conceptually understandable MD

PIM. Our solution thus focuses on the creation of the

CIM and the MD PIM as well as the respective

automatic transformation (see figure 1).

3 OBJECTIVES

FOR A SOLUTION

First of all, the system should serve to reduce time

and cost for DW development by supporting and

automating steps in the process of ascertaining

information needs and deriving MD PIM. It is

obvious, that the pure mechanism to automate

process steps holds the potential to save time and

costs, but it is not sufficient. In fact, the costly

process of formulating multidimensional business

problems (see section 2.2) should be supported by

predefined business content reflecting a

comprehensive collection of knowledge about

potential measures in a knowledge base. The

knowledge base stores the knowledge of diverse

business domains such as HRM, sales and financial

controlling. The measures should be defined clearly

and in case of derived measures also concise

calculation rules should be provided. Furthermore,

all potential dimensions and hierarchy levels to

analyze these measures should be stored in the

knowledge base. This means that the knowledge

base comprises all dimensions and hierarchy levels

that are reasonable from a business point of view.

This knowledge base then can serve to satisfy future

information needs. So beyond necessary technical

functionalities the predefined knowledge base is an

essential part of the system to be developed.

Therefore, the following first requirement for the

system can be derived.

(R1) The system must allow the collection and

management of knowledge about multidimensional

business problems.

BridgingtheGapbetweenMultidimensionalBusinessProblemFormulationandtheImplementationofMultidimensional

DataModels

Whereas the system should provide a

comprehensive and complete knowledge base about

multidimensional business problems within a

business domain, only a subset of this knowledge

base is relevant in specific DW projects. Therefore,

the system should support the user to ascertain the

information needs with the feature to identify and

select a relevant subset of the knowledge base. This

specified subset then is used to derive the respective

MD PIM. This objective leads to the following

system requirement.

(R2) The system must support the user in

selecting relevant measures and dimensions from the

provided knowledge base according to specific

information needs.

As elaborated before, the system should serve to

avoid misunderstandings between business and IT

professionals. Therefore, the usage of domain

specific business vocabulary is mandatory and

should be automatically transformed into a model

language, such as UML, which IT professionals can

easily understand and work with. This ensures a fail-

safe communication between business professionals

and IT professionals. Furthermore, media breaks are

diminished as the formulation of the

multidimensional information needs as well as the

generation of a MD PIM are integrated in one

system. According to the aforementioned objectives,

the system must meet the following requirements:

(R3) The system must represent the information

in ways understandable by IT professionals as well

as business professionals.

(R4) The system must automatically generate a

MD PIM based on the information needs of the

business professionals in a common

multidimensional model language.

The elaborated requirements are realized in a

software artifact which is introduced in the

following.

4 ARTIFACT DESIGN

4.1 Actors

As information systems are sociotechnical systems,

the definition of potential actors is necessary (De

Bruijn and Herder, 2009); (Bostrom et al., 2009).

According to the objectives of the system, it aims at

supporting business professionals and IT

professionals who are therefore two potential actors

of the system.

Business professionals represent the actual core

user of the system. They are employees of

organizations who want to execute multidimensional

analyses in their business domain. While they have

some idea of the information they want to generate

they are usually not experts of multidimensional

analysis or data warehouses. Their goal is to define

and formulate their multidimensional business

problem using domain specific concepts stored in

the system.

IT professionals are the persons tasked with

creating the DW. They usually have no or limited

specific domain knowledge and therefore depend on

the business professionals to provide them with their

information needs. The system provides the IT

professionals with understandable information by

generating a MD PIM.

Furthermore, the management of the

knowledgebase requires the definition of another

actor, the domain expert. Domain experts represent

persons who have a broad understanding of the

relevant business domains and hence of relevant

measures (their structure as well as their

dimensions). This actor type includes business

professionals as well as researchers and consultants

with wide experience in specific business domains.

The knowledge of the domain experts provides the

basis for the knowledge base integrated into the

system. Domain experts manage the knowledge

base.

4.2 Components of the System

The underlying idea for the general design of the

system is to gather the knowledge of the domain

experts in a knowledge base which then can be

accessed and used by other business professionals in

order to generate a model the IT professional can

work with (Earl, 2001). For this the System must

provide three core functionalities: the management

of the knowledge base, the selection of measures and

dimensions (creating a MD requirements model) and

the model generation (creating a PIM). The overall

architecture of the system resulting from this design

choice is shown in figure 2. Each component is

further described in the following sections.

The repository comprises the knowledge base

and a collection of MD requirements models. The

knowledge base bases its models on the MD domain

meta model describing the structure of the concepts

in the domain and each MD requirements model is

based on the MD requirements meta model.

As such, a first step in the creation of a system

offering the required functionalities is the

representation of the problem domain in the system:

the MD domain meta model and the MD

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

Figure 2: Components of the system.

requirements meta model. There are three types of

primitives that make up the structure of these

models: concepts, properties and relations between

concepts. Concepts are central objects in the domain

(e.g. a dimension or a measure). Each concept can

have (one or more) properties. Each property has a

name and a value. Concepts can be related to each

other by relations. Relations are used for hierarchical

ordering of concepts as well as other semantic

meanings. For example the dimension “time” can be

viewed at the dimension level “day” and “month”

(hierarchical ordering), and the derived measure can

be derived from other measures.

The concepts of these meta models are shown in

figure 3. The main concept around which all other

concepts are centred is the measure. Measures

represent a numerical value uniquely determined by

the value of the dimensions that provide its context

(Chaudhuri and Dayal, 1997). Measures can be

differentiated into simple measures and derived

measures which are calculated using other measures.

Dimensions allow the representation of measures on

different aggregation levels.

The concept function can be used to categorize

measures into specific business functions. Each

function represents one business function and can be

associated to as many measures as are relevant for

that business function. Measures are not limited to

one specific function, but can be relevant to multiple

functions. The measures “actual working time” and

“target working time” are relevant in the business

function personnel retention as well as the business

function personnel planning for example. This

information is especially relevant for business

professionals as they search relevant measures for

the specific business problem they try to formulate.

The knowledge base stores the knowledge about

multidimensional business problems in entities that

are created based on the concepts defined in the MD

domain meta model. Known and useful measures,

calculation rules, dimensions and dimension levels

as well as the functional association relevant for a

business domain and relevant for multidimensional

business problems are specified and stored in the

system. For example, when the measure “actual

working time” is described in the knowledge base,

all potential dimensions such as time, region,

organization etc. with respective dimension levels

such as minute, hour, day, etc. are specified and

personnel functions are determined and associated.

Figure 3: Meta models used by the system.

BridgingtheGapbetweenMultidimensionalBusinessProblemFormulationandtheImplementationofMultidimensional

DataModels

These entities of the knowledge base also have their

properties assigned. While the properties used can

differ from concept to concept, generally they all

include a name and description property. Following

the previous example, the property “description” of

the measure contains a general description of this

measure allowing other actors to better understand

its meaning. Furthermore, the property “literature”

of a measure can contain bibliographic content

pointing to literature with comprising and detailed

definition of the measure. If a measure is derived

from other measures, the property calculation rule

contains the rule by which this measure is derived

from others. Beyond that, the source property of a

measure provides information about the source

systems from which a specific measure can be

obtained.

The concepts cube and measure instance

depicted in the MD requirements meta model (figure

3) are relevant for representing specific information

needs contained in the MD requirements models.

The MD requirements models each link to a specific

part of the knowledge base according to the specific

information needs they represent. While the

knowledge base comprises the general knowledge

about multidimensional business problems within

different business domains, the specific information

needs of business professionals constitute a partition

of the whole knowledge base. By determining the

information needs, measures are instantiated with

specific dimensions and levels. As for example the

measure “actual working time” has the potential

dimension levels “minute”, “hour”, “day” up to

“year” stored in the knowledge base, a specific

information need comprises relevant dimension

levels on a more aggregated level from “day” up to

“year” neglecting the more granular levels. To allow

this the measure instance concept links between the

general measure and a specific dimension level at

which it is measured.

4.3 Core Functions

The system offers three core functions providing the

interaction of the actors with the system: the

management of the knowledge base, the selection of

requirement specific measures and the generation of

a platform independent model from the knowledge

base and the specific requirements.

The management component allows domain

experts to access the knowledge base and to create,

edit and delete respective entities. The domain

experts are able to define functions, measures,

dimensions and dimension levels and to specify the

respective properties such as name, description,

calculation rule, literature source etc.

Based on the knowledge stored in the knowledge

base, business professionals can select measures,

dimensions and dimension levels to define and

formulate their multidimensional business problem,

resulting the specific MD requirements models The

process of formulating multidimensional business

problems is facilitated by selecting relevant

measures and determining the needed level and

hence the business professional creates the MD

requirements model, which can be seen as a CIM

according to the MDA introduced in chapter 2.1.

Based on the MD requirements model and

further information available in the knowledge base,

a multidimensional platform independent model

(MD PIM) based on ADAPTed UML (Priebe and

Pernul, 2001) is generated. This model can then be

used by IT professionals for further development of

the DW.

4.4 Implementation Detail

The system architecture described is realized as a

web application created using Java technologies. In

the prototype the raw data managed in the repository

is saved in a graph database Neo4j (Neo Technology

inc., www. neo4j.org/) since the interconnected

concepts in the domain and requirements model

suggest that form of storage. Furthermore Neo4j

offers a simple and clearly defined API, allowing the

wrapping of nodes directly to java objects. The

different components of the system can then work

on those objects without having any knowledge of

the underlying persistence layer. The inference

mechanisms needed by the selection and generation

components are achieved by extending the java

objects, i.e. adding new methods that will support

the inference based on the underlying graph nodes

and relations. For example the measure instance

“actual working time” gathered at the level “day” is

also available at all aggregated levels such as

“week”, “month” and “year” which are inferred by

the method. In the database the node representing

the measure instance “actual working time”

specified at the level “day” is only connected to the

node of the dimension level “day”. By traversing the

graph the system can then infer the other dimension

levels available for the measure.

The representation on the client side is achieved

by different dynamic pages. They allow users to

access the management functionality of the

knowledge base as well as the selection and

generation component.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

The generation component furthermore allows a

visual representation of PIMs in the ADAPTed

UML notation on the dynamic page, as well as the

export of these models as an XML document, so that

they can be further edited with independent

modeling tools.

5 APPLICATION IN HRM

The prototype implemented so far contains a

comprehensive knowledge base for the business

domain Human Resource Management (HRM). A

comprehensive literature review in the domain of

HRM was performed. To identify relevant literature

a keyword search using a scholarly Internet search

engine (scholar.google.com), several scholarly

online databases (ABI/Inform, Business Source

Premier, Scopus, and Science Direct) and online-

libraries was employed. Search terms such as “

HRM” and “controlling” in combination with

“metrics” or “measures” as well as “recruiting” or

“performance” were applied. The literature review

revealed 208 domain specific HR measures such as

“headcount”, “level of overtime” or “absence quote”

with nine dimensions such as “organization”, “time”

or “communication channel”. Within every

dimension detailed hierarchies are stored which

reflect the most granular and widest possible spread

of the respective dimension. Also heterarchies,

parallel hierarchies and different path lengths are

possible. The dimension “time” for example

contains “minutes” as the most granular level and

can be aggregated into “weeks”, “month” and

“years” also considering the specific aggregation of

“days” in “weeks” and in “months”. To give an

example considering the knowledge base: The

measure “actual working time” can be analyzed via

the dimensions “time”, “organization”, “employee”

and “region” where all of these dimensions contain

comprehensive hierarchies such as depicted for the

dimension “time”. Furthermore a clear definition of

the measure “actual working time” is provided. The

“actual working time” is a simple measure and has

no calculation rule. But the “level of overtime” for

example uses the “actual working time” and rates

this against the “target working time” which is

defined in the respective calculation rule. Beyond

that, it is also possible to enter the source system in

which the measure or part of the measures are

stored. If for example the “actual working time” is

so far stored in Excel files, this information can be

entered into the system so as to provide the IT

professionals with information for the ETL-process

in the DW project.

The screenshots provided in figure 4 represent

the functionality to select certain measure instances

(a) and to specify the needed dimension levels (b)

Figure 4: Screenshots of the prototype.

BridgingtheGapbetweenMultidimensionalBusinessProblemFormulationandtheImplementationofMultidimensional

DataModels

and the generated MD PIM using ADAPTed UML

(c).

In order to support the business professional with

the selection of relevant measures, the measures

within the domain model are structured based on HR

functions derived from the Michigan Approach

(Tichy et al., 1982). This is a well known and wide

spread approach to categorize HR functions and is

relevant for practice as well as for research. Hence,

this categorization supports the business professional

to select the relevant measures and diminishes the

costs of structuring specific multidimensional

business problems. The possibility to select relevant

measures and relevant dimensions supports the

ascertainment of user requirements. Furthermore,

domain experts can also expand the knowledge base

in case new requirements appear.

In the screenshot (figure 4, a), the two measures

“actual working time” and “target working time”

associated to the HR function personnel retention are

selected. As these measures are relevant for multiple

HR functions such as personnel retention, personnel

planning and cost planning for example, these

measures can be found within all of these HR

functions. The business professional selects these

two measures from one of the mentioned HR

functions and further determines the relevant

hierarchy levels by selecting the appropriate

dimension level (figure 4, b). The functionality of

the system is demonstrated for the case that the

“actual working time” and the “target working time”

should be analyzed from multiple perspectives such

as time and organizational unit (among others).

Depending on the information needs the aggregation

levels can be flexibly selected such as for example

the dimension level “day” (and higher levels) within

the dimension “time” and the dimension level

“position” (and higher levels) within the dimension

“organizational unit” (figure 4, b). In a last step the

IT professional can trigger the model generation and

receives a conceptual multidimensional data model

in ADAPTed UML (figure 4, c), the MD PIM. IT

professionals can use this model to further

implement the DW. The XML-export function also

allows the export of the specific requirements model

into a modeling tool.

6 CONCLUSIONS

It could be demonstrated that the developed

prototype supports the specification of user

requirements concerning the formulation of

multidimensional business problems. The business

professionals can work with the predefined

knowledge base and specify a requirements model

using their familiar business terms. This

requirements model then serves as a basis for the

generation of the MD PIM. Due to the knowledge

base, time intensive and hence costly processes of

ascertaining information needs and of formulating

the multidimensional business problems can be

avoided. Furthermore the communication with the

IT professional can be improved since the business

professional still works in his business domain using

domain specific vocabulary and the conceptual

model for the IT professional can be generated

automatically. The HR business professional works

only with concepts known in the HR domain and

does not have to think about e.g. parallel hierarchies

or the number of cubes to model. All the concepts

are provided within a predefined knowledge base

familiar to the HR business professional. Given the

generated MD PIM, the DW development process

can further take advantage of the transformation

steps demonstrated by Mazón and Trujillo (2008)

and Gluchowski, Kurze and Schieder (2009).

Nevertheless, there are some limitations and

further work to be done. First of all, the developed

system has to be evaluated by professionals within

DW projects. The evaluation could for example be

based on the Technology Acceptance Model (Davis,

1989) or the Information System Success Model

(DeLone and McLean, 2003) to research acceptance

and success of the system.

Furthermore, it is obvious that the knowledge

base, especially its completeness, is essential for the

applicability and success of the prototype. So far, the

prototype only contains HR measures, dimensions

and dimension levels found in literature. It is

planned to integrate the knowledge of domain

experts that are business professionals as well as

researchers or consultants in the specific business

domain into the knowledge base. For this purpose

the prototype will be provided via web so the

domain experts can access the system and

incorporate their knowledge and enter measures,

definitions, calculation rules, dimensions and

dimension levels. To ensure the consistency and

validity of the system, an authorization and releasing

concept for the changes made by the domain experts

has to be worked out and implemented. So far only

the HRM domain is implemented but other business

domains such as sales or financial controlling can be

integrated as well.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

REFERENCES

Bostrom, R. P., S. Gupta, and D. Thomas. 2009. “A Meta-

Theory for Understanding Information Systems

Within Sociotechnical Systems.” Journal of

Management Information Systems 26 (1): 17–48.

De Bruijn, H., and P. M. Herder. 2009. “Systems and

Actor Perspectives on Sociotechnical Systems.” IEEE

Transactions on Systems, Man and Cybernetics, Part

A: Systems and Humans 39 (5): 981–992.

Chaudhuri, Surajit, and Umeshwar Dayal. 1997.

“Overview of Data Warehousing and OLAP

Technology.” ACM SIGMOD Record 26 (1): 65–74.

Davis, F. D. 1989. “Perceived Usefulness, Perceived Ease

of Use, and User Acceptance of Information

Technology.” MIS Quarterly 13: 319–340.

DeLone, W. H., and E. R. McLean. 2003. “The DeLone

and McLean Model of Information Systems Success:

A Ten-Year Update.” Journal of Management

Information Systems 19 (4): 9–30.

Earl, Michael. 2001. “Knowledge Management

Strategies.” Journal of Management Information

Systems 18 (1) (October): 215–233.

Gluchowski, P., C. Kurze, and C. Schieder. 2009. “A

Modeling Tool for Multidimensional Data Using the

ADAPT Notation.” In The 42nd Hawaii International

Conference on System Sciences (HICSS), 3575–3584.

IEEE Press.

Kimball, Ralph. 1996. The Data Toolkit: Practical

Techniques for Building Dimensional Warehouses.

New York, NY, USA: John Wiley & Sons.

Kurze, Christian, and Peter Gluchowski. 2008.

“Computer-Aided Warehouse Engineering (CAWE):

Leveraging MDA and ADM for the Development of

Data Warehouses.” In AMCIS 2010 Proceedings.

Paper 282.

Mazón, J. -N., and J. Trujillo. 2008. “An MDA Approach

for the Development of Data Warehouses.” Decision

Support Systems 45: 41–58.

Mazón, J. -N., J. Trujillo, M. Serrano, and M. Piattini.

2005. “Applying MDA to the Development of Data

Warehouses.” In The 8th ACM International

Workshop on Data Warehousing and OLAP (DOLAP),

57–66. ACM Press.

Neo Technology, Inc. “Neo4j – The World’s Leading

Graph Database.” http://www.neo4j.org/.

Olbrich, S., J. Pöppelbuß, and B. Niehaves. 2012. “Critical

Context Success Factors for Business Intelligence: A

Delphi Study on Their Relevance, Variability, and

Controllability.” In 45th IEEE International

Conference in System Sciences, 4148–4157. IEEE

Press.

OMG Group. 2012. “OMG Model Driven Architecture.”

http://www.omg.org/mda/.

Peffers, K., T. Tuuanen, M. A. Rothenberger, and S.

Chatterjee. 2007. “A Design Science Research

Methodology for Information Systems Research.”

Journal of Management Information Systems 24: 45–

77.

Priebe, T., and G. Pernul. 2001. “Metadaten-gestützter

Data-Warehouse-Entwurf Mit ADAPTed UML.” In

5th Internationale Tagung Wirtschaftsinformatik.

Paper 8.

Tichy, Noel M., Charles J. Fombrun, and Mary Anne

Devanna. 1982. “Strategic Human Resource

Management.” Sloan Management Review 23 (2): 47–

61.

Winter, R., and B. Strauch. 2003. “A Method for Demand-

driven Information Requirements Analysis in Data

Warehousing Projects.” In The 36th Hawaii

International Conference on System Sciences (HICSS),

1359–1365. IEEE Press.

Xu, Hongjiang, and Mark I. Hwang. 2008. “A Structural

Model of Data Warehousing Success. Journal of

Computer Information Systems.” The Journal of

Computer Information Systems 49: 48–56.

Yeoh, William. 2011. “Business Intelligence Systems

Implementation: Testing a Critical Success Factors

Framework in Multiple Cases.” International Journal

of Business Information Systems 8 (2): 192–209.

Yeoh, William, and Andy Koronios. 2010. “Critical

Success Factors for Business Intelligence Systems.”

Journal of Computer Information Systems 50 (3): 23–

32.

BridgingtheGapbetweenMultidimensionalBusinessProblemFormulationandtheImplementationofMultidimensional

DataModels