The learnability of the dimensional view of data and what to do with it
| Published date | 21 January 2019 |
| DOI | https://doi.org/10.1108/AJIM-05-2018-0125 |
| Date | 21 January 2019 |
| Pages | 38-53 |
| Author | Dušan Vujošević,Ivana Kovačević,Milena Vujošević-Janičić |
| Subject Matter | Library & information science,Information behaviour & retrieval,Information & knowledge management,Information management & governance,Information management |
The learnability of the
dimensional view of data and
what to do with it
Dušan Vujošević
Faculty of Computer Science, Union University, Belgrade, Serbia
Ivana Kovačević
Faculty of Organizational Sciences, University of Belgrade, Belgrade, Serbia, and
Milena Vujošević-Janičić
Faculty of Mathematics, University of Belgrade, Belgrade, Serbia
Abstract
Purpose –The purpose of this paper is to examine the usability of the dimensional view of data in the
context of its presumed learnability.
Design/methodology/approach –In total, 303 participants were asked to solve 12 analytical problems in
an experiment using the dimensional view of data for half of the problems and an operational view of data for
the other half. Inferential statistics and structural equation modeling were performed with participants’
objective results and affective reactions.
Findings –Showing that the order of exposure to the two views of data impacts the overall usability of ad
hoc querying, the study provided evidence for the learnability potential of the dimensional view of data.
Furthermore, the study showed that affective reactions to the different views of data follow objective usability
parameters in a way that can be explained using models from affective computing research.
Practical implications –The paper proposes a list of guidelines for use of the dimensional view of data in
business analytics.
Originality/value –This study is the first to confirm the learnability of the dimensionalview of data and the
first to take a deeper look at affective reactions to an ad hoc business analytics solution. Also, it is one of few
studies that examined the usability of different views of data directly on these views, rather than using paper
representations of data models.
Keywords Business intelligence, Affective computing, Business analytics, Ad hoc querying,
Data modelling, View of data
Paper type Research paper
1. The introduction
Business users can query data with ad hoc querying tools. These tools follow the direct
manipulation paradigm –theusersdonotneedtocodetheirqueries,insteadtheypickuptables
and columns that they are interested in and obtain the data through simple drag-and-dropping,
double clicks or insert options. The obtained data can intuitively be sorted, filtered, summed up
or presented with charts. Ad hoc querying tools belong to the business intelligence family of
technologies, which is closely related to data warehousing technologies used for storing the data.
Business intelligence is used for business analytics, together with other technologies such as
predictive analytics, machine learning, or operation research software.
Business intelligence is in a phase of maturity (Fischer, 2018) and users are increasingly
empowered to move from the exploitation of data to their exploration (Alpar and Schulz,
2016). Business user analysis shifts from simple consumption of information to preparation
of diverse data representations and analyses (Stodder, 2015). The scope of business
intelligence has been extended from strategic questions to operational tasks, so the number
of employees that should be ready to leverage it increases (Alpar and Schulz, 2016).
In spite of this, business intelligence tools have not been widely accepted and in none of
the organizations included in a recent study did top management use business intelligence
Aslib Journal of Information
Management
Vol. 71 No. 1, 2019
pp. 38-53
© Emerald PublishingLimited
2050-3806
DOI 10.1108/AJIM-05-2018-0125
Received 13 June 2018
Revised 18 September 2018
Accepted 24 September 2018
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/2050-3806.htm
38
AJIM
71,1
as much as expected before the introduction of the system (Grublješičand Jaklič, 2015).
A reason for that might be an inadequate introduction of the business intelligence tools and
their adjustment to users. Since the contemporary economy is the economy of information
seeking and knowing (Huvila, 2016), in which technologies like business intelligence should
integrate non-intrusively into users’day-to-day activities (Pawlowski et al., 2015), and since
the way these tools are designed will inevitably impact users’day-to-day well-being
(Harbich and Hassenzahl, 2017), more attention should be payed to human-computer
interaction in this field.
The first step in bringing business intelligence tools closer to users was taken when
business intelligence architecture moved from client/server applications to web applications
(Alpar and Schulz, 2016). Social features such as a star rating or the number of access points
could also help business users evaluate the quality of reports (Alpar and Schulz, 2016).
Another example of attempts to boost the acceptance of business intelligence technologies is
the preparation of data views based on studies of the usability of differently modeled data,
one of which is presented in this paper. This study considers how the dimensional view of
data effects how well users can complete tasks as well as examines their affective response
to different views of data.
2. Related work
Previous research on the usability of the dimensional view of data measured usability either
in experiments with paper, focusing on data models, or in experiments with computers,
focusing on data. An interesting development that implied learnability was detected in an
analysis of the timeline of the solution of tasks in an experiment. Affective responses were
not approached systematically.
2.1 Usability of dimensional view of data
The usability and benefits gained from everyday business intelligence use depend on data
modeling (Hänel and Felden, 2017). The standard data modeling, used for databases of
operational or transactional systems, aims to eliminate data repetition and redundancy, so
that when a change happens to data it is only necessary to change data in one place (Bethke,
2017). Such an approach, also referred to as 3NF modeling, enables high data quality in
those systems where updates of information occur frequently.
The dimensional approach is mainly used for analytical applications. Dimensionally
modeled data are usually made from operationally modeled data, through processes of
extraction, transformation and loading, which make multiple changes to the structure of
data. Data from hierarchically separated data tables, such as those with information about
organizational structure or geographically distributed customers, are packed into single
new tables, called dimensions, which may contain redundant data. If necessary, data are
cleaned and codes converted to easily intelligible words.
The outlines of the two models are presented in Figure 1, where each rectangle
represents an object of the model, usually implemented in a database as a data table.
The dimensional model collects the same data as the transactional in a smaller number of
wider tables, which is indicated here by wider rectangles. As the figure shows, the
contextual description of an item is no farther than one network node away in a dimensional
model. In contrast, an operational model’s item can be described by data in tables that are
many nodes away from it. The occurrence of multiple paths between some tables makes the
operational model’s network even more complex.
The literature on dimensional modeling (Kimball and Ross, 2013) states that it is widely
accepted as the preferred technique for presenting analytic data because it delivers data that
is understandable to business users. However, a vibrant debate rages between the
technique’s supporters (Ross, 2017) and those specialists who question it. Considering even
39
Learnability
of the
dimensional
view of data
To continue reading
Request your trial