Empowering Collaborative Business Intelligence by the use of Online
Social Networks
Jens Kaufmann
and Peter Chamoni
Department of Technology and Operations Management, University of Duisburg-Essen, Lotharstr. 63, Duisburg, Germany
Keywords: Collaborative Business Intelligence, Online Social Networks.
Abstract: Online Social Networks (OSNs) gain more and more attention in nearly all systems or concepts that deal
with the exchange of information between human actors. While the analysis process in Business Intelligence
systems normally is based on structured and pre-filtered data, OSNs promise a lot of customer insights for
companies. On the other hand, OSNs are a common product used by the analysts themselves to interconnect
and discuss. Therefore they may serve the inter-human exchange about the analyzed figures and support
collaboration of the users. We show how OSNs can support the collaboration and communication in a joint
analysis process and prove the feasibility with a small prototype.
1 INTRODUCTION
Business Intelligence (BI) does not follow a single,
undisputable definition. It mainly is recognized as
the use of concepts, methods and systems that
support management decisions (Kemper et al.,
2010). In the whole context of Decision Support
Systems (DSS), BI systems can be sorted into the
category of data-driven DSS (Power, 2011). An
often quoted approach for a structure of BI traces
back to (Gluchowski, 2001). He distinguishes
different scopes of BI regarding the focus of the
system (technical or business-driven) and the
process phase (data supply or data analysis). We
understand BI in its broad scope and therefore
include the data gathering, processing and
provisioning (summarized under the term Data
Warehousing) as well as the data analysis by
humans and algorithms (called Data Mining).
The input data has to be put into structures that
analysts or deciders use to make and formulate
decisions. So the data is reduced, aggregated and
(concerning top decision makers) provided on a
level that does not explain every single number or
deviation to previous reported figures. This creates
the need to discuss about individual figures with
experts who do have a more detailed or special
knowledge of the data. This communication then
will lead to deeper analysis and maybe even to a
joint decision on how to react to the event. Over the
past few years, the term Collaborative Business
Intelligence (CBI) has emerged, that describes this
co-working in the analysis process. Today, the
definition for CBI neither is unanimous, nor is CBI
the only term that covers this joint analysis process.
Group decision making has been a major issue
for decision makers and decision support system
builders and users since the 1980’s especially in the
field of Operations Research (Vetschera, 1991). The
idea although has regained interest with the strong
emergence of online social networks (OSNs) in
private areas, e.g. facebook, or business cases, e.g.
Yammer. According to the Hype Cycle for Business
Intelligence 2012, published by Gartner Inc.,
Collaborative Decision Making (CDM) is presented
as an upcoming technology that is supposed to
substantially change the understanding and the
market of BI systems. In this context CBI enriches
BI solutions with social or collaborative capabilities
as known from OSNs (Bitterer, 2012), (Dayal et al.,
2008), (Muntean, 2012).
Today, OSNs often are considered as huge
databases that provide a mass of information about
(and from) customers. There are different
approaches to analyze the information and to include
it into the systems as e.g. (Costa et al., 2012) and
(Böhringer and Helmholz, 2011) do. These efforts
are characterized by the terms Social Business
Intelligence (Hinchcliffe and Kim, 2012) or Social
Analytics (Roe, 2011).
While OSNs provide the analysts with a great
125
Kaufmann J. and Chamoni P..
Empowering Collaborative Business Intelligence by the use of Online Social Networks.
DOI: 10.5220/0004335901250128
In Proceedings of the 9th International Conference on Web Information Systems and Technologies (WEBIST-2013), pages 125-128
ISBN: 978-989-8565-54-9
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
amount of data, the usage of the networks
themselves for the analysis process has not been
developed very far. Many vendors have released a
proprietary solution for communication in their BI
applications. Three factors can explain the lack of
direct interaction between BI systems and OSNs.
First, there would exist the need for a standardized
protocol to exchange messages and analysis
information over different systems and networks.
Second, security issues arise, when alien networks
are used. Third, it probably is not in many vendor’s
main interest, to provide an external access to their
systems, because using a best-of-breed-approach
instead of a single vendor solution would become
easier for companies.
Despite these concerns, we use a small piece of
the BI analysis world, namely ad-hoc-reporting, to
provide a simple discussion platform over OSNs
(which provide a communication platform as well as
the needed relationship information) with direct
linkage to reported figures. In section 2 we discuss
related work to this approach. Section 3 describes
the analysis process and the possible points of
application. In section 4 we present our findings
with a simple solution for collaboration. Section 5
gives an introduction on possible fields for research.
2 RELATED WORK
As stated before, in this paper we focus on Col-
laborative Business Intelligence and understand it as
the joint process of analysis between people in one
or more enterprises. The first workflow-oriented
proposals, what CBI should consist of, are from
(Rasmussen, 1999) and (Collins, 2001). In 2008
(Dayal et al., 2008) described a virtual cockpit for
collaborative decision making in enterprises.
Figure 1: Business Intelligence Architecture.
Since then, OSNs have become a wide-spread
product. (Berthold et al., 2010) explicitly mentioned
the combination of social software functions and BI
applications over an ad-hoc and analysis layer.
Approaches that present a more partner-driven or
cross-company view on the collaborative part of CBI
come from (Golfarelli et al., 2011). They presented
Business Intelligence Networks (BIN), that should
provide data over a peer-to-peer network to other
users. The authors focus on the technical aspect of
OLAP queries rather than on a communicative
aspect. (Liu and Daniels, 2012) take this idea, but
focus on the analysis itself, rather than on the data
supply, and state the benefit of a cross-company
analysis. None of the mentioned articles, however,
uses existing relationship information from OSNs to
let people (interal or external) communicate about
figures and reports. To the best of our knowledge,
this idea has not been discussed yet.
3 ARCHITECTURE
Business Intelligence applications encompass
different levels of data gathering, data processing,
data storing, data provisioning, and data presentation
(see Figure 1). The following subsections describe
the considered architecture of the system that
underlies the collaboration process.
3.1 The General Collaborative BI
Process
The more the data is processed and structured, the
simpler it is for a company to analyze. On the other
hand, structuring and filtering leads to possible data
loss and to predefined structures that might not be
universal, e.g. product hierarchies may even differ
from the production department to the sales
department. There obviously is a trade-off, regarding
structure and interchangeability with other
enterprises/ departments, which hardens
communication.
We therefore decide between two scenarios. In
the first scenario, we have one where everyone can
access the same information up to the lowest edited
level. In this case, questions may arise, that need
detailed fact knowledge about the circumstances of
numbers. The communication between the decider
and the expert will work on the same basis, only the
figure in focus has to be marked.
In our second scenario, the communicating
parties are from different companies or departments
that may even use different BI systems. They do not
WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies
126
have access to the exact same reports, but at least to
the same data. (If this is not the case, data exchange
must be a part of the process, see Figure 2.) In the
simple case, to ensure a useful communication, it is
necessary to define a consensual level of structure
and at the specific level a consistent and
homogenous definition of the objects in focus. For
the actual discussion, this can be understood as meta
data. The interchange of this information is a
prerequisite to the joint analysis process. From the
whole process of collaborative decision making in
the sense of CBI, we focus on the connecting and the
discussion itself rather than on the data exchange.
Given a BI architecture with two parties, we
argue that at a first step the parties must connect and
agree on the cooperation. After this, meta data has to
be exchanged as far as needed. That concerns the
data model, access structures, security issues, and so
on. Subsequently, data can be exchanged, if needed
and is not already made available, and the analyses
can be conducted.
3.2 Using OsNs for Connecting
and Discussing
Although some present BI solutions link reports to
authors or provide emails in and out the BI system,
these functions do not directly provide a basic
solution for inter-company or inter-system
communication. Since many companies nowadays
use social networks
1
, it seems natural to use
the already built-up relations there for connecting
Figure 2: General CBI Process.
people. Our approach uses the existing relations to
provide the user with a direct access to the network
information. Because many of the networks allow
1
According to Forbes, 85% of the Fortune 500 companies use
Yammer, an enterprise social network.
http://www.forbes.com/sites/ilyapozin/2012/06/26/the-rise-of-
social-enterprise/, as of 22.10.2012.
for a direct access to their functions via a
programming interface (called API), the relations
can be extracted from the network and messages can
be sent to other users instantly out of the BI
application. The following discussion can also use
the same way (as shown in Figure 3), which allows
for communication that afterwards can be decoupled
from the BI solution.
4 EXEMPLARY SOLUTION
WITH PALO AND YAMMER
To prove the feasibility of the idea, we chose two
widely used applications in the field of BI and social
networking and implemented a prototype of a
communications interface. The goal was to be able
to select a figure from an ad-hoc-report and to send a
question to another user (B) that had to be in the
contacts of the first person (A).
We chose PALO as an open-source BI
application and used an easily understandable data
cube that consists of the dimensions product, region,
and month/year. We only took the sales amount as a
figure (called fact). One of the features of PALO is
its Microsoft Excel-based frontend, which allows for
plugin-implementations via Microsoft Visual Basic
for Applications. With this plugin, we connected to
our virtual company at Yammer, where we used five
employees in different departments to simulate a
working environment. Yammer offers an API based
on JSON, which is a text-based, human-readable
data exchange format similar to XML. This allowed
for a simple interaction between our application and
the social network. The security of the connection
itself is provided by SSL and OAuth.
Figure 3: CBI Process with OSN Support.
EmpoweringCollaborativeBusinessIntelligencebytheuseofOnlineSocialNetworks
127
When a user sees a figure in the report he does
not understand, he is able to right-click on the value
and can chose a Yammer contact that he wants to
ask about it. The plugin then will send a message to
the Yammer account of the other user, serving the
information of the data cube, the current chosen
dimensions to describe the figure (that is a
rudimentary form of meta data), and a text message
with the question of user A. User B can then look for
new messages and import them. The figure to be
discussed can be highlighted in the respective report.
B answers the question and A will receive the
appropriate text, which can be linked to the report in
the same way or just be read via the network user
interface. Figure 4 shows a very basic example on
how the communication looks like. In our prototype,
in fact the dataset is as simple and therefore easy to
understand and to implement.
Figure 4: Exemplary Question from user to user.
5 CONCLUSIONS
While BI is one of the most thriving concepts in
today’s enterprises and OSNs, and social media in
general, are vigilantly observed, the combination of
these two is mostly reduced to using networks as
another data source. Then again, collaborative BI
gets more and more attention as today’s employees
use mobile devices and social networks on their own
and in their daily work.
Future research should focus on the question
how a bigger model of information, data and meta
data sharing could look like. Unified data models or
very flexible peer-to-peer architectures are aspects
that are already being discussed. The question also
still stands, how missing data can safely be
transferred. Last not least, security issues will have
to be discussed. While the communication itself can
be encrypted by SSL connections, the data could be
client-side encrypted to prevent third parties from
understanding possibly captured data.
We showed that the usage of already existing
structures can ease up the process of information
sharing and that the necessary means for this only
lead to small efforts. Future work will show, if
OSNs can provide even more support to the decision
making process.
REFERENCES
Berthold, H., Rösch, P., Zöller, S., Wortmann, F.,
Carenini, A., Campbell, S., et al. (2010). An
architecture for ad-hoc and collaborative business
intelligence. In EDBT '10 Proceedings of the 2010
EDBT/ICDT Workshops .
Bitterer, A. (2012). Hype Cycle for Business Intelligence,
2012. Gartner RAS Core Research Note G00227572,
Böhringer, M., & Helmholz, P. (2011). “What are they
Thinking?” - Accessing Collective Intelligence in
Twitter. In BLED 2011 Proceedings .
Collins, H. (2001). Corporate portals: revolutionizing
information access to increase productivity and drive
the bottom line: Amacom Books.
Costa, P. R., Souza, F. F., Times, V. C., & Benevenuto, F.
(2012). Towards integrating Online Social Networks
and Business Intelligence. International Conference
on Web Based Communities and Social Media 2012,
Dayal, U., Vennelakanti, R., Sharma, R., Castellanos, M.,
Hao, M., & Patel, C. (2008). Collaborative Business
Intelligence: Enabling Collaborative Decision Making
in Enterprises. In Lecture Notes in Computer Science
(pp. 8–25).
Gluchowski, P. (2001). Business Intelligence - Konzepte,
Technologien und Einsatzbereiche. HMD Praxis der
Wirtschaftsinformatik, (222), 5–15.
Golfarelli, M., Mandreoli, F., Penzo, W., Rizzi, S., &
Turricchia, E. (2011). BIN: Business intelligence
networks. Business Intelligence Applications and the
Web, IGI Global, 244–265.
Hinchcliffe, D., & Kim, P. (2012). Social Business By
Design: Transformative Social Media Strategies for
the Connected Company: Jossey-Bass.
Kemper, H.-G., Baars, H., & Mehanna, W. (2010).
Business Intelligence (3rd ed.). Wiesbaden:
Vieweg+Teubner Verlag, GWV Fachverlage GmbH.
Liu, L., & Daniels, H. (2012). Towards a Value Model for
Collaborative, Business Intelligence-supported Risk
Assessment. In Proceedings of the 6th International
Workshop on Value Modeling and Business Ontology
(VMBO 2012). Vienna.
Muntean, M. (2012). Business Intelligence Approaches.
Mathematical Models & Methods in Applied Sciences,
Vol. I, 192–196.
Power, D. J. (2011). A Brief History of Decision Support
Systems, version 4.1. Retrieved September 27, 2012,
from http://dssresources.com/history/dsshistory.html.
Rasmussen, R. (1999). SAS Institute Releases SAS
Collaborative Server. Retrieved September 26, 2012,
from http://www.information-management.com/news/
1435-1.html.
Roe, C. (2011). Business Intelligence 3.0 – Social
Analytics Part 1. Retrieved September 27, 2012, from
http://www.dataversity.net/business-intelligence-3-0-
social-analytics-part-1/6309/.
Vetschera, R. (1991). Entscheidungsunterstützende
Systeme für Gruppen: Ein rückkopplungsorientierter
Ansatz. Physica-Schriften zur Betriebswirtschaft: Vol.
35. Heidelberg: Physica-Verl.
<server>localhost/demo
<cube>MyCompany_Sales
<row>product|ACMEprod
<row>region|USA
<col>year|2012
<message>Hi John, can you explain to me why
we lost 5% of our revenue here?
WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies
128
