information sharing on top of BI tools that include
data warehouses, analytical tools and reporting tools.
CBI enables organizations to gain timely access to
quality information and competitive advantages.
This paper introduces our CBI framework, which
is an accessible, open-source BI platform that
implements the data warehousing process in
Software-as-a-Service (SaaS) mode. CBI users
connect to the platform where they can share and
explore data, help each other to construct data cubes,
and formalize different visualizations via traditional
graphs and charts. CBI users can talk with each
other, build collaborative analysis, comment and
annotate their findings along the compelling graphs
and finally store their analysis in the dashboard. We
also design a CBI ontology that stores annotations,
rather than a database or text files. Ontology-based
storage of annotations indeed makes them machine-
processable, interpretable and they enable high
precision when searching and retrieving knowledge.
The remainder of the paper is organized as
follows. Section 2 presents related work on CBI.
Section 3 details our CBI platform. Finally, Section
4 concludes this paper and hints at future research.
2 RELATED WORKS
There is diverse work in the field of CBI. We
classify the existing literature on CBI into several
categories. The first category consists of research
works focused on collaborative query management.
Giacometti et al. (2011) recommend
multidimensional queries based on a distance
measure calculated by comparing log sessions and
the current session. However, they do not consider
user context or preferences, which are addressed by
Eirinaki et al. (2014). Their system is based on a
lookup mechanism where similar users and queries
help recommend queries to the current user.
Khoussainova et al. (2011) propose another research
work for auto-completion and query management.
They develop a context-aware system that helps
novice users formulating SQL queries. Their system
does not recommend complete queries, but possible
additions to various clauses in the user’s queries.
Sapia (2000) differs from these approaches, as they
use predictive prefetching in on-line analytical
processing (OLAP) to minimize query execution
time. They use a Markov model based on the user’s
multidimensional data. Moreover, Golfarelli et al.
(2012) enhance the decision making process by
sharing knowledge and operational data across
networks of peers. They develop a language for
semantic mappings between multidimensional
schemata of peers, as well as a query reformulation
framework.
The second category we propose consists of
research works that focus on OLAP session analysis
for preference-based recommendations within BI
platforms. It may be difficult to differentiate them
from the first category, but they actually focus on
analyzing previous sessions and not recommend
single OLAP queries. Jerbi et al. (2009) investigate
the interactive and navigational nature of user query
behavior. They extract relevant elements from the
user profile and enrich the query answer before
presenting it to the end-user. Next, Aligon et al.
(2015) present a collaborative filtering approach that
recommends OLAP sessions by analyzing previous
sessions, while not recommending single OLAP
queries. Their work is unique in the sense that they
compare sessions rather than queries. Aufaure et al.
(2013) also use recent analytical sessions to
recommend queries based on a probabilistic user
behavior model and query similarity metrics. They
aim to reduce latency time by the use of a cache
manager that prefetches objects. Wu et al. (2007)
combine the aggregation power of OLAP and
keyword-driven analytical processing. They develop
scalable algorithms for subspace generation, novel
ranking and dynamic facet construction. Cabanac et
al. (2007) develop relational OLAP to add
annotations on multidimensional data. Such
annotations enable decision-makers to share and
communicate data among all collaborators. This
work is very close to ours, but we store annotations
in an ontology rather than a database. Ontology-
based storage of annotations makes them machine-
processable, interpretable and they enable high
precision for searching and retrieving knowledge.
Ontology-based dashboard let different collaborators
to get a better understanding of the data by making it
easier for them to find, manipulate and access
information related to collaborative sessions held
between different collaborators.
3 COLLABORATIVE BUSINESS
INTELLIGENCE PLATFORM
This section details our CBI platform, which is
accessible, open source and free. It implements the
data warehousing process in Software-as-a-Service
mode. In the following subsections, we first present
the architecture of our CBI Framework. Then, we
discuss its implementation and elaborate a use case