Big data has gained much interesting in recent years due to the rapid expansion of the massive amount of data that is available for solving different types of tasks within many different application domains. However, today's big data is still on a fairly low level of abstraction when it comes to complex decision support tasks, subject to e.g. high dimensionality and significant portions of uncertainty regarding which patterns to look for in the data.
So far, research on big data has been increasingly busy with focusing on different types of platforms, e.g., Hadoop and Spark, for performing big data analytics utilizing traditional machine learning algorithms.
At the same time, within the research field of information fusion there has been an increase in research regarding exploratory analytics, i.e., finding and understanding patterns in data, and predictive modelling, i.e., estimating future values of some important variable, from a higher-level of abstraction typically involving several unknown and uncertain variables that gives some indication of the current situation, e.g., in surveillance scenarios.
Information fusion in this respect offers a new perspective on big data that can make a substantial contribution. More specifically, the three main aspects of information fusion:
- use of multiple sources for identifying patterns and performing predictions
- modelling of uncertain information and reliability of sources
- modelling of conceptual, non-measurable and high-dimensional variables
are crucial aspects when it comes to exploiting big data. We denote this view of big data as big data fusion. Big data fusion has the capacity to provide intelligence for decision makers that facilitates more complex accounts of problem domains.
Moreover, it has been shown that data-driven decision-making improves organisational performance. However, the role of big data and information fusion in decision support has been very little theorized, making it cross-disciplinary research problem in urgent need of attention.
The project addresses societal and industrial problems in achieving value through improved decision-making from the rapidly increasing volume of digital data. It employs information fusion as the governing decision technology and the principal application areas are bio-informatics and telecommunications. The four-year work programme, examines the research question:
- How can big data fusion enable reliable exploratory and predictive modelling on big data for the purpose of decision support?
The underlying decision support model (problematisation, big data fusion, exploratory analysis, predictive modelling, and interpretation) underpins three scientific sub-projects in Bioinformatics, Business Intelligence and Operational Decision Support. The collaborating partners are the University of Skövde, Takara Bio Europe, AstraZeneca, Advectas, and Huawei and the total budget is 29.8 MSEK, of which the Knowledge Foundation contributes with 14.8 MSEK, partner companies contribute 12.3 MSEK, and the University of Skövde contributes 2.7 MSEK.
The outcomes of the project will be:
- new techniques for exploratory analytics and predictive modelling,
- methods and techniques for big data fusion in bioinformatics, business intelligence, and operational decision support,
- demonstrator solutions which may be used to develop commercially viable, systems, and
- integration of the University of Skövde’s research capacity within decision technology and big data.
BISON is hosting four sub projects:
Synergy: Project leader Göran Falkman
Bioinformatics: Project leader Jane Synnergren
Business Intelligence: Project leader Mikael Berndtsson
Operational Decision Support: Project leader Joe Steinhauer