Big data research flagship project gets R9m boost

The University of KwaZulu-Natal (UKZN) has invested just over R9 million to kickstart the Big Data for Science and Society (BDSS) Flagship Project, one of three such projects to receive financial backing from a programme known as the Research Flagship Programme.

The Research Flagship Programme concentrates on four broad areas: social cohesion, African health, big data and informatics, and African cities of the future.

Professor Deresh Ramjugernath, the UKZN's deputy vice-chancellor for research, says the purpose of the flagship programme is "to promote research that has socio-economic impact and relevance to society and UKZN stakeholders, with the emphasis being on projects that are impactful and implementable, and that spur economic growth and social upliftment".

The BDSS project will support this mission through multi-disciplinary, cross-flagship and cross-college collaboration in big data research that will have a significant scientific and social impact. The benefits of the cross-disciplinary approach include leveraging big data expertise to advance university-wide big data research projects in disciplines where large data sets currently exist, but are not being fully exploited.

Three principal investigators (PIs) will lead the flagship project. They are Professor Kavilan Moodley, representing the field of astrophysics; Professor Maheshvari Naidu, representing Geospatial Humanities; and Professor Onisimo Mutanga, representing Earth Observation. In addition, the project has a number of co-investigators from the fields of computer science and computer engineering, physics, statistics, psychology, management and information systems, and information and communication services.

The BDSS team comprises members from three colleges: the College of Law and Management Studies; the College of Agriculture, Engineering and Science; and the College of Humanities.

The project will deploy interdisciplinary and trans-disciplinary data-driven techniques that will blur the boundaries between disciplines, resulting in scientific and societal benefits spanning a wide variety of topics, ranging from dark energy to climate change and urbanisation, to mobility and migration.

Naidu is a professor in anthropology, as well as the academic leader of research in the School of Social Sciences at UKZN and a PI on the BDSS project. She highlights the interdisciplinary nature of the BDSS Flagship Project by saying: "This particular project cements meaningful collaboration among a wide spectrum of disciplines. It's particularly exciting for aspects of what can be described as computational social sciences.

"One of the initiatives will be to work with colleagues from physics and computer science to explore machine learning (ML) tools in the context of social science narrative data. The aim is to work towards a model that is adapted to heterogeneous and unstructured social science textual data.

"This kind of work will, in turn, allow social scientists to probe social science issues on a scale not imaginable in the past in the context of ML, geographic information systems (GIS) and data mining," she says.

Innovative research in the BDSS Flagship Project will include:

* Utilisation of big data from key astrophysics telescope projects, such as the Hydrogen Intensity and Real-time Analysis eXperiment, known as HIRAX; the Large Synoptic Survey Telescope; and the Karoo Array Telescope, known as the MeerKAT. This, to develop big data pipelines, which will allow scientists to perform cutting-edge science in the topics of dark energy and fast radio bursts.

* Extraction and analysis of big data from Earth Observation sources to conduct invasive species mapping in the context of climate change for better understanding of the ecological impacts of invasive species and the overall impacts of climate change.

* Curation and analysis of big data from Geospatial Humanities to map spatial patterns and human mobility through GIS.

* Utilisation of large data sets from Twitter to develop new, and expand on existing, ML sentiment algorithms that will provide insights into social enactments, as well as race and gender relations.

* Development of domain-specific ML and artificial intelligence (AI) techniques and algorithms that are tailormade to discover patterns in data sets, intelligent data mining and quantum computing techniques for big data analysis.

* Utilisation of statistical clustering and data mining techniques for big data analysis, including spatial clustering and data mining methods that will create a cohesive "picture" that can be used to investigate selected social science phenomena.

Moodley is an associate professor in astrophysics at UKZN, as well as a PI on the HIRAX and a PI on the flagship project. He has extolled the benefits of this substantial project by saying: "The BDSS Flagship Project will support big data research in a number of high-profile astronomy projects that seek to uncover the secrets of the universe. New techniques developed for these projects will also be applied to big data sets in other disciplines, resulting in important scientific and societal advances."

The scientific and socio-economic impact of this research will be magnified through collaboration with a strategic group of partners that will enable the project to be sustainable and broaden the impact of programme outputs. These partnerships and key links include national government organisations such as the National Research Foundation, the South African Radio Astronomy Observatory, the Council for Scientific and Industrial Research, and South African National Parks.

These organisations will support big data research in a multitude of topics and disciplines such as radio astronomy, biodiversity, land use planning and management, and wildlife management.

There will also be partnerships with local government through shared collaborative centres that will promote human capital development through big data innovations and technology spin-offs into industry.

BDSS will also work with a number of multi-disciplinary research centres, including the Centre for AI Research, the Astrophysics Research Centre, and the UKZN-based Andrew Mellon-funded Spatial Humanities Project. These research centres provide data processing, ML and AI expertise and linkages with governments and end users.

Finally, the project boasts a growing number of international partners that currently span nine institutions in six different countries.

Mutanga is a professor in Ecological Remote Sensing and GIS at UKZN, as well as the SA Research Chairs' Initiative Chair in land use planning and management, and a PI on the flagship project. He says: "The direction pointed to by the fourth industrial revolution when it comes to big data research opens up avenues for integrating remote sensing data, advanced climate information, crop/agro hydrological models and ML tools for early warning systems on environmental perturbations. This will help us to make informed decisions on climate risk and adaptive land use management."

In addition to the societal and scientific impacts resulting from the research, the flagship's projects are also geared at developing local staff and students. BDSS will focus on equity and gender transformation of students, postdoctoral fellows and senior researchers, and support the funding of additional postdoctoral fellows and graduate students to work together across disciplines and with the project's many external partners.

This cross-disciplinary co-supervision and collaboration model will promote skills development in both big data techniques and their application across disciplines. The training of trans-disciplinary PhD-level students and postdoctoral fellows will have a long-term impact on society by boosting student entrepreneurship, increasing job growth and raising the local and national gross domestic product.

Words: Raylene Captain-Hasthibeer and Carolyn Crichton