Student working on UN project to ensure universal understanding of sustainable development goals

Mark Jensen knows how to structure information. Give him gigabytes of data and he’ll organize the disparate elements into a universally accessible whole.

Jensen, a doctoral candidate in the Department of Biomedical Informatics, Jacobs School of Medicine and Biomedical Sciences, uses ontologies to create structured representations of knowledge and information across and within specific disciplines.

“It’s engaging material at a fundamental level and thinking about how the information that’s being generated in a particular course of study can be clearly and precisely defined, organized and accessed,” Jensen explains.

Ontologists come from many backgrounds. Jensen’s training is in cognitive science and philosophy, and his research focuses on the development of ontologies in the domain of mental disorders, cognitive impairment and neurological diseases.

For five years, he has developed biomedical ontologies for UB researchers. Now he is working with the United Nations Environment Programme (UNEP) to provide the information infrastructure needed to monitor the UN’s recently adopted Sustainable Development Goals (SDGs), a set of goals created to help frame policy for the next 15 years on the issues of poverty, inequality and injustice, and climate change.

The term “ontology” evolves from metaphysics, the branch of philosophy that deals with the nature of being. Outside of philosophy, ontology describes ways of linking data within the fields of computer science and information technology — and UB’s new Department of Biomedical Informatics (DBMI) is the first academic unit in the world with a division devoted to ontology, says Barry Smith, SUNY Distinguished Professor of Philosophy and director of the National Center for Ontological Research at UB.

Jensen is pursuing his PhD under Smith’s supervision.

The 17 Sustainable Development Goals, along with their targets and indicators, replace and expand upon the eight Millennium Development Goals (MDGs) that expired at the end of 2015.

The SDGs are a collaborative engagement of all countries that call for private-sector support and are rooted in areas of basic human rights.

Each of the 17 goals has a subset of targets, 179 in all, but the terms used in those goals and their targets often are broad, imprecise and contextual, such as “build,” “ensure” and “promote.”

To solve the resulting language problem, Jacqueline McGlade, a chief scientist with UNEP, assembled a team to develop an ontology for the SDGs.

Jensen started work in the fall with Smith, who, along with Werner Ceusters, chief of the division of biomedical ontology in the DBMI, leads most of the ontology work being conducted at UB. Other members of the UNEP team include Pier Luigi Buttigieg, Alfred Wegener Institute for Polar and Marine Research; Chris Mungall, Berkeley Labs; Lynn Schriml, University of Maryland; and Ramona Walls, iPlant Collaborative.

“The approach to ontologies used by Mark and the UNEP team draws on work that has been taking place in Buffalo for over 10 years, including large-scale ontology building for U.S. Army intelligence, the Air Force and Army Research Labs, the Federal Highways Administration and also UB’s Clinical and Translational Research Center,” says Smith.

There are 193 UN member states and hundreds of interest groups concerned with specific SDGs and representing different cultures and many languages.

Collaboration and integration are among the characteristics separating the SDGs from their MDG predecessors, so a consistent understanding of objectives and the accompanying necessity to uniformly monitor goal progress are critical.

“Polysemy in language makes the task of organizing the data problematic,” says Jensen. “As ontologists working on this project, we have to investigate the use of each term within the goals and targets. Many terms appear multiple times and we have to determine how their meaning varies relative to context. Even the word ‘sustainable’ occurs dozens of times. What does each instance of the term mean for a particular domain?”

And this is not simply a matter of nuanced translation.

A translator might take a work of Homer from its original Greek to another target language. While that provides access to a specific population, it doesn’t create a universal work of literature since no one outside of each chosen target language would be able to read the text. UNEP ontologists, meantime, must be globally comprehensive in their task to create one source of structured information that is accessible and understandable to all of the SDG stakeholders.

One way of structuring information — via relational databases — though well suited to small scale projects is not an option as the data scale grows, as is the case with the type of monitoring relevant to the SDGs, says Jensen.

“Relational databases represent knowledge that is specific to a particular domain. They don’t scale up or handle major revision particularly well, though. With the SDGs, the language used doesn’t always match from one goal to the next,” he says. “So we create needed consensus definitions that link together in a hierarchy as a way of building a common SDG ontology.”

Jensen says a goal in developing the ontology is to create a way of structuring the data by acting as an interface to provide a high level of connection between more specialized domain ontologies that will offer cross-reference capabilities from fields as varied as reducing inequalities (Goal 10) to life below water (Goal 14). This integration captures casual associations that allow information to be shared across disciplines.

“The next stage will be in developing a complete representation for all the targets and their associated data, and beginning to work closely with a team of collaborators from various specialized domains among the member states. This process provides the needed feedback and grounded approach to understand the specific uses and application for each of the targets,” says Jensen.