News item
UB Ontology Scholars explore avenues for collaboration with the Toronto group
from the INSIGHTS Newsletter of the UB Institute for Artificial Intelligence and Data Science
In 2005, the NIH established the $60 million National Center for Biomedical Ontology as a multi-site consortium with three US partners: the Stanford University Department of Medical Informatics, the Mayo Clinic, and the UB Department of Philosophy. Since then, UB has established itself as an important center for ontological research, involving not only the Departments of Philosophy and Geography but also the Division of Biomedical Ontology in UB’s Department of Biomedical Informatics.
The idea of using ontologies for computational purposes was born in Stanford in the 1970s as one strand in the development of AI and of what is called ‘knowledge representation’. An ontology is, in simple terms, a representation of the types of entities in reality and of the relations between such entities. An ontology is distinguished from a database in that it codifies a body of knowledge in a semantically precise way. Ontologies so defined can be used as a means of regimenting the ways in which entities in the world are represented for example in order to make data deriving from different sources more easily discoverable and integratable. Ontologies can thereby play a role in allowing data to be combined in such a way as to provide larger quantities of representative sample data for purposes of training a neural net.
Large numbers of ontologies have been developed in the last 30 years. But these were typically one-off constructions, useful to their authors in relation to given bodies of locally collected data, but of no lasting significance. The UB work in ontology is, however, well known for its contribution to creating modular suites of ontologies that are deliberately built to cover a broad array of different sorts of data in a way which allows them to be of lasting utility to broader audiences. This idea led to the creation in 2021 of a new international standard (ISO/IEC 21838) based on the methodology used in UB since around 2005 for the building of such ontology suites.
As the documentation of ISO/IEC 21838 points out, however, the first suite of modular ontologies was in fact built already by 1998 under the name “Toronto Virtual Enterprise” or “TOVE”, by a group of ontologists working in the University of Toronto. The group is headed by Mark Fox, Distinguished Professor of Urban Systems Engineering at the University of Toronto, (pictured above) director, there, of the Urban Data Centre in the School of Cities, the Centre for Social Services Engineering, and the Enterprise Integration Laboratory. On October 24 Fox led a team of Toronto ontologists on a visit to Buffalo for the first Ontology Day encounter.photograph of a building The ontologists in Toronto have extended the ontology method to address data relating to a wide range of issues in the social world, and they have developed ontologies that have proved useful to the solving of problems especially on the part of government agencies. One special emphasis is on problems at the city level, where their work gave rise to the ISO/IEC 5087 series of City Data Model standards.
An example is the Toronto Common Data Model Project, whose goal is to create an ontology-based framework for the communication of city data among Toronto city divisions and departments as well as external stakeholders. A second example is the Toronto Water Asset Management Ontology Project, which is developing an ontology for modeling water and sanitation physical assets and for management of the associated documentation. Further examples include work within Common Approach, a consortium of over 100 organizations devoted to exploring the use of digital technology to enhance social service provision broadly conceived through the use of a common Social Services Sector Ontology.
A particular contribution within Common Approach is in the area of impact modeling. The Toronto ontology group addressed specifically the question of how data can be exploited to gauge the impact of social service provision by creating a data standard for representing impact models and reporting on social service performance, via measured outcomes and indicators. This standard is used to develop a natural language understanding architecture called SeMantIc roLe Extraction (SMILE), which relies on a hybrid of symbolic and subsymbolic methods. SMILE extracts ontological concepts and roles from unstructured text describing real-life services. The extracted knowledge can then be used to evaluate the impact of a real-life network of service providers by simulating service provisioning using AI planning and scheduling methods.
Finally, in 2021 the ontologists in Toronto joined the Compass project, an initiative funded by Canada’s Digital Technology Supercluster to deliver a technology platform that leverages AI to streamline the experience of those looking for help and provide better insights for those delivering or funding different kinds of support. Building on its previous work, the group has produced the Compass suite of ontologies, which together form a framework for representing a wide range of topics, from the needs of clients and communities, the need satisfiers and the services that deliver them, and the insights that can be gained via data analytics. The Compass suite of ontologies also serves as the basis for the Social Needs Marketplace (SNM), a core component of the Compass project. SNM is an AI-powered platform that matches supply to demand in order to ensure better service planning, coordination, and delivery to maximize client outcomes and community impact.
The follow-up match between the Toronto and UB ontologist teams will take place in Toronto on December 5, when the two teams will explore concrete avenues for collaboration, including the use of common upper-level ontology standards and coordination of the Toronto ontology work with the contributions of UB ontologists in areas such as government policy and industrial manufacturing. UB scholars working in any of the areas mentioned in the above are invited to contact Barry Smith (phismith@buffalo.edu) with a view to expanding the range of possible collaborations.
Click here for further details
