DE4A Machine Learning for semantic interoperability

In De4a context, we will investigate the possibilities of applying machine learning algorithms to understand the usage patterns of the services and the potential of self-emerging ontologies from a collaborative tagging system to be used as an information classification structure that can be updated dynamically and continuously. The idea is not to automatically derive a fully functional ontology but expanding and integrating it with concepts and terms identified by user participatory data. For instance, analytic data, data that track the integrated e-services will be derived from piloting of the use cases, and machine learning algorithms can be applied upon these data to establish, for example, new tags and semantics upon the services utilised. Usage data can be used to draw weights in order to reinforce or deactivate initially designed relationships among properties within classes in the original ontologies. In other words, data can feed machine learning algorithms to adapt the ontology accordingly. As a result, new classes and properties may appear, synonyms may be identified, relationships among classes and properties may change. Special focus will be put on investigation of the potential of machine learning and self-emerging ontologies in providing semantic interoperability within European eGovernment networks. The idea is that the semantics themselves (e.g. ISA2 Core Vocabularies) are to be dynamically enhanced by machine learning algorithms on (training) data. These data will be initially related to public service usage data.This page describes the work in progress on machine learning methodologies for semantic problems.

Automated Vocabulary search

In progress...