This laboratory develops technical applications and theoretical foundations for advancing a research that supports the modeling and the usage of knowledge inside artificial systems. Two are the main objectives: i) to provide practical and efficient solutions for enabling artificial agents to address tasks like recognition, classification, semantic interoperability, meaning negotiation and learning; ii) to get closer in developing systems that bridge the gap between today’s state-of-the-art AI solutions and the human-level ability to manipulate knowledge. To reach these objectives, the K-LAB follows an interdisciplinary research approach, on the border between computer science, artificial intelligence, philosophy and cognitive science.
Currently, the activities of the K-LAB can be grouped according to the following research directions:
- Knowledge Recognition. Here the main goal is to provide tools and theoretical foundations for supporting the identification and the integration of multiple information coming from different knowledge resources, such as schemas, ontologies and semi-structured data.
- Knowledge Diversity. Here the main goal is to provide mining and processing solutions, along with evaluation metrics for analyzing and profiling the diversity of existing knowledge resources (e.g., schemas, ontologies and semi-structured data), in order to understand and exploit the usage of knowledge in supporting or guiding specific practical tasks (see knowledge recognition, engineering, evolution or classification).
- Knowledge Engineering. Here the main goal is to develop methodologies and tools to support the representation of knowledge in computer interpretable formats, with a particular focus on the relations between natural language and knowledge, knowledge interoperability, knowledge compositionality and knowledge adaptation.
- Classification and Identification. This research direction, by exploiting machine learning and classical-symbolic AI technologies, explores new perspective on the formalization and implementation of the human ability to acquire and manipulate knowledge, in order to: i) run generalization over multiple domains; ii) draw (new) inferences, and iii) run complex reasoning.