C8891 - Cognitive Semantics for Artificial Intelligence (1) (LM)

Learning outcomes

The course introduces logical, cognitive and computational methods to deal with multimodal data and content, and their explicit and implicit interpretations. The methods address different ontological levels. A 4E (embodied, enactive, embedded, extended) semiotic approach will be assumed. At the end of the Course students will be able to master the basic usage of state-of-the-art neurosymbolic artificial intelligence to characterize and resolve complex problems. The acquired competency will be expanded and evaluated within a dedicated project.

Course contents

The course is divided into seven modules, delivered in ten three-hour slots that alternate lectures and laboratory exercises. Each module allows the production of an intermediate artifact that can be reused in the exam project.

1. Semiotic and cognitive foundations of interpretation (1 slot). Sign, meaning, and levels of interpretation. Why linguistic semantics does not exhaust the interpretation of content. Semantic universals. Introduction to the 4E program in cognitive neuroscience and its implications for neurosymbolic AI and knowledge engineering.

2. Cognitive semantics: schemas, metaphors, simulation (2 slots). Image schemas and their sensorimotor bases; metaphor and conceptual blending; neural theory of language and simulation semantics. Critical discussion of supporting evidence and objections. Reference: Lakoff & Narayanan (2025).

3. Ontological levels and knowledge patterns (2 slots). Foundational ontologies and categorical distinctions; Descriptions and Situations as a mechanism for reifying context; Ontology Design Patterns and their relationship with cognitive patterns. Workshop: Modeling a domain fragment from competency questions. Reference: Gangemi (2020).

4. Computational frame semantics and knowledge graph extraction (2 slots). From FrameNet to integrated frame-based resources; machine reading and knowledge graph extraction from text; lexical-ontological alignment and ambiguity management. Workshop: Graph extraction and auditing from a corpus. Reference: Gangemi et al. (2017).

5. Perspectives, stances, and values (1 slot). Formal representation of point of view: factuality, attribution, subjectivity, moral evaluation; automatic extraction of perspectives and its applications to public discourse analysis. Reference: Gangemi & Presutti (2022).

6. Neurosymbolic architectures (2 slots). Neurosymbolic enrichment of graphs and grounded world models; Logic-Augmented Generation: differences from RAG, the role of deductive reasoning, hallucination control, inference traceability. Steering LLMs as pipeline components rather than oracles. Workshop: Experiment design and auditing. References: De Giorgis et al. (2025); Gangemi & Nuzzolese (2025).

7. Evaluation, Epistemology, and Design Workshop (1 slot + supervision). What counts as validation of a cognitively inspired model: negative competency questions, SPARQL unit tests, SHACL constraints, baselines, and ablations; limits of inference from system performance to cognitive plausibility. Definition and peer review of project proposals.

Prior qualifications (optional): First-order logic, Web engineering, Experimental cognitive science. Module 1 provides a bridge for students with a background in computer science and humanities.

Readings/Bibliography

De Giorgis, S., Gangemi, A., Russo, A. Neurosymbolic graph enrichment for Grounded World Models. Information Processing & Management, 62, 4, 2025, https://www.sciencedirect.com/science/article/pii/S030645732500069X

Gangemi, A. and Nuzzolese, A.G. Logic-Augmented Generation. Journal of Web Semantics, 2025, https://www.sciencedirect.com/science/article/pii/S1570826824000453

Lakoff, George, and Srini Narayanan. The Neural Mind: How Brains Think. University of Chicago Press, 2025. (Lakoff, George, and Srini Narayanan. La mente neurale: Come pensa il nostro cervello. ROI Edizioni, 2025.)

Ciroku, F., De Giorgis, S., Gangemi, A., Martinez-Pandiani, D.S., Presutti, V. Automated multimodal sensemaking: Ontology-based integration of linguistic frames and visual data. Computers in Human Behavior, 107997, 2024, https://drive.google.com/file/d/1_2ju5yg_3vxrQXW-KgP903DhNJ6csLan/view

Gangemi, A., Presutti, V., 2022. Formal Representation and Extraction of Perspectives. Vossen, P. et al. (eds.): Creating a More Transparent Internet: The Perspective Web, 2022, https://www.dropbox.com/scl/fi/qmaoifmxwqfklvixs629n/WebOfPerspectives-GangemiPresutti.pdf?rlkey=e82o49dz4crwgf84yue0eiupv&dl=0

Gangemi, A. Closing the Loop between Knowledge Patterns in Cognition and the Semantic Web. Semantic Web, Volume 11, Issue 1, 2020, Pages 139-151, https://semantic-web-journal.net/system/files/swj2334.pdf

Gangemi, A., Presutti, V., Recupero, D.R., Nuzzolese, A.G., Draicchio, F. Semantic Web Machine Reading with FRED. Semantic Web Journal 8 (6), 2017, 873-893, https://scholar.google.com/citations?view_op=view_citation&hl=en&user=-iVGcoAAAAAJ&citation_for_view=-iVGcoAAAAAJ:dX-nQPao9noC

Teaching methods

The course will be given in 3-hours slots, with mixed lectures and hands-on sessions. Practical experiments will be reproduced or newly designed on specific sessions.

Assessment methods

The final exam will consist of a project implementing a model of cognitively-inspired sense-making on a specific topic agreed with the teacher, and a discussion.

AI can be a useful tool to support individual study with in-depth analysis, summaries, and self-assessment. As for learning assessment, the exam may consist in a section that can exploit substantial use of AI (problem solving, content generation), accompanied by a mandatory critical analysis section.

Students with disabilities and Specific Learning Disorders (SLD)

Students with disabilities or Specific Learning Disorders are entitled to special adjustments according to their condition, subject to assessment by the University Service for Students with Disabilities and SLD. Please do not contact teachers or Department staff, but make an appointment with the Service. The Service will then determine what adjustments are specifically appropriate, and get in touch with the teacher. For more information, please visit the page:
https://site.unibo.it/studenti-con-disabilita-e-dsa/en/for-students

Teaching tools

Online and desktop apps with artificial intelligence capabilities will be regularly used during the course.

Office hours

See the website of Aldo Gangemi