Knowledge and Language

Recommended Prerequisites

Knowledge on Intelligent Agents; Machine Learning; Programming; Mathematics and Statisticts; reading and writing in English.

Teaching Methods

The course will include two main activities:

- Presentation and discussion of concepts in theoretical classes.

- Research work and practical project, monitored in laboratory classes.

The work must delve deeper into a (sub)topic of the program, through reading and analysis of scientific literature. In the practical project, some concepts will be applied to a medium-sized practical challenge. Ideally, they will be complementary, addressing the same (sub)topic. The main findings must be described in a report to be presented orally to colleagues and teachers.

Learning Outcomes

The course targets the research and discussion of the role of humans in Artificial Intelligence (AI), namely as a producer of data and as a consumer of results, then studying specific human aspects of AI.

It is expected to acquire knowledge about:

(1) Human-interpretable AI techniques, namely: symbolic methods of logic, knowledge representation and sharing, and probabilistic methods for dealing with situations of uncertainty;

(2) Types of interaction between humans and AI, focusing on natural language processing, including associated challenges, as they are addressed by AI, and common applications;

(3) Leveraging symbolic and probabilistic methods for explaining or complementing more complex methods, based on black boxes, in the so-called Explainable AI and Neuro-symbolic AI.

Work Placement(s)

No

Syllabus

1. Human-AI Interaction
1.1 Types of Interaction
1.2 Humans as producers of data and knowledge
1.3 Data Annotation
1.4 Crowdsourcing
2. Structural Knowledge Representation
2.1 Logic fundamentals
2.2 Knowledge Structures
2.3 Representation Languages
2.4 Querying Structural Knowledge
2.5 Sharing: Linked Data
3. Uncertain Knowledge
3.1 Quantifying Uncertainty
3.2 Probabilistic Reasoning
3.3 Probabilistic Reasoning over Time
4. Natural Language Processing Foundations
4.1 Challenges in Human Communication
4.2 Linguistic Knowledge
4.3 Typical AI Problems
4.4 Embedding Text
4.5 Language Modelling
5. Natural Language Processing Applications
5.1 Sentiment Analysis
5.2 Knowledge Extraction
5.3 Question Answering
5.4 Chatbots
6. Advanced Topics:
6.1 Human-Interpretable AI
6.2 Approaches for Explainability
6.3 Neuro-symbolic AI.

Head Lecturer(s)

Hugo Ricardo Gonçalo Oliveira

Assessment Methods

Assessment
Research work: 20.0%
Exam: 30.0%
Project: 50.0%

Bibliography

-- Russell, S. & Norvig, P. (2021). Artificial Intelligence: A Modern Approach, Global Edition, 4th Edition, Pearson Education Limited.

-- Liyang Yu (2014). A Developer's Guide to the Semantic Web, 2nd edition. Springer.

-- Jurafsky, D. & Martin, J. H. (2024). Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics and Speech Recognition. Pearson Education International, 3rd edition draft.

-- Molnar, C. (2020). Interpretable Machine Learning. Lulu.com.