Text and Argument Mining

Learning objectives

The Text Mining Module (3 cfu) provides an introduction to the issue of extracting meaningful information from texts, which is at the basis of several applications of Natural Language processing, including text classification and clustering, opinion and argument mining. After briefly introducing the basic pre-processing phases, with particular emphasis on the tokenizing phase, the main techniques for part-of-speech tagging and entity recognition will be introduced, as well as the issue of dependency parsing. The tasks of opinion mining and sentiment classification, as well as the issue of unsupervised classification of texts, will be introduced, and the main techniques to address them will be presented. The students will learn the basic tasks and techniques to make sense of textual content.

The Argument Mining module (3 cfu) provides an introduction to formal argumentation theory, a key area within symbolic AI that focuses on mimicking human reasoning and debate processes. By examining formal models of argumentation, students will learn about their applications to multi-agent systems and common-sense reasoning. By integrating argumentation theory into their studies, students gain valuable skills that are directly applicable to the development and improvement of explainable AI systems, an interdisciplinary approach that ensures that AI not only makes decisions accurately but also communicates them transparently.

Expected learning outcomes

The Text Mining Module (3 cfu) provides basic knowledge and technical skills in the following topics:
· Basic Text Processing: Tokenization, Normalization, Stemming, Lemmatization
· Methods of tokenization
· Parts of Speech and Named Entities recognition
· Dependency Parsing
· Opinion Mining
· Sentiment classification
· Text Clustering
· Topic Modelling

Skills acquisition and ability to apply knowledge:

At the end of the module, students are expected to be able to:
- Apply to texts the main preprocessing phases;
- Coping with the issues of entity recognition and part-of-speech tagging;
- Understanding the task of unsupervised text classification, and the main related techniques;
- Deepening the tasks of opinion and sentiment mining

The Argument Mining module (3 cfu) provides basic knowledge and technical skills in the following topics:
· Argumentation and argument mining;
· Tulmin's argumentative model;
· Abstract argumentation theory (Dung-style);
· Argumentative semantics;
· Bipolar argumentation frameworks;
· Structured argumentation frameworks;
· Weighted argumentation frameworks;
· Dynamics of argumentation.


Skills acquisition and ability to apply knowledge:

At the end of the module, students are expected to be able to:
- Define the argumentative frameworks introduced;
- Compute the main argumentative semantics;
- Evaluate a given argumentative framework;
- Formalize realistic reasoning problems in terms of argumentation frameworks.