Problem Description

In an era where information is abundant and often polarizing, the ability to extract, analyze, and evaluate arguments from textual data is more crucial than ever. Argument mining is a subfield of natural language processing (NLP) that focuses on identifying and structuring argumentative components such as claims, premises, and conclusions in written or spoken discourse. With applications in legal texts, online debates, policy-making, and misinformation detection, argument mining enhances critical thinking and decision-making by uncovering the underlying logical structure of arguments.

This project is triggered by the increasing need to automate argument identification in large-scale textual datasets, such as social media discussions, news articles, or research papers. Traditional manual annotation of argumentative structures is labor-intensive and infeasible for big data applications. By developing computational techniques to extract and classify arguments, we can improve debate analysis, fact-checking, and AI-driven reasoning systems.

Project Definition

This project guides students in building a modular argument mining pipeline using NLP and machine learning. Key activities include:

1. Understanding Argumentation Basics of claims, premises, and rebuttals
2. Dataset Exploration Working with labeled argument mining datasets
3. Preprocessing & Feature Engineering Applying NLP techniques (e.g., TF-IDF, embeddings, dependency parsing)
4. Argument Classification Training machine learning models to identify argument structures
5. (optional). Relation Extraction Running ML models to identify attack and support relationships between arguments
6. Evaluation & Refinement Measuring model performance and optimizing results
7. Architectural concept and technology selection for the modular argument mining pipeline
8. Implementation, hosting, and test of the pipeline
9. Real-life Showcase Definition of a suitable demo use case (live debate, social discussion, legal debates, )
10. Application & Visualization Demonstrating results through an interactive tool or visualization.

By the end, students will have a functional argument mining system. The students will focus on specific artifacts, depending on the chosen project type (ACS or GAK).

Learning Outcome

Students will:

• Understand argumentation structures in text
• Gain hands-on experience in NLP preprocessing and feature engineering
• Train and evaluate machine learning models for argument classification
• Apply argument mining in real-world contexts like misinformation detection
• Design & implement a suitable software architecture for a real-world AI application

Participation Requirements

Students in the GAK project flavor should have:

• Basic Python programming skills
• Willingness to make yourself familiar with NLP concepts (e.g., tokenization, embeddings)
• Some experience with machine learning (recommended but not mandatory)
• Interest in computational linguistics and AI applications

Students in the ACS project flavor should have:

• Some familiarity with designing and implementing build pipelines and system (hosting) architectures
• Programming skills in Python and other relevant languages
• Interest in computational linguistics and AI applications
• Willingness to make yourself familiar with NLP concepts (e.g., tokenization, embeddings)
• Openness for working in an interdisciplinary student team

This project is ideal for those interested in NLP, Machine Learning, argument analysis, and AI application architecture.

External Partner

If there is an external partner, give the contact person(s) and a brief description here; otherwise leave blank