Natural Language Processing - Methods and Applications

People in organizations, society and institutions communicate via natural language. Typically, large amounts of unstructured information are stored in text documents. However, it is very challenging for machines to query relevant content fast and extract structured information.

Machine-based solutions for text analysis, indexing and generation are often summarized under the terms natural language processing, text mining or natural language generation. In the past decades, many different methods and solutions have been developed. The diversity of the different problems and the fast development of new technologies, e.g. machine learning, underline the large potential of these methods. 

The seminar aims to give an overview over the technological building blocks and their applications. The participants independently work on a scientific topic, based on existing literature and present and discuss their findings in a presentation and a seminar paper.

Proposed Seminar Topics:

Foundations of NLP:

1. Word Embeddings: Techniques and Applications
   
2. Language Models: From N-grams to Transformer Architectures
   
3. Attention Mechanisms in NLP: Foundations and Innovations -

Techniques in NLP:

1. Document Classification and Topic Modeling
   
2. Named Entity Recognition (NER) in Natural Language Processing
   
3. Part-of-Speech Tagging and Dependency Parsing: Models, Methods, Evaluation and Applications
   
4. Machine Translation: Approaches and Evaluations
   

Large Language Models:

1. Transfer Learning and Domain Adaptations: Challenges and Solutions 
   
2. Natural Language Generation (Auto Regressive Models): Techniques and Use Cases 
   
3. Prompt-Tuning 
   
4. Finetuning LLMs and Reinforcement Learning from Human Feedback 
   
5. Efficiency and Context Window in Large Language Models 
   
6. Text Summarization: Extractive and Abstractive Approaches 
   
7. Question Answering Systems: Challenges and Approaches 
   
8. Model Hallucination
   

Conversational AI:

1. Task-based & Social Conversational Agents 
   
2. Dialogue Management (Dialogue State Tracking & Policy) 
   
3. Conversational Search Systems 
   

Privacy & Security in Natural Language Processing:

1. Ethical Societal, and Legal Aspects of Large Language Models 
   
2. Differential Privacy in Natural Language Processing 
   
3. Adversarial Attacks in and Privacy Risks of (Large) Language Models 
   

Miscellaneous:

1. Natural Language Inference 
   
2. Explainability in NLP 
3. Knowledge Graphs in NLP: Construction and Applications 
   

Weekly Sessions

Deliverables

Submissions