Course 2025-2026 a.y.

Neural Network Foundations: The course introduces fundamental neural network architectures, backpropagation algorithms, and optimization techniques, with hands-on implementation using Python frameworks. Students will build basic neural networks and understand their mathematical foundations for subsequent applications in economic and social science contexts.

Word Embeddings in Social Science: Students will explore vector representations of language, covering word2vec, GloVe, and contextual embeddings. Technical implementation focuses on applying these methods to economic texts, policy documents, and social media data to extract quantitative insights from qualitative information.

Neural Embeddings for Marketing Research: Building on embedding techniques, this section demonstrates practical applications in consumer behavior analysis, brand positioning, and market segmentation. Students will develop custom embedding models for marketing datasets and implement clustering and similarity algorithms for business intelligence.

Large Language Model Architecture and Implementation: Deep dive into transformer architectures, attention mechanisms, and pre-training methodologies. Students will work with state-of-the-art models (GPT, BERT, Claude) through APIs and fine-tuning techniques, developing custom applications for economic research and data analysis.

Text Data Mining and Insight Extraction with LLMs: Students will master advanced techniques for extracting structured insights from unstructured text using large language models. Topics include automated content analysis, theme identification, entity extraction, and sentiment classification applied to economic datasets such as earnings calls, policy documents, and consumer reviews. Programming projects will focus on building scalable text analysis pipelines.

Societal Impact Assessment: Analysis of LLM effects on labor markets, information systems, and economic structures. Students will conduct empirical studies on AI's economic impact, including automation effects, productivity changes, and market disruptions using computational methods and econometric analysis.

• Understand the fundamental principles of neural networks and transformer architectures that power large language models
• Explain how word embeddings and vector representations capture semantic meaning for computational analysis of text data
• Describe the technical methods for extracting structured insights from unstructured text using large language models and NLP techniques
• Analyze the economic and societal impacts of AI deployment, including effects on labor markets and information systems
• Apply programming techniques for implementing and fine-tuning large language models using APIs and Python frameworks for economic research applications

• Implement neural network models and large language model APIs using Python to solve real-world economic and business problems
• Process and analyze large-scale text datasets to extract quantitative insights for economic research and market analysis
• Design custom text mining pipelines that transform unstructured data into structured information for decision-making purposes
• Develop AI-powered applications for specific economic research questions, including sentiment analysis, topic modeling, and content classificatio

• Lectures
• Practical Exercises
• Individual works / Assignments

• Practical Exercises
  • In-class exercises
• Individual works / Assignments
  • Project Work
  • Project Presentation

For each topic in the course, we will combine lecture with hands-on exercises. Students will have opportunity to work with data to practice in LLM knowledge and techniques.

• Participation (20%) 
  • Engagement and In-class Exercise.
• Assignments (30%) 
  • Multiple assignments to help students LLM techniques.
• Final Exam (50%) 
  • Test on both conceptual knowledge and programming skills learnt in this course.

Attendance will be registered at the beginning of all the sessions. To get the attending student status, students should be present in at least 75% of the lessons.

• Final Exam (100%)

Course materials posted on Blackboard.