10.1 Expert Systems
Computer systems that emulate human expert decision-making:
- Knowledge Base: Facts and rules about the domain
- Inference Engine: Applies rules to solve problems
- User Interface: How users interact with the system
Examples: Medical diagnosis, financial planning
10.2 Natural Language Processing (NLP)
Techniques for understanding and generating human language:
| Task | Description | Example |
|---|---|---|
| Tokenization | Splitting text into words/tokens | "Hello world" → ["Hello", "world"] |
| POS Tagging | Identifying parts of speech | "run" → verb or noun? |
| NER | Named Entity Recognition | "Apple in Cupertino" → ORG, LOC |
| Sentiment Analysis | Determining emotional tone | "I love this!" → positive |
10.3 NLP Pipeline
- Text Preprocessing (cleaning, normalization)
- Feature Extraction (TF-IDF, word embeddings)
- Model Training (classifiers, neural networks)
- Evaluation (accuracy, precision, recall)
10.4 Modern NLP Techniques
- Word Embeddings (Word2Vec, GloVe)
- Transformer Models (BERT, GPT)
- Attention Mechanisms
- Transfer Learning in NLP
10.5 Frequently Asked Exam Questions
- Explain the architecture of an expert system.
- Compare rule-based vs machine learning approaches in NLP.
- What are word embeddings and why are they useful?
- Describe the transformer architecture.
- How does sentiment analysis work?