معالجة اللغات الطبيعية

Estimated duration: 36 hours (6 modules, 18 lessons + project)

What you will learn

• Prepare text for modelling: tokenisation, normalisation, Arabic diacritics and orthographic variants, morphology and dialect issues
• Represent text with bag-of-words, n-grams and TF-IDF and measure similarity
• Build and evaluate text classifiers and sequence labellers (NER with BIO tags)
• Explain word and contextual embeddings and the limits of recurrent models
• Explain attention and transformers, subword tokenisation, pretrained models and fine-tuning
• Build LLM-based features responsibly: prompting, retrieval-augmented generation, evaluation, bias, hallucination and privacy

Who it is for: university students and developers building text classification, search, extraction and LLM-based applications

Prerequisites: CRS-GEN-ML-FUNDAMENTALS; Python, NumPy, basic probability