Machine Learning Natural Language Processing Syllabus

Natural Language Processing syllabus — full chapter and topic list

Expand any chapter to see its topics and sub-topics. This is the whole examinable outline for Natural Language Processing in Machine Learning, not a summary of it.

1. 1

   Introduction to NLP

   3 topics

   • Definition and Applications
   • History of NLP
   • Challenges in NLP
2. 2

   Text Preprocessing

   6 topics

   • Tokenization
   • Stop Words Removal
   • Stemming and Lemmatization
   • Text Normalization
   • Part-of-Speech Tagging
   • Named Entity Recognition
3. 3

   Feature Extraction

   6 topics

   • Bag of Words
   • TF-IDF
   • Word Embeddings
   • Word2Vec
   • GloVe
   • FastText
4. 4

   Language Models

   7 topics

   • N-grams
   • Recurrent Neural Networks (RNNs)
   • Long Short-Term Memory Networks (LSTMs)
   • Gated Recurrent Units (GRUs)
   • Transformers
   • BERT
   • GPT
5. 5

   Text Classification

   6 topics

   • Naive Bayes Classifier
   • Support Vector Machines
   • Decision Trees
   • Random Forests
   • Neural Networks
   • Evaluation Metrics
6. 6

   Sequence-to-Sequence Tasks

   4 topics

   • Machine Translation
   • Text Summarization
   • Question Answering
   • Chatbots
7. 7

   Advanced NLP Techniques

   5 topics

   • Attention Mechanisms
   • Self-Attention
   • Transfer Learning in NLP
   • Pre-trained Models
   • Fine-Tuning Models
8. 8

   NLP Libraries and Tools

   6 topics

   • NLTK
   • spaCy
   • Gensim
   • Hugging Face Transformers
   • OpenNLP
   • Stanford NLP
9. 9

   Ethics and Bias in NLP

   4 topics

   • Bias in Data
   • Bias in Models
   • Mitigating Bias
   • Ethical Considerations
10. 10

    NLP in Practice

    4 topics

    • Building NLP Pipelines
    • Deploying NLP Models
    • Case Studies
    • Future Trends in NLP

Natural Language Processing flashcards for Machine Learning

21 of 50 cards from the Natural Language Processing deck — real questions with worked answers.

1. What is Natural Language Processing (NLP)?

   NLP is a subfield of artificial intelligence and computational linguistics concerned with enabling computers to understand, interpret, generate, and manipulate human (natural) language in both text and speech form.

2. Name four common real-world applications of NLP.

   Machine translation, sentiment analysis, chatbots/virtual assistants, spam detection, speech recognition, text summarization, and question answering (any four).

3. In NLP, what is the difference between Natural Language Understanding (NLU) and Natural Language Generation (NLG)?

   NLU focuses on machine reading comprehension — extracting meaning, intent, and structure from input language. NLG focuses on producing coherent, human-readable language as output from structured data or internal representations.

4. What landmark 1950 paper/test is considered a starting point of NLP and AI, and who proposed it?

   The Turing Test, proposed by Alan Turing in his paper 'Computing Machinery and Intelligence' (1950), which judged machine intelligence by its ability to converse indistinguishably from a human.

5. What was ELIZA, and when was it created?

   ELIZA was an early rule-based chatbot created by Joseph Weizenbaum at MIT in 1966 that simulated a Rogerian psychotherapist using simple pattern matching and substitution rules.

6. Describe the three major historical paradigms of NLP in order.

   1) Symbolic/rule-based NLP (1950s–1980s, hand-crafted grammar rules); 2) Statistical NLP (late 1980s–2000s, probabilistic models learned from corpora); 3) Neural/deep learning NLP (2010s–present, embeddings, RNNs, and Transformers).

7. Why is ambiguity a central challenge in NLP? Give the main types.

   Natural language allows multiple interpretations. Key types: lexical ambiguity (a word with multiple meanings, e.g. 'bank'), syntactic/structural ambiguity (multiple parse trees), and semantic/referential ambiguity (unclear pronoun reference or meaning).

8. What linguistic phenomenon does the sentence 'I saw the man with the telescope' illustrate?

   Syntactic (structural) ambiguity — it is unclear whether 'with the telescope' attaches to the act of seeing or describes the man.

9. List three challenges in NLP beyond ambiguity.

   Context dependence, sarcasm/irony detection, handling of synonyms and polysemy, language diversity and low-resource languages, spelling/grammar errors, idioms, and coreference resolution (any three).

10. What is tokenization in NLP?

    Tokenization is the process of splitting raw text into smaller units called tokens (typically words, subwords, or characters), which serve as the basic units for further processing.

11. Distinguish word tokenization from subword tokenization, and name a subword algorithm.

    Word tokenization splits text into whole words; subword tokenization breaks words into smaller meaningful units to handle rare/unknown words and reduce vocabulary size. Example algorithm: Byte-Pair Encoding (BPE), WordPiece, or SentencePiece.

12. Why can tokenization be difficult in languages like Chinese or Japanese?

    These languages typically do not use whitespace to separate words, so word boundaries must be inferred algorithmically rather than read off delimiters.

13. What are stop words, and why are they often removed?

    Stop words are extremely common words (e.g. 'the', 'is', 'and', 'of') that carry little discriminative meaning. They are removed to reduce dimensionality and noise, focusing on more informative content words.

14. Give one situation where removing stop words is harmful.

    In tasks like sentiment analysis or machine translation, words such as 'not' or 'no' are critical for meaning; removing them can flip or destroy the intended sense (e.g. 'not good' becoming 'good').

15. What is stemming?

    Stemming is a heuristic process that crudely chops word endings to reduce inflected words to a common root (stem), which may not be a valid dictionary word (e.g. 'studies' -> 'studi').

16. What is lemmatization, and how does it differ from stemming?

    Lemmatization reduces a word to its dictionary base form (lemma) using vocabulary and morphological analysis, considering part of speech (e.g. 'better' -> 'good'). Unlike stemming, it produces valid words and is more accurate but computationally heavier.

17. Name a classic stemming algorithm for English.

    The Porter stemmer (Porter stemming algorithm); other examples include the Snowball (Porter2) and Lancaster stemmers.

18. Apply stemming vs lemmatization to the word 'caring'.

    Stemming typically yields 'car' (incorrect crude chop in some stemmers) or 'care'; lemmatization (with verb POS) yields 'care', the correct base form.

19. What is text normalization in NLP?

    Text normalization transforms text into a single canonical form, including lowercasing, removing punctuation/special characters, expanding contractions and abbreviations, converting numbers/dates to standard forms, and correcting spelling.

20. Why is lowercasing during normalization sometimes undesirable?

    Lowercasing can erase meaningful distinctions — e.g. 'US' (country) vs 'us' (pronoun), or 'Apple' (company) vs 'apple' (fruit) — which hurts tasks like named entity recognition.

21. What is Part-of-Speech (POS) tagging?

    POS tagging assigns each token a grammatical category (such as noun, verb, adjective, adverb, pronoun, preposition) based on its definition and context within the sentence.

See more Natural Language Processing flashcards →

Planning Natural Language Processing for Machine Learning

Natural Language Processing (Machine Learning) FAQ