Tech World Software SchoolTWSS

from first concept to confident practice

NLP

Teach computers to work with words, one useful idea at a time.

Start learning 18 lessons · first lesson about 15 min · free · no account needed
About this track

Explore how computers process and learn from human language - the ideas behind search engines, spam filters, autocorrect and chat assistants.

You’ll start with text and tokens, count and compare words, label and split real text data, build and evaluate classifiers, extract facts from documents, and work up to word vectors, attention and Transformers. You’ll also inspect generated summaries for missing or unsupported facts. Interactive games make the ideas visible: repair a leaky dataset, tag spans, tune a spam filter and solve word analogies. Every lesson ends with hands-on Python in your browser, and the capstone builds a working search engine.

Before you start
  • No prior NLP experience needed.
  • Basic Python helps; examples introduce the code as you go.
  • A modern browser with JavaScript enabled.
Loading progress…
  1. Unit 1 · 0/4 lessons

    Text foundations

    Normalize text, split it into tokens, clean it, and learn subword tokens. Badge: Token tamer

    1. 0x001Turn text into tokensNormalize text and split it into pieces a language system can process. 15 min 0/2 exercises solved 15 min 0/2 exercises solved
    2. 0x102Choose a tokenization strategySplit sentences consistently and make punctuation handling explicit. 18 min 0/2 exercises solved 18 min 0/2 exercises solved
    3. 0x203Clean text: stop words, stems and lemmasDecide which words to drop and how to reduce words to a common form. 18 min 0/2 exercises solved 18 min 0/2 exercises solved
    4. 0x304Learn subword tokens with byte-pair encodingSee how modern tokenizers learn reusable word pieces from data. 20 min 0/1 exercises solved 20 min 0/1 exercises solved
  2. Unit 2 · 0/3 lessons

    Count and compare words

    Weigh words with TF-IDF, measure spelling distance, and predict words from counts. Badge: Pattern finder

    1. 0x405Represent documents with word countsBuild a bag-of-words view and weigh words by how much they tell documents apart. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
    2. 0x506Measure word similarity with edit distanceCount the edits between two words and use it to correct typos. 22 min 0/2 exercises solved 22 min 0/2 exercises solved
    3. 0x607Model short sequences with n-gramsCount nearby word sequences, generate text from them, and meet their limits. 20 min 0/2 exercises solved 20 min 0/2 exercises solved
  3. Unit 3 · 0/4 lessons

    Classify text

    Label and split real text data, classify it with rules and Naive Bayes, then evaluate it honestly. Badge: Spam slayer

    1. 0x708Classify text with simple featuresBuild a tiny sentiment baseline, try to fool it, and learn to evaluate it fairly. 18 min 0/2 exercises solved 18 min 0/2 exercises solved
    2. 0x809Build a trustworthy text datasetWrite a labeling guide, handle ambiguous messages, and split by conversation to prevent leakage. 25 min 0/1 exercises solved 25 min 0/1 exercises solved
    3. 0x9010Learn a classifier with Naive BayesLet labeled examples, not hand-written word lists, decide which words matter. 25 min 0/1 exercises solved 25 min 0/1 exercises solved
    4. 0xA011Evaluate a classifier honestlyGo beyond accuracy with precision, recall, F1 and the decision threshold. 20 min 0/1 exercises solved 20 min 0/1 exercises solved
  4. Unit 4 · 0/2 lessons

    Language structure

    Tag parts of speech, find entities and extract useful fields from messy messages. Badge: Tagger

    1. 0xB012Tag words: parts of speech and named entitiesLabel every word with its grammatical role, and find the people, places and organizations in text. 25 min 0/1 exercises solved 25 min 0/1 exercises solved
    2. 0xC013Extract facts from messy documentsTurn support messages into structured fields with spans, rules, validation and human review. 25 min 0/1 exercises solved 25 min 0/1 exercises solved
  5. Unit 5 · 0/4 lessons

    Embeddings and Transformers

    Represent meaning as vectors, compute attention and verify generated text against source evidence. Badge: Context builder

    1. 0xD014Represent meaning with word vectorsPlace words as points in space so that similar words sit close together. 22 min 0/1 exercises solved 22 min 0/1 exercises solved
    2. 0xE015Add context: from word vectors to TransformersSee why one vector per word isn’t enough, and how Transformers build context-aware representations. 20 min 0/1 exercises solved 20 min 0/1 exercises solved
    3. 0xF016Compute attention step by stepTurn word-to-word scores into attention weights with softmax, and mix information by those weights. 22 min 0/2 exercises solved 22 min 0/2 exercises solved
    4. 0x10017Generate text, then verify itCompare summarization and translation, inspect source evidence, and catch unsupported numbers. 25 min 0/1 exercises solved 25 min 0/1 exercises solved
  6. Unit 6 · 0/1 lessons

    Capstone

    Build a working search engine from the pieces you learned. Badge: Search engineer

    1. 0x11018Capstone: build a mini search engineCombine tokenizing, stop words, an inverted index and TF-IDF into a working search engine. 35 min 0/2 exercises solved 35 min 0/2 exercises solved