{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"provenance":[],"machine_shape":"hm","gpuType":"A100","authorship_tag":"ABX9TyMX6G6zwnmndl1+m49u4/ca"},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","source":["\n","\n","# 📊 Comparison of Word Embedding Models\n","\n","| Feature            | CBOW                          | Skip-Gram                     | FastText                                    | GloVe                                      |\n","| ------------------ | ----------------------------- | ----------------------------- | ------------------------------------------- | ------------------------------------------ |\n","| Full Name          | Continuous Bag of Words       | Skip-Gram                     | FastText (Extension of Word2Vec)            | Global Vectors                             |\n","| Type               | Predictive model              | Predictive model              | Predictive + Subword                        | Count-based model                          |\n","| Training Idea      | Context ➜ predict center word | Center word ➜ predict context | Same as Word2Vec but uses character n-grams | Factorizes global co-occurrence matrix     |\n","| Learns From        | Local context window          | Local context window          | Local context + subword info                | Global word-word statistics                |\n","| Handles Rare Words | Medium                        | Good                          | Very Good                                   | Medium                                     |\n","| Handles OOV Words  | ❌ No                          | ❌ No                          | ✅ Yes (via subwords)                        | ❌ No                                       |\n","| Speed              | Faster                        | Slower                        | Slightly slower than Word2Vec               | Depends (often heavy matrix factorization) |\n","| Memory Usage       | Low                           | Low                           | Slightly higher                             | Higher                                     |\n","| Good For           | Frequent words                | Rare words                    | Morphologically rich languages              | Capturing global structure                 |\n","| Example Library    | Gensim (`sg=0`)               | Gensim (`sg=1`)               | Gensim `FastText`                           | Pretrained via Gensim / Stanford           |\n","\n","---\n","\n","# 🔍 Core Concept Difference\n","\n","| Model Type       | How It Learns                              |\n","| ---------------- | ------------------------------------------ |\n","| CBOW / Skip-Gram | Neural network predicts words              |\n","| FastText         | Neural network + character pieces          |\n","| GloVe            | Matrix factorization of word co-occurrence |\n","\n","---\n","\n","# 🧠 Intuitive Summary\n","\n","* **CBOW** → “Fill in the missing word.”\n","* **Skip-Gram** → “Given a word, predict neighbors.”\n","* **FastText** → “Words are made of smaller pieces.”\n","* **GloVe** → “Words that appear together globally are similar.”\n","\n","---\n","\n","# 🎯 When To Use What?\n","\n","| Scenario                                               | Best Choice |\n","| ------------------------------------------------------ | ----------- |\n","| Small dataset                                          | CBOW        |\n","| Rare words matter                                      | Skip-Gram   |\n","| Morphologically rich language (Turkish, Persian, etc.) | FastText    |\n","| Want strong pretrained vectors                         | GloVe       |\n","\n"],"metadata":{"id":"B0wOpYG7rE05"}},{"cell_type":"markdown","source":["## 0) Setup (install + imports)"],"metadata":{"id":"ARI4ZswxlqnF"}},{"cell_type":"code","source":["!pip -q install gensim\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"r6i6-MWmluPb","executionInfo":{"status":"ok","timestamp":1770912288055,"user_tz":360,"elapsed":4116,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"2a65abfe-3ef7-4099-efa8-166f40dc97e0"},"execution_count":1,"outputs":[{"output_type":"stream","name":"stdout","text":["\u001b[2K   \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m27.9/27.9 MB\u001b[0m \u001b[31m90.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n","\u001b[?25h"]}]},{"cell_type":"code","execution_count":2,"metadata":{"id":"q_ELF50nlX8u","executionInfo":{"status":"ok","timestamp":1770912291001,"user_tz":360,"elapsed":2941,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"outputs":[],"source":["import gensim\n","import gensim.downloader as api\n","from gensim.models import Word2Vec, FastText\n","from gensim.utils import simple_preprocess"]},{"cell_type":"markdown","source":["\n","Word2Vec → CBOW and Skip-gram (same class, different sg)\n","\n","FastText → subword embeddings (handles rare/OOV better)\n","\n","GloVe → we’ll load pre-trained GloVe (most reliable in class), and optionally show a simple “train-your-own” approach later"],"metadata":{"id":"H16m_u91l7b6"}},{"cell_type":"markdown","source":["### 1) Load a corpus (text8)"],"metadata":{"id":"y43EbxgXmZTj"}},{"cell_type":"code","source":["# text8 is a cleaned Wikipedia subset commonly used in word embedding demos\n","\n","#Embedding models need lots of word co-occurrence.\n","\n","#text8 is already tokenized into sentences (lists of tokens), ideal for Word2Vec/FastText.\n","\n","corpus = api.load(\"text8\")  # iterable of tokenized sentences (lists of tokens)\n","\n","# Peek\n","for i, sent in enumerate(corpus):\n","    print(sent[:20])\n","    if i == 2:\n","        break\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"KOALlLu2l06_","executionInfo":{"status":"ok","timestamp":1770912359394,"user_tz":360,"elapsed":5740,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"e8eceb13-6ee7-45d8-bda5-9ba9429873e0"},"execution_count":3,"outputs":[{"output_type":"stream","name":"stdout","text":["[==================================================] 100.0% 31.6/31.6MB downloaded\n","['anarchism', 'originated', 'as', 'a', 'term', 'of', 'abuse', 'first', 'used', 'against', 'early', 'working', 'class', 'radicals', 'including', 'the', 'diggers', 'of', 'the', 'english']\n","['reciprocity', 'qualitative', 'impairments', 'in', 'communication', 'as', 'manifested', 'by', 'at', 'least', 'one', 'of', 'the', 'following', 'delay', 'in', 'or', 'total', 'lack', 'of']\n","['with', 'the', 'aegis', 'of', 'zeus', 'when', 'he', 'goes', 'to', 'the', 'battlefield', 'the', 'entire', 'trojan', 'army', 'flees', 'behind', 'the', 'walls', 'of']\n"]}]},{"cell_type":"markdown","source":["### 2) Helper: quick evaluation (similarity + analogies)"],"metadata":{"id":"m96HYJm3mxOc"}},{"cell_type":"code","source":["def probe(model, word=\"king\"):\n","    print(\"Most similar to:\", word)\n","    for w, s in model.wv.most_similar(word, topn=10):\n","        print(f\"  {w:12s} {s:.3f}\")\n","\n","def analogy(model, a, b, c, topn=5):\n","    # a is to b as c is to ?\n","    print(f\"\\nAnalogy: {a} : {b} :: {c} : ?\")\n","    for w, s in model.wv.most_similar(positive=[b, c], negative=[a], topn=topn):\n","        print(f\"  {w:12s} {s:.3f}\")\n"],"metadata":{"id":"KFF-DqUCmgsE","executionInfo":{"status":"ok","timestamp":1770912401791,"user_tz":360,"elapsed":19,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"execution_count":4,"outputs":[]},{"cell_type":"markdown","source":["🔹 Conceptual Difference (for your students)\n","\n","**CBOW**\n","\n","Context → Predict center word\n","\n","Example:\n","[\"the\", \"cat\", __ , \"on\", \"mat\"]\n","Predict → \"sat\"\n","\n","✔ Faster\n","✔ Works well for frequent words"],"metadata":{"id":"pDj6FzAwnYHN"}},{"cell_type":"markdown","source":["Skip-gram\n","**bold text**\n","Center word → Predict context\n","\n","Example:\n","Input: \"sat\"\n","Predict: \"the\", \"cat\", \"on\", \"mat\"\n","\n","✔ Better for rare words\n","✔ Slightly slower"],"metadata":{"id":"m1FQA9_dnZ50"}},{"cell_type":"markdown","source":["## CBOW\n","### 3) Train CBOW (Word2Vec with sg=0)\n","\n","\n","Predicts the center word from context words.\n","\n","Usually faster and can work well for frequent words.\n","\n","In Gensim: sg=0 ⇒ CBOW."],"metadata":{"id":"9Ji723wlnDsv"}},{"cell_type":"code","source":["cbow = Word2Vec(\n","    sentences=corpus,     # The training data: list (or iterable) of tokenized sentences\n","    vector_size=100,      # Dimensionality of word vectors (embedding size)\n","    window=5,             # Maximum distance between current word and context words\n","    min_count=5,          # Ignore words that appear fewer than 5 times\n","    workers=2,            # Number of CPU cores used for training (parallelization)\n","    sg=0,                 # Training algorithm: 0 = CBOW, 1 = Skip-gram\n","    negative=10,          # Number of negative samples for negative sampling\n","    epochs=5              # Number of times the model sees the entire corpus\n",")\n","\n","\n","probe(cbow, \"king\")\n","analogy(cbow, \"man\", \"woman\", \"king\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"TM-fWKhzmsfx","executionInfo":{"status":"ok","timestamp":1770912585999,"user_tz":360,"elapsed":148301,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"f0c9e5ff-c6da-446b-a272-e07d572ff6a3"},"execution_count":5,"outputs":[{"output_type":"stream","name":"stdout","text":["Most similar to: king\n","  prince       0.712\n","  kings        0.675\n","  throne       0.671\n","  queen        0.662\n","  aragon       0.649\n","  constantine  0.646\n","  emperor      0.646\n","  vii          0.638\n","  charlemagne  0.633\n","  sigismund    0.633\n","\n","Analogy: man : woman :: king : ?\n","  queen        0.662\n","  prince       0.601\n","  isabella     0.592\n","  princess     0.587\n","  elizabeth    0.587\n"]}]},{"cell_type":"markdown","source":["##Skip-gram\n","###4) Train Skip-gram (Word2Vec with sg=1)\n","\n","Predicts context words from the center word.\n","\n","Often better for rare words, but can be slower.\n","\n","In Gensim: sg=1 ⇒ Skip-gram."],"metadata":{"id":"DsYGjoF7n6Om"}},{"cell_type":"code","source":["skipgram = Word2Vec(\n","    sentences=corpus,\n","    vector_size=100,\n","    window=5,\n","    min_count=5,\n","    workers=2,\n","    sg=1,          # <-- Skip-gram\n","    negative=10,\n","    epochs=5\n",")\n","\n","probe(skipgram, \"king\")\n","analogy(skipgram, \"man\", \"woman\", \"king\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"91T9lsUGm1Qi","executionInfo":{"status":"ok","timestamp":1770913355089,"user_tz":360,"elapsed":613025,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"818ff2d2-d268-4eac-9905-c8df1ade730e"},"execution_count":6,"outputs":[{"output_type":"stream","name":"stdout","text":["Most similar to: king\n","  prince       0.777\n","  haakon       0.750\n","  valdemar     0.740\n","  vii          0.739\n","  kings        0.734\n","  queen        0.733\n","  sweyn        0.731\n","  canute       0.730\n","  throne       0.722\n","  sobieski     0.712\n","\n","Analogy: man : woman :: king : ?\n","  queen        0.672\n","  consort      0.656\n","  daughter     0.645\n","  valois       0.635\n","  alexandra    0.633\n"]}]},{"cell_type":"markdown","source":["##FastText\n","###5) Train FastText (subword embeddings)\n","\n","Represents a word as a sum of character n-gram vectors.\n","\n","Advantage: can produce vectors for OOV words (words not seen in training) via subwords."],"metadata":{"id":"CPmarLIIocvm"}},{"cell_type":"code","source":["fasttext = FastText(\n","    sentences=corpus,\n","    vector_size=100,\n","    window=5,\n","    min_count=5,\n","    workers=2,\n","    sg=1,          # can do CBOW(0) or Skip-gram(1); Skip-gram is common\n","    negative=10,\n","    epochs=5\n",")\n","\n","probe(fasttext, \"king\")\n","analogy(fasttext, \"man\", \"woman\", \"king\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"sUwrvb6In_lu","executionInfo":{"status":"ok","timestamp":1770914244642,"user_tz":360,"elapsed":889547,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"e7a924fe-ee08-4c60-b1e3-35192258c261"},"execution_count":7,"outputs":[{"output_type":"stream","name":"stdout","text":["Most similar to: king\n","  prince       0.774\n","  son          0.735\n","  kingship     0.725\n","  kings        0.720\n","  haakon       0.710\n","  kingssonar   0.703\n","  kintyre      0.700\n","  reigning     0.700\n","  throne       0.690\n","  reigned      0.684\n","\n","Analogy: man : woman :: king : ?\n","  queen        0.730\n","  daughter     0.700\n","  matilda      0.688\n","  elisabeth    0.686\n","  stepdaughter 0.686\n"]}]},{"cell_type":"code","source":["# These may or may not exist in vocab; FastText can still build vectors via subwords\n","test_words = [\"computer\", \"computerr\", \"microsoft\", \"microsoft\"]\n","\n","for w in test_words:\n","    try:\n","        sims = fasttext.wv.most_similar(w, topn=5)\n","        print(w, \"=>\", sims[:5])\n","    except KeyError as e:\n","        print(\"KeyError for:\", w, e)\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"rTGovR40okC9","executionInfo":{"status":"ok","timestamp":1770914244645,"user_tz":360,"elapsed":65,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"ae28d72a-601c-4893-a1c0-5b51d328196b"},"execution_count":8,"outputs":[{"output_type":"stream","name":"stdout","text":["computer => [('omputer', 0.9314585328102112), ('multicomputer', 0.8884887099266052), ('computers', 0.8599362373352051), ('minicomputer', 0.8568235635757446), ('microcomputer', 0.8417887687683105)]\n","computerr => [('computer', 0.9398351311683655), ('computers', 0.9105141758918762), ('comput', 0.89348965883255), ('multicomputer', 0.8828125596046448), ('minicomputer', 0.8810953497886658)]\n","microsoft => [('windows', 0.8205729126930237), ('netbsd', 0.8185611367225647), ('netware', 0.8095643520355225), ('openbsd', 0.8092650771141052), ('vmware', 0.8006311655044556)]\n","microsoft => [('windows', 0.8205729126930237), ('netbsd', 0.8185611367225647), ('netware', 0.8095643520355225), ('openbsd', 0.8092650771141052), ('vmware', 0.8006311655044556)]\n"]}]},{"cell_type":"markdown","source":["##GloVe (pre-trained)\n","###6) Load pre-trained GloVe vectors (fast + reliable for class)\n","\n","Trains on a global co-occurrence matrix (word-word counts).\n","\n","Unlike Word2Vec’s local sliding-window prediction, GloVe factorizes co-occurrence statistics to learn vectors."],"metadata":{"id":"xSnuOyZaovsF"}},{"cell_type":"code","source":["glove = api.load(\"glove-wiki-gigaword-50\")  # 50-dim pretrained\n","print(glove.vector_size)\n","\n","print(\"Most similar to 'king':\")\n","for w, s in glove.most_similar(\"king\", topn=10):\n","    print(f\"  {w:12s} {s:.3f}\")\n","\n","print(\"\\nAnalogy: man:woman :: king:?\")\n","for w, s in glove.most_similar(positive=[\"woman\", \"king\"], negative=[\"man\"], topn=5):\n","    print(f\"  {w:12s} {s:.3f}\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"o_2BJ0cdoszk","executionInfo":{"status":"ok","timestamp":1770914275680,"user_tz":360,"elapsed":31046,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"a5941732-eea3-4a00-96a8-f3867f1fd962"},"execution_count":9,"outputs":[{"output_type":"stream","name":"stdout","text":["[==================================================] 100.0% 66.0/66.0MB downloaded\n","50\n","Most similar to 'king':\n","  prince       0.824\n","  queen        0.784\n","  ii           0.775\n","  emperor      0.774\n","  son          0.767\n","  uncle        0.763\n","  kingdom      0.754\n","  throne       0.754\n","  brother      0.749\n","  ruler        0.743\n","\n","Analogy: man:woman :: king:?\n","  queen        0.852\n","  throne       0.766\n","  prince       0.759\n","  daughter     0.747\n","  elizabeth    0.746\n"]}]},{"cell_type":"markdown","source":["### 7) Compare models quickly (same probe across CBOW / Skip-gram / FastText / GloVe)"],"metadata":{"id":"ai2zHOA9o7R8"}},{"cell_type":"code","source":["def compare(word=\"bank\"):\n","    print(\"=== CBOW ===\")\n","    probe(cbow, word)\n","    print(\"\\n=== Skip-gram ===\")\n","    probe(skipgram, word)\n","    print(\"\\n=== FastText ===\")\n","    probe(fasttext, word)\n","    print(\"\\n=== GloVe (pretrained) ===\")\n","    for w, s in glove.most_similar(word, topn=10):\n","        print(f\"  {w:12s} {s:.3f}\")\n","\n","compare(\"bank\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"tE3exY9Mo0yv","executionInfo":{"status":"ok","timestamp":1770914275733,"user_tz":360,"elapsed":40,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"4c064165-3bfe-4fc3-886b-69c40099767e"},"execution_count":10,"outputs":[{"output_type":"stream","name":"stdout","text":["=== CBOW ===\n","Most similar to: bank\n","  banks        0.625\n","  fund         0.587\n","  monetary     0.577\n","  commission   0.548\n","  imf          0.544\n","  loans        0.530\n","  bureau       0.528\n","  reserve      0.528\n","  trade        0.524\n","  hedge        0.508\n","\n","=== Skip-gram ===\n","Most similar to: bank\n","  monetary     0.724\n","  banks        0.723\n","  bundesbank   0.670\n","  fund         0.661\n","  suntrust     0.640\n","  brokers      0.636\n","  imf          0.634\n","  convertibility 0.633\n","  subsidy      0.632\n","  escb         0.627\n","\n","=== FastText ===\n","Most similar to: bank\n","  banks        0.798\n","  ewbank       0.760\n","  citibank     0.753\n","  medibank     0.723\n","  eubank       0.723\n","  monetary     0.703\n","  banknote     0.699\n","  fund         0.694\n","  commerzbank  0.692\n","  databank     0.690\n","\n","=== GloVe (pretrained) ===\n","  banks        0.870\n","  securities   0.800\n","  banking      0.797\n","  investment   0.785\n","  exchange     0.781\n","  financial    0.767\n","  credit       0.765\n","  lender       0.752\n","  capital      0.738\n","  brokerage    0.737\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"awRdRufCriSD"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"uHIzkhLYriPl"},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":["###Demo 0 — Setup + Train quickly (CBOW / Skip-gram / FastText) + Load GloVe"],"metadata":{"id":"g5RAMHlQwmcw"}},{"cell_type":"code","source":["# Use text8 so training is fast and results look decent\n","corpus = api.load(\"text8\")\n","\n","cbow = Word2Vec(corpus, vector_size=100, window=5, min_count=5, workers=2, sg=0, negative=10, epochs=5)\n","sg   = Word2Vec(corpus, vector_size=100, window=5, min_count=5, workers=2, sg=1, negative=10, epochs=5)\n","ft   = FastText(corpus, vector_size=100, window=5, min_count=5, workers=2, sg=1, negative=10, epochs=5)\n","\n","glove = api.load(\"glove-wiki-gigaword-50\")  # pretrained\n"],"metadata":{"id":"BZXSJnHyriMl","executionInfo":{"status":"ok","timestamp":1770915923880,"user_tz":360,"elapsed":1648130,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"execution_count":11,"outputs":[]},{"cell_type":"code","source":["def topn(model, word, n=8):\n","    if hasattr(model, \"wv\"):\n","        return model.wv.most_similar(word, topn=n)\n","    return model.most_similar(word, topn=n)\n","\n","def show_neighbors(word):\n","    print(f\"\\nWORD: {word}\")\n","    for name, model in [(\"CBOW\", cbow), (\"Skip-gram\", sg), (\"FastText\", ft), (\"GloVe(pre)\", glove)]:\n","        try:\n","            sims = topn(model, word, 8)\n","            print(f\"  {name:10s} -> \" + \", \".join([f\"{w}({s:.2f})\" for w,s in sims]))\n","        except Exception as e:\n","            print(f\"  {name:10s} -> [no result: {e}]\")\n"],"metadata":{"id":"CmegbnBfriJi","executionInfo":{"status":"ok","timestamp":1770915923916,"user_tz":360,"elapsed":10,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"execution_count":12,"outputs":[]},{"cell_type":"markdown","source":["###Demo 1 — OOV + Misspellings: FastText “magic”\n","\n","**Goal:** show that FastText can produce vectors for unseen words (subwords), while CBOW/SG/GloVe usually fail."],"metadata":{"id":"9AuxJaZYwxOa"}},{"cell_type":"markdown","source":["CBOW/Skip-gram/GloVe: often KeyError (unknown word)\n","\n","FastText: still gives reasonable neighbors for misspellings / variants"],"metadata":{"id":"G_DTFc2Xw8hT"}},{"cell_type":"code","source":["test_words = [\"internet\", \"internets\", \"interneet\", \"computerr\", \"microsof\", \"mexicoo\"]\n","\n","for w in test_words:\n","    print(\"\\n====================\")\n","    print(\"Word:\", w)\n","\n","    for name, model in [(\"CBOW\", cbow), (\"Skip-gram\", sg), (\"FastText\", ft), (\"GloVe(pre)\", glove)]:\n","        try:\n","            sims = topn(model, w, 5)\n","            print(f\"{name:10s}:\", [x[0] for x in sims])\n","        except Exception as e:\n","            print(f\"{name:10s}: FAIL ({type(e).__name__})\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"jYowTz-JriF9","executionInfo":{"status":"ok","timestamp":1770915923995,"user_tz":360,"elapsed":84,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"7850d9eb-8a41-4ac5-8d9a-68e1f7532a0f"},"execution_count":13,"outputs":[{"output_type":"stream","name":"stdout","text":["\n","====================\n","Word: internet\n","CBOW      : ['usenet', 'arpanet', 'aol', 'ip', 'fidonet']\n","Skip-gram : ['bitnet', 'dsl', 'isps', 'compuserve', 'providers']\n","FastText  : ['internetwork', 'internetworking', 'undernet', 'telnet', 'usenet']\n","GloVe(pre): ['web', 'online', 'users', 'networks', 'google']\n","\n","====================\n","Word: internets\n","CBOW      : FAIL (KeyError)\n","Skip-gram : FAIL (KeyError)\n","FastText  : ['internet', 'internetwork', 'internetworking', 'undernet', 'internecine']\n","GloVe(pre): ['esps', 'particuliers', 'googlers', 'shakhas', 'geiko']\n","\n","====================\n","Word: interneet\n","CBOW      : FAIL (KeyError)\n","Skip-gram : FAIL (KeyError)\n","FastText  : ['internecine', 'intern', 'internetwork', 'internetworking', 'interned']\n","GloVe(pre): FAIL (KeyError)\n","\n","====================\n","Word: computerr\n","CBOW      : FAIL (KeyError)\n","Skip-gram : FAIL (KeyError)\n","FastText  : ['computer', 'computers', 'comput', 'multicomputer', 'omputer']\n","GloVe(pre): FAIL (KeyError)\n","\n","====================\n","Word: microsof\n","CBOW      : FAIL (KeyError)\n","Skip-gram : FAIL (KeyError)\n","FastText  : ['microsoft', 'micros', 'netbsd', 'microsystems', 'microserfs']\n","GloVe(pre): FAIL (KeyError)\n","\n","====================\n","Word: mexicoo\n","CBOW      : FAIL (KeyError)\n","Skip-gram : FAIL (KeyError)\n","FastText  : ['mexico', 'mexica', 'mexicali', 'mexicana', 'cubango']\n","GloVe(pre): FAIL (KeyError)\n"]}]},{"cell_type":"markdown","source":["###Demo 2 — “Rare word survival”: Skip-gram usually wins\n","\n","**Goal:** create an artificial “rare word” situation and show Skip-gram tends to keep rare words better.\n","\n","Skip-gram usually gives more meaningful neighbors for the rare term."],"metadata":{"id":"8MBpzpGJxBAJ"}},{"cell_type":"markdown","source":["####2.1 Create a tiny corpus with a rare word"],"metadata":{"id":"SypI8ngoxFa9"}},{"cell_type":"code","source":["toy = [\n","    \"the doctor treated the patient in the hospital\".split(),\n","    \"the nurse helped the patient in the clinic\".split(),\n","    \"the surgeon operated in the hospital\".split(),\n","] * 200\n","\n","# Rare word appears only a few times\n","toy += [\n","    \"the hematologist studied blood disorders\".split(),\n","    \"hematologist helps diagnose blood diseases\".split(),\n","] * 3\n"],"metadata":{"id":"OTNoyXZsrh8o","executionInfo":{"status":"ok","timestamp":1770915924036,"user_tz":360,"elapsed":37,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"execution_count":14,"outputs":[]},{"cell_type":"markdown","source":["####2.2 Train CBOW vs Skip-gram on the toy corpus"],"metadata":{"id":"pH9kc8HhxI3A"}},{"cell_type":"code","source":["toy_cbow = Word2Vec(toy, vector_size=50, window=3, min_count=1, workers=2, sg=0, epochs=50)\n","toy_sg   = Word2Vec(toy, vector_size=50, window=3, min_count=1, workers=2, sg=1, epochs=50)\n","\n","def toy_neighbors(word=\"hematologist\"):\n","    for name, m in [(\"CBOW\", toy_cbow), (\"Skip-gram\", toy_sg)]:\n","        sims = m.wv.most_similar(word, topn=5)\n","        print(f\"{name:8s} -> \" + \", \".join([w for w,_ in sims]))\n","\n","toy_neighbors(\"hematologist\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"y_acnhtmryGN","executionInfo":{"status":"ok","timestamp":1770915924273,"user_tz":360,"elapsed":270,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"7fdab2da-2fc5-4051-96ce-ad053a1969db"},"execution_count":15,"outputs":[{"output_type":"stream","name":"stdout","text":["CBOW     -> blood, diagnose, studied, helps, diseases\n","Skip-gram -> diagnose, blood, diseases, studied, disorders\n"]}]},{"cell_type":"markdown","source":["###Demo 3 — “Polysemy pain”: One word, one vector (bank = river + money)\n","\n","**Goal:** show the limitation of static embeddings (all these are static).\n","\n","See mixed neighborhoods (finance + river etc.).\n","\n","“This is why contextual embeddings (BERT) exist.”"],"metadata":{"id":"XMwEMswsxXxC"}},{"cell_type":"code","source":["for w in [\"bank\", \"apple\", \"python\", \"jaguar\"]:\n","    show_neighbors(w)\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"IRJFNtSDryDW","executionInfo":{"status":"ok","timestamp":1770915924448,"user_tz":360,"elapsed":157,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"3de0b3a6-ecb5-4c7f-dcb9-81ba58132b7a"},"execution_count":16,"outputs":[{"output_type":"stream","name":"stdout","text":["\n","WORD: bank\n","  CBOW       -> fund(0.60), banks(0.60), imf(0.56), monetary(0.56), commission(0.54), trade(0.53), loans(0.52), bureau(0.52)\n","  Skip-gram  -> monetary(0.74), banks(0.71), bundesbank(0.67), fund(0.66), ecb(0.65), suntrust(0.65), imf(0.63), banking(0.63)\n","  FastText   -> banks(0.79), ewbank(0.73), citibank(0.73), eubank(0.72), medibank(0.72), monetary(0.70), fund(0.70), databank(0.70)\n","  GloVe(pre) -> banks(0.87), securities(0.80), banking(0.80), investment(0.78), exchange(0.78), financial(0.77), credit(0.76), lender(0.75)\n","\n","WORD: apple\n","  CBOW       -> macintosh(0.76), amiga(0.72), intel(0.72), ibm(0.69), atari(0.68), amd(0.67), hypercard(0.65), intellivision(0.65)\n","  Skip-gram  -> macintosh(0.81), iic(0.78), amiga(0.74), iigs(0.73), iie(0.73), ibm(0.73), iix(0.73), iicx(0.72)\n","  FastText   -> macintosh(0.84), applesoft(0.83), macintoshes(0.82), mcintosh(0.77), microsoft(0.77), iigs(0.75), amiga(0.75), hypercard(0.75)\n","  GloVe(pre) -> blackberry(0.75), chips(0.74), iphone(0.74), microsoft(0.73), ipad(0.73), pc(0.72), ipod(0.72), intel(0.72)\n","\n","WORD: python\n","  CBOW       -> monty(0.89), animaniacs(0.68), doraemon(0.66), grail(0.65), gilliam(0.63), capcom(0.62), sketch(0.61), blaxploitation(0.61)\n","  Skip-gram  -> monty(0.86), pythons(0.68), gilliam(0.68), perl(0.67), cleese(0.66), grail(0.64), sketch(0.63), spamalot(0.62)\n","  FastText   -> pythons(0.91), monty(0.86), pytheas(0.70), sketchpad(0.67), spamalot(0.66), sketch(0.66), kylix(0.63), cleese(0.63)\n","  GloVe(pre) -> reticulated(0.69), spamalot(0.66), php(0.64), owl(0.63), mouse(0.63), reticulatus(0.63), perl(0.63), monkey(0.62)\n","\n","WORD: jaguar\n","  CBOW       -> toy(0.72), nes(0.72), ds(0.71), chevrolet(0.71), snes(0.71), famicom(0.71), mazda(0.71), falcon(0.70)\n","  Skip-gram  -> coleco(0.75), lynx(0.74), mpv(0.74), intv(0.74), isuzu(0.73), wonderswan(0.73), acura(0.73), snes(0.73)\n","  FastText   -> jaguars(0.88), saguaro(0.77), lynx(0.76), mazdas(0.74), miata(0.72), supercar(0.72), mazda(0.72), supercars(0.71)\n","  GloVe(pre) -> xk(0.79), rover(0.78), falcon(0.77), xjs(0.77), xkr(0.75), xj6(0.74), puma(0.73), xk8(0.71)\n"]}]},{"cell_type":"markdown","source":["####Demo 4 — Word arithmetic (king - man + woman)\n","\n","**Goal:** show classic “vector geometry”. Usually best with Skip-gram / GloVe.\n","\n","GloVe/Skip-gram often look “smarter” here."],"metadata":{"id":"H77FnEKhxqig"}},{"cell_type":"code","source":["def analogy(model, a, b, c, topn=5):\n","    if hasattr(model, \"wv\"):\n","        res = model.wv.most_similar(positive=[b, c], negative=[a], topn=topn)\n","    else:\n","        res = model.most_similar(positive=[b, c], negative=[a], topn=topn)\n","    return res\n","\n","pairs = [\n","    (\"man\",\"woman\",\"king\"),\n","    (\"paris\",\"france\",\"rome\"),\n","    (\"good\",\"better\",\"bad\"),\n","]\n","\n","for a,b,c in pairs:\n","    print(f\"\\nAnalogy: {a}:{b} :: {c}:?\")\n","    for name, model in [(\"CBOW\", cbow), (\"Skip-gram\", sg), (\"FastText\", ft), (\"GloVe(pre)\", glove)]:\n","        try:\n","            ans = analogy(model, a,b,c, topn=3)\n","            print(f\"  {name:10s} -> {ans}\")\n","        except Exception as e:\n","            print(f\"  {name:10s} -> FAIL ({e})\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"X7Ly4X4dryAP","executionInfo":{"status":"ok","timestamp":1770915924452,"user_tz":360,"elapsed":20,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"0ff66f27-1f31-424a-cdef-3394eb55e8af"},"execution_count":17,"outputs":[{"output_type":"stream","name":"stdout","text":["\n","Analogy: man:woman :: king:?\n","  CBOW       -> [('queen', 0.6870896816253662), ('prince', 0.6478095650672913), ('princess', 0.644791305065155)]\n","  Skip-gram  -> [('queen', 0.6678494811058044), ('consort', 0.6352888941764832), ('daughter', 0.6315451860427856)]\n","  FastText   -> [('queen', 0.7003143429756165), ('dethrone', 0.6939550638198853), ('throne', 0.6888295412063599)]\n","  GloVe(pre) -> [('queen', 0.8523604273796082), ('throne', 0.7664334177970886), ('prince', 0.7592144012451172)]\n","\n","Analogy: paris:france :: rome:?\n","  CBOW       -> [('greece', 0.6862745881080627), ('spain', 0.6832908987998962), ('gaul', 0.6781083345413208)]\n","  Skip-gram  -> [('italy', 0.7439185976982117), ('papacy', 0.7385503649711609), ('greece', 0.7334614992141724)]\n","  FastText   -> [('italy', 0.7533741593360901), ('visigoth', 0.7380224466323853), ('greece', 0.7365857362747192)]\n","  GloVe(pre) -> [('italy', 0.8614554405212402), ('spain', 0.7927343249320984), ('portugal', 0.7596487998962402)]\n","\n","Analogy: good:better :: bad:?\n","  CBOW       -> [('worse', 0.7095265984535217), ('quicker', 0.668522834777832), ('slower', 0.6636506915092468)]\n","  Skip-gram  -> [('worse', 0.6934384703636169), ('talkative', 0.6342303156852722), ('gimmick', 0.6320604681968689)]\n","  FastText   -> [('betters', 0.6575916409492493), ('getter', 0.6323222517967224), ('betterment', 0.6268559098243713)]\n","  GloVe(pre) -> [('worse', 0.901099443435669), ('too', 0.8323724269866943), ('unfortunately', 0.8223986625671387)]\n"]}]},{"cell_type":"markdown","source":["###Demo 5 — Visualize embeddings (2D map) and compare models\n","\n","**Goal**: show clusters like (king, queen, prince, princess) and differences per model.\n","\n","Clusters form, but shape differs across models.\n","\n","Pretrained GloVe often looks more semantically “organized”."],"metadata":{"id":"SCfV8HMjx04W"}},{"cell_type":"code","source":["import matplotlib.pyplot as plt\n","from sklearn.manifold import TSNE\n","import numpy as np\n","\n","\n","words = [\"king\",\"queen\",\"man\",\"woman\",\"prince\",\"princess\",\n","         \"paris\",\"france\",\"rome\",\"italy\",\"london\",\"england\",\n","         \"doctor\",\"nurse\",\"hospital\",\"clinic\"]\n","\n","def get_vec(model, w):\n","    if hasattr(model, \"wv\"):\n","        return model.wv[w]\n","    return model[w]\n","\n","def plot_model(model, title):\n","    valid = []\n","    vecs = []\n","    for w in words:\n","        try:\n","            vecs.append(get_vec(model, w))\n","            valid.append(w)\n","        except:\n","            pass\n","\n","    X = np.array(vecs)\n","    X2 = TSNE(n_components=2, perplexity=5, init=\"random\", random_state=0).fit_transform(X)\n","\n","    plt.figure(figsize=(8,6))\n","    plt.scatter(X2[:,0], X2[:,1])\n","    for i, w in enumerate(valid):\n","        plt.text(X2[i,0]+0.5, X2[i,1]+0.5, w, fontsize=10)\n","    plt.title(title)\n","    plt.show()\n","\n","plot_model(cbow, \"CBOW (trained on text8)\")\n","plot_model(sg, \"Skip-gram (trained on text8)\")\n","plot_model(ft, \"FastText (trained on text8)\")\n","plot_model(glove, \"GloVe (pretrained)\")\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000},"id":"Hnb0F1oPrx84","executionInfo":{"status":"ok","timestamp":1770916269657,"user_tz":360,"elapsed":1216,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"7f0a8437-9998-4302-fd88-bbd2d0a076d2"},"execution_count":19,"outputs":[{"output_type":"display_data","data":{"text/plain":["<Figure size 800x600 with 1 Axes>"],"image/png":"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\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":["<Figure size 800x600 with 1 Axes>"],"image/png":"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\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":["<Figure size 800x600 with 1 Axes>"],"image/png":"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\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":["<Figure size 800x600 with 1 Axes>"],"image/png":"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\n"},"metadata":{}}]},{"cell_type":"markdown","source":["###Demo 6 — Domain shift: Train small “class corpus” vs pretrained GloVe\n","**Goal**: show why pretrained vectors help when your data is tiny.\n","\n","Small corpus produces “class-specific” associations.\n","\n","Pretrained gives broader, more general semantics.\n","\n","Great discussion: “Do we need domain embeddings or general ones?”"],"metadata":{"id":"3pqSlkgWx-8X"}},{"cell_type":"markdown","source":["####**6.1** Make a small domain corpus (e.g., “NLP class”)"],"metadata":{"id":"r6qpfVGGyGv-"}},{"cell_type":"code","source":["domain = [\n"," \"tokenization removes punctuation and splits text into tokens\",\n"," \"word embeddings represent words as vectors in a space\",\n"," \"cbow predicts a target word using surrounding context words\",\n"," \"skip gram predicts surrounding words using a center word\",\n"," \"fasttext uses subword information based on character n grams\",\n"," \"glove uses global co occurrence statistics and matrix factorization\",\n"," \"transformers provide contextual embeddings unlike word2vec\"\n","]\n","\n","domain = [s.lower().split() for s in domain]\n","dom_sg = Word2Vec(domain, vector_size=50, window=3, min_count=1, sg=1, epochs=200)\n"],"metadata":{"id":"O0tfarVAr9Yn","executionInfo":{"status":"ok","timestamp":1770916274007,"user_tz":360,"elapsed":186,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}}},"execution_count":20,"outputs":[]},{"cell_type":"markdown","source":["####6.2 Compare neighbors: domain-trained vs pretrained"],"metadata":{"id":"JctLGXr5yN7R"}},{"cell_type":"code","source":["def compare_domain(word):\n","    print(\"\\nWORD:\", word)\n","    print(\"Domain Skip-gram:\", [w for w,_ in dom_sg.wv.most_similar(word, topn=6)])\n","    try:\n","        print(\"GloVe pretrained:\", [w for w,_ in glove.most_similar(word, topn=6)])\n","    except Exception as e:\n","        print(\"GloVe pretrained: FAIL\", e)\n","\n","for w in [\"embeddings\",\"context\",\"transformers\",\"tokenization\"]:\n","    compare_domain(w)\n"],"metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"0YXMVlqLr9V5","executionInfo":{"status":"ok","timestamp":1770916275077,"user_tz":360,"elapsed":46,"user":{"displayName":"fazl phd","userId":"11176626483823360507"}},"outputId":"5d2771e6-5bd0-4bc3-97e5-dac1da219f68"},"execution_count":21,"outputs":[{"output_type":"stream","name":"stdout","text":["\n","WORD: embeddings\n","Domain Skip-gram: ['vectors', 'and', 'a', 'n', 'based', 'occurrence']\n","GloVe pretrained: ['conjugates', 'mini-game', 'floaters', 'voltammetry', 'endomorphisms', 'generaciones']\n","\n","WORD: context\n","Domain Skip-gram: ['a', 'and', 'matrix', 'word', 'words', 'based']\n","GloVe pretrained: ['aspects', 'perspective', 'particular', 'aspect', 'defining', 'subject']\n","\n","WORD: transformers\n","Domain Skip-gram: ['using', 'co', 'a', 'into', 'splits', 'contextual']\n","GloVe pretrained: ['marvel', 'dc', 'multiverse', 'toyline', 'wildstorm', 'energon']\n","\n","WORD: tokenization\n","Domain Skip-gram: ['factorization', 'based', 'information', 'punctuation', 'word', 'splits']\n","GloVe pretrained: ['turtling', 'hydrops', 'fenestrata', 'sooraya', 'silvestrii', 'pyy']\n"]}]},{"cell_type":"code","source":[],"metadata":{"id":"ZAI4kiaRr9Sx"},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":[],"metadata":{"id":"0N7QLoAvqJNM"}},{"cell_type":"markdown","source":["\n","\n","# 📌 1️⃣ Text Classification (Sentiment, Spam, Topic)\n","\n","### Example Task\n","\n","* Movie review sentiment analysis\n","* Spam detection\n","* News topic classification\n","\n","### What to use?\n","\n","| Scenario                                                 | Best Choice      | Why                                 |\n","| -------------------------------------------------------- | ---------------- | ----------------------------------- |\n","| Small dataset                                            | CBOW             | Faster, stable for frequent words   |\n","| Many rare words                                          | Skip-Gram        | Better representation of rare words |\n","| Morphologically rich language (Persian, Turkish, German) | FastText         | Handles prefixes/suffixes           |\n","| Want strong baseline quickly                             | Pretrained GloVe | Already trained on large corpus     |\n","\n","\n","\n","---\n","\n","# 📌 2️⃣ Word Similarity Task\n","\n","### Example\n","\n","Find similar words:\n","\n","```python\n","model.wv.most_similar(\"doctor\")\n","```\n","\n","### What to use?\n","\n","| Goal                        | Best Choice |\n","| --------------------------- | ----------- |\n","| Semantic similarity         | GloVe       |\n","| Local contextual similarity | Skip-Gram   |\n","| Noisy text (social media)   | FastText    |\n","\n","Example:\n","\n","* “king” → “queen”\n","* “car” → “vehicle”\n","\n","---\n","\n","# 📌 3️⃣ Analogy Tasks\n","\n","### Example\n","\n","```\n","man : woman :: king : ?\n","```\n","\n","### Best Choice?\n","\n","| Model     | Performance |\n","| --------- | ----------- |\n","| Skip-Gram | Very good   |\n","| GloVe     | Very good   |\n","| CBOW      | Good        |\n","| FastText  | Good        |\n","\n","If teaching classic analogy → **Skip-Gram or GloVe**\n","\n","---\n","\n","# 📌 4️⃣ Handling Misspellings / OOV Words\n","\n","### Example\n","\n","```\n","computerr\n","walked, walking, walks\n","```\n","\n","### Best Choice?\n","\n","✅ **FastText**\n","\n","Why?\n","Because it uses character n-grams:\n","\n","```\n","\"walking\" → walk + ing\n","```\n","\n","Very useful for:\n","\n","* Social media NLP\n","* Student essays\n","* Morphologically rich languages\n","\n","---\n","\n","# 📌 5️⃣ Low-Resource Language\n","\n","Example:\n","\n","* Small Persian dataset\n","* Small Arabic dataset\n","\n","### Best Choice?\n","\n","| Situation                      | Pick      |\n","| ------------------------------ | --------- |\n","| Very small corpus              | CBOW      |\n","| Need better rare word learning | Skip-Gram |\n","| Rich morphology                | FastText  |\n","\n","---\n","\n","# 📌 6️⃣ Information Retrieval (Search)\n","\n","Example:\n","\n","* Similar document retrieval\n","* Query expansion\n","\n","### Best Choice?\n","\n","| Task                    | Pick      |\n","| ----------------------- | --------- |\n","| Simple search expansion | GloVe     |\n","| Query similarity        | Skip-Gram |\n","| Noisy queries           | FastText  |\n","\n","---\n","\n","# 📌 7️⃣ Named Entity Recognition (Traditional ML)\n","\n","If using embeddings as features:\n","\n","* Pretrained GloVe → Good baseline\n","* FastText → Better for unknown names\n","\n","---\n","\n","# 🎯 Quick Decision Chart (Give This to Students)\n","\n","```\n","Do you have many unknown words?\n","   → Yes → FastText\n","   → No → continue\n","\n","Do you want strong pretrained vectors?\n","   → Yes → GloVe\n","   → No → continue\n","\n","Is dataset small?\n","   → Yes → CBOW\n","   → No → Skip-Gram\n","```\n","\n","---\n","\n","# 🧠 Practical Summary\n","\n","| If you want…               | Use       |\n","| -------------------------- | --------- |\n","| Speed                      | CBOW      |\n","| Rare word quality          | Skip-Gram |\n","| Morphology / OOV           | FastText  |\n","| Strong pretrained baseline | GloVe     |\n","\n","---\n","\n","# ⚠️ Important  Note\n","\n","\n","\n","> These are **static embeddings**.\n","> They do NOT understand context like BERT.\n","\n","Example:\n","\n","```\n","bank (river)\n","bank (money)\n","```\n","\n","All these models give **one vector per word**, regardless of context.\n","\n"],"metadata":{"id":"HfYiJKZDqo3i"}},{"cell_type":"code","source":[],"metadata":{"id":"wTp-7G56pAJk"},"execution_count":null,"outputs":[]}]}