Word embeddings concept (Word2Vec) in ML Python - Model Metrics & Evaluation

Metrics & Evaluation - Word embeddings concept (Word2Vec)

Which metric matters for Word2Vec and WHY

Word2Vec creates word vectors that capture meaning. To check if it works well, we use cosine similarity. This measures how close two word vectors are in meaning. A higher cosine similarity means words are used in similar ways.

We also look at analogy accuracy. For example, if the model can solve "king - man + woman = ?" and find "queen", it shows the embeddings learned meaningful relationships.

Confusion matrix or equivalent visualization

Word2Vec does not use a confusion matrix like classification. Instead, we visualize word vectors in 2D using t-SNE or PCA. Words with similar meanings cluster together.

    Example 2D plot:
    
    [king]      [queen]
       \        /
        \      /
         [royal]
          |
    [man]  [woman]
    
    Words close together show the model learned meaning well.
    

Precision vs Recall tradeoff with examples

In Word2Vec, precision and recall are less direct but relate to how well the model finds similar words.

• High precision: When the model returns similar words, they are truly related (few wrong words). Good for search engines to show relevant results.
• High recall: The model finds most related words, even if some are less relevant. Useful for creative writing tools that want many options.

Balancing these helps depending on the task: strict similarity or broad relatedness.

What "good" vs "bad" metric values look like

Good Word2Vec model:

• Cosine similarity between synonyms > 0.7
• Analogy accuracy > 70% on test questions
• Clear clusters of related words in visualization

Bad Word2Vec model:

• Cosine similarity near 0 for related words
• Analogy accuracy close to random guess (~25%)
• Random scatter in visualization with no clusters

Common pitfalls in Word2Vec metrics

• Overfitting: Model memorizes frequent word pairs but fails on new words.
• Data bias: If training data is small or biased, embeddings reflect that and give poor results.
• Ignoring context: Word2Vec gives one vector per word, so it can confuse words with multiple meanings.
• Misinterpreting cosine similarity: High similarity does not always mean synonyms; it can reflect related but different meanings.

Self-check question

Your Word2Vec model shows cosine similarity 0.95 between "cat" and "dog", but only 0.2 between "cat" and "kitten". Is this good? Why or why not?

Answer: This is not good. "Cat" and "kitten" are closely related (adult and baby), so similarity should be high. A low value means the model did not learn this relationship well. The high similarity between "cat" and "dog" might reflect they are both animals but is less precise than expected.

Key Result

Cosine similarity and analogy accuracy are key to check if Word2Vec captures meaningful word relationships.

Practice

1. What is the main purpose of Word2Vec in machine learning?

easy

A. To translate text from one language to another

B. To count the number of words in a sentence

C. To convert words into numerical vectors that capture their meanings

D. To generate random words for testing

Solution

1. Step 1: Understand Word2Vec's role

   Word2Vec creates vectors representing words based on their context, capturing meaning.

2. Step 2: Compare options

   Only To convert words into numerical vectors that capture their meanings describes converting words into meaningful numerical vectors.

3. Final Answer:

   To convert words into numerical vectors that capture their meanings -> Option C

4. Quick Check:

   Word2Vec = word vectors [OK]

Hint: Word2Vec turns words into numbers showing meaning [OK]

Common Mistakes:

• Thinking Word2Vec translates languages
• Confusing Word2Vec with word counting
• Assuming Word2Vec generates random words

2. Which of the following is the correct way to import Word2Vec from the gensim library in Python?

easy

A. import gensim.Word2Vec

B. import Word2Vec from gensim.models

C. from gensim import Word2Vec

D. from gensim.models import Word2Vec

Solution

1. Step 1: Recall correct Python import syntax

   Python uses 'from module import class' to import specific classes.

2. Step 2: Match with gensim syntax

   The correct import is 'from gensim.models import Word2Vec'. Other options have wrong syntax.

3. Final Answer:

   from gensim.models import Word2Vec -> Option D

4. Quick Check:

   Correct import syntax = from gensim.models import Word2Vec [OK]

Hint: Use 'from module import class' for specific imports [OK]

Common Mistakes:

• Using 'import Word2Vec from ...' which is invalid syntax
• Trying to import directly from 'gensim' without 'models'
• Using dot notation in import statements incorrectly

3. Given the following code snippet, what will be the output of print(model.wv.most_similar('king', topn=1))?

from gensim.models import Word2Vec
sentences = [['king', 'queen', 'man', 'woman'], ['prince', 'princess', 'boy', 'girl']]
model = Word2Vec(sentences, vector_size=10, window=2, min_count=1, epochs=10)

medium

A. [('man', similarity_score)] where similarity_score is close to 0

B. [('queen', similarity_score)] where similarity_score is close to 1

C. Error because 'king' is not in vocabulary

D. Empty list []

Solution

1. Step 1: Check if 'king' is in vocabulary

   'king' appears in the sentences, so it is in the model's vocabulary.

2. Step 2: Understand most_similar output

   Word2Vec finds words with similar context. 'queen' is closely related to 'king' in the data, so it will have highest similarity.

3. Final Answer:

   [('queen', similarity_score)] where similarity_score is close to 1 -> Option B

4. Quick Check:

   Most similar to 'king' = 'queen' [OK]

Hint: Most similar words share context, 'queen' pairs with 'king' [OK]

Common Mistakes:

• Assuming 'king' is missing from vocabulary
• Expecting unrelated words like 'man' as most similar
• Thinking output is an error or empty list

4. Identify the error in this Word2Vec training code and select the fix:

from gensim.models import Word2Vec
sentences = [['cat', 'dog', 'mouse'], ['dog', 'cat', 'elephant']]
model = Word2Vec(sentences, vector_size=5, window=2, min_count=2, epochs=5)
model.train(sentences, total_examples=1, epochs=5)

medium

A. Set min_count=1 to include all words in training

B. Change total_examples to len(sentences) in train()

C. Remove the train() call because Word2Vec trains on initialization

D. Increase vector_size to at least 10

Solution

1. Step 1: Analyze min_count effect

   min_count=2 excludes words appearing less than twice. 'mouse' and 'elephant' appear once, so they are ignored.

2. Step 2: Fix by lowering min_count

   Setting min_count=1 includes all words, fixing vocabulary size and training coverage.

3. Final Answer:

   Set min_count=1 to include all words in training -> Option A

4. Quick Check:

   min_count controls word inclusion [OK]

Hint: min_count filters words; lower it to include rare words [OK]

Common Mistakes:

• Ignoring min_count effect on vocabulary
• Thinking train() is unnecessary or wrongly used
• Changing vector_size without addressing vocabulary issue

5. You want to find the word that best fits the analogy: 'king' is to 'man' as 'queen' is to ____. Using Word2Vec, which operation would you perform to find this word?

hard

A. vector('king') - vector('man') + vector('queen')

B. vector('king') + vector('man') - vector('queen')

C. vector('queen') - vector('king') + vector('man')

D. vector('man') - vector('king') + vector('queen')

Solution

1. Step 1: Understand analogy vector math

   Word2Vec analogies use vector arithmetic: 'king' - 'man' + 'queen' approximates the vector for the missing word.

2. Step 2: Apply formula to find the answer

   Performing vector('king') - vector('man') + vector('queen') finds the word similar to 'queen' but related like 'king' is to 'man'.

3. Final Answer:

   vector('king') - vector('man') + vector('queen') -> Option A

4. Quick Check:

   Analogy = king - man + queen [OK]

Hint: Use king - man + queen to solve analogies [OK]

Common Mistakes:

• Mixing up vector addition and subtraction order
• Using unrelated vector combinations
• Ignoring analogy vector math concept