AI Glossary Term
What does “Word Embedding” mean?
Definition
A dense vector representation of words capturing semantic relationships.
How Word Embedding works
Example
In a word-embedding model, words used in similar contexts get similar vectors. 'Doctor' ends up near 'nurse' and far from 'banana'. Classic examples such as word2vec showed that directions in the space can capture relationships.
Why it matters
Word embeddings were an early step toward machines handling meaning numerically. They also pick up biases that appear in text.
Common questions
Which methods made them popular?
Word2vec and GloVe, around 2013 and 2014.
Word vs contextual embeddings?
A word embedding gives each word one vector. Modern models produce vectors that change with the surrounding context.
Do they carry bias?
Yes. They can reflect stereotypes present in the training text.