Before it
word2vec learned from local context windows, one sample at a time.
The problem
Local windows ignore the global co-occurrence statistics of the whole corpus.
What it introduced
Factorise the global word co-occurrence matrix directly, giving vectors with a cleaner interpretation.
Why it matters
It gave the embedding idea a second, complementary derivation and became the default pretrained vectors for years.
What modern models inherit
Count-based and predictive embeddings as two views of the same geometry.
What came next
Absorbed into ELMo and other models as an input layer.