Revision history [back]

Yes, Gensim can be used for counted-based embeddings, such as using the word co-occurrence matrix. The gensim.models module provides the CoOccurrences class, which can be used to create a word co-occurrence matrix from a corpus of documents. This matrix can then be used to train word embeddings using the Word2Vec class or other methods provided by the gensim.models module. However, compared to neural network-based embeddings like Word2Vec, counting-based embeddings are less common as they may not capture more complex relationships between words.