In natural language processing (NLP), doc2vec and word2vec are two important techniques used for converting text into numerical representations that machine learning models can understand.

word2vec is a method introduced by Google in 2013 to create vector representations of words. It uses a neural network model to learn word associations from a large corpus of text. There are two main architectures within word2vec: Continuous Bag of Words (CBOW) and Skip-Gram. CBOW predicts a target word based on its context, while Skip-Gram does the opposite, predicting the context from a target word. The resulting word vectors capture semantic relationships, so words with similar meanings have similar vector representations. This technique has revolutionized many NLP tasks by providing a way to use semantic information in models.

doc2vec, on the other hand, extends word2vec to larger pieces of text like sentences, paragraphs, or entire documents. Introduced by the same team that developed word2vec, doc2vec aims to create a fixed-length vector representation for any length of text. There are two primary models in doc2vec: Distributed Memory (DM) and Distributed Bag of Words (DBOW). DM extends the CBOW model by adding a document vector, which influences the prediction of a target word. DBOW is similar to the Skip-Gram model but focuses on predicting words from a document's context without considering the order of words. These document vectors help in tasks such as document classification, sentiment analysis, and information retrieval.