Сравнение методов

	Векторные представления BERT	Встраивания GloVe	Word2Vec
Область	Интеллектуальный анализ текста	Интеллектуальный анализ текста	Интеллектуальный анализ текста
Семейство	Process / pipeline	Process / pipeline	Process / pipeline
Год появления	2019	2014	2013
Автор метода	Devlin, Chang, Lee & Toutanova (Google AI)	Pennington, Socher & Manning	Tomas Mikolov et al.
Тип	Contextual transformer text-representation method	Static word-embedding model	Neural word-embedding model
Основополагающий источник	Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. (2019). BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. NAACL-HLT, 4171-4186.	Pennington, J., Socher, R. & Manning, C. D. (2014). GloVe: Global Vectors for Word Representation. EMNLP.	Mikolov, T., Chen, K., Corrado, G. & Dean, J. (2013). Efficient Estimation of Word Representations in Vector Space.
Другие названия	contextual embeddings, transformer embeddings, BERT Tabanlı Metin Gömülmeleri	GloVe, global vectors, GloVe Kelime Gömülmeleri	word embeddings, skip-gram, continuous bag-of-words, Word2Vec Kelime Gömülmeleri
Связанные	4	3	4
Сводка	BERT-based text embeddings, introduced by Devlin and colleagues at Google AI in 2019, turn text into context-sensitive dense vectors using a bidirectional Transformer encoder. Because the meaning of a word shifts with its context, BERT produces richer representations than static methods such as Word2Vec or topic models like LDA.	GloVe (Global Vectors for Word Representation) is a static word-embedding model introduced by Pennington, Socher and Manning (2014) that learns word vectors directly from global word-word co-occurrence statistics gathered across an entire corpus. The resulting vectors place semantically related words close together and perform strongly on semantic analogy tasks.	Word2Vec is a neural word-embedding technique introduced by Mikolov and colleagues in 2013 that maps each word in a text corpus to a dense numeric vector. Words that appear in similar contexts end up close together in the vector space, so the embeddings capture semantic similarity that can be measured arithmetically.