ELMo (Emergent Language Model) is a deep learning-based Natural Language Processing technique developed by researchers at Google, NY Times, Stanford University and the Allen Institute for Artificial Intelligence. ELMo uses a deep learning technique called bidirectional language modeling to give a state-of-the-art accuracy in recognizing natural language.
ELMo is a recurrent neural network (RNN) based language model that gives word embeddings. The word embeddings are trained on a corpus of text in order to understand and capture syntactic and semantic relations between words. When using ELMo for natural language processing (NLP) tasks such as sentiment analysis and machine translation, the resulting model has state-of-the-art accuracy.
Whereas traditional word embedding techniques such as Word2Vec and GloVe can represent words of a language with finite-dimensional vector vectors, ELMo adds a context-dependent meaning to the vectors. This means that words used in different contexts get different vectors. This contextualization of word vectors was found to improve the performance of many NLP tasks.
ELMo is widely used in many systems including Google Translate, Google Assistant, Amazon Alexa and Apple’s Siri. In addition, ELMo has been used for speech recognition, text classification, machine translation, and question answering.
ELMo is an important step forward for natural language processing, as it enables new systems to recognize and interpret natural language more accurately. The use of deep learning-based language models is expected to further increase in the future.
