Methods for word encoding: A survey
Gilles Bernard, Georges Lebboss · 2017
Internal representation of words has gained a renewed interest with the apparition and extent of word embedders. It is crucial in many fields of research, information retrieval, natural language processing, document indexing, knowledge extraction. The goal of this paper is to revise most word encoding methods and models, from ngram sets to vector space models, through feature models, including word embedding. We propose some typological keys on the models and characterize the emergence of further needs and models in the domain.