Pronunciation-Enhanced Chinese Word Embedding

Abstract Chinese word embeddings have recently garnered considerable attention. characters and their sub-character components, which contain rich semantic information, are incorporated to learn embeddings. can represent a combination of meaning, structure, pronunciation. However, existing embedding learning methods focus on the structure meaning characters. In this study, we aim develop an method that make complete use information represented by characters, including phonology, morphology, semantics. Specifically, propose pronunciation-enhanced method, where pronunciations context target simultaneously encoded into Evaluation similarity, analogy reasoning, text classification, sentiment analysis validate effectiveness our proposed method.