Long named entities are often abbreviated in oral Chinese lan-guage, and this usually leads to out-of-vocabulary(OOV) prob-lems in speech recognition applications. The generation of Chi-nese abbreviations is much more complex than English abbrevi-ations, most of which are acronyms and truncations. In this pa-per, we propose a new method for automatically generating ab-breviations for Chinese named entities and we perform vocabu-lary expansion using output of the abbreviation model for voice search. In our abbreviation modeling, we convert the abbrevi-ation generation problem into a tagging problem and use the conditional random field (CRF) as the tagging tool. In the vo-cabulary expansion, considering the multiple abbreviation prob-lem and limited coverage of top-1 abbreviation candidate, we add top-10 candidates into the vocabulary. In our experiments, for the abbreviation modeling, we achieved the top-10 cover-age of 88.3% by the proposed method; for the voice search, we improved the voice search accuracy from 16.9% to 79.2% by incorporating the top-I0 abbreviation candidates to vocabulary.
展开▼
