论文部分内容阅读
Named Entity Recognition(NER)is a tough task in Chi-nese social media due to a large portion of informal writings.Existing research uses only limited in-domain annotated data and achieves low performance.In this paper,we utilize both limited in-domain data and enough out-of-domain data using a domain adaptation method.We pro-pose a multichannel LSTM-CRF model that employs different channels to capture general patterns,in-domain patterns and out-of-domain pat-terns in Chinese social media.The extensive experiments show that our model yields 9.8%improvement over previous state-of-the-art methods.We further ifnd that a shared embedding layer is important and ran-domly initialized embeddings are better than the pretrained ones.