An Overview of Named Entity Recognition

摘要

Named Entity Recognition (NER) is essential for some Natural Language Processing (NLP) tasks. Previous researchers gave a survey of NER in statistical machine learning era, however, research on NER has already changed a lot in recent decade. On the one hand, more and more NER systems adopt deep learning, transfer learning, knowledge base and other methods. On the other hand, multilingual and low resource languages NER researches increase rapidly. To reflect these changes, we here give an overview of NER based on 162 papers of NLP related conferences from 1996 to 2017. In this survey, we discuss two main aspects of NER research - target languages and technical approaches with statistical analysis. Finally, we summarize some conclusions and explore potential future issues in NER research.