Abstract: Conditional Random Field (CRF) Model is one of the important methods in natural language processing (NLP). To understand the research progress in this field, the paper comprehensively expounds and analyzes the research results of this model. The paper introduces the CRF model and summarizes the extended model research in six aspects: multi-label, hidden state, semantics hierarchy, spatial information, semi-supervised learning and model fusion. The applications of this model in word segmentation, labeling and marking, identification and detection, extraction and classification, filling and matching are summarized. Finally, the future research in this field is forecasted, including designing feature generation methods, optimizing training and inference algorithms, and extending the graph structure of the model.

Key words: Conditional Random Field (CRF), Natural language processing (NLP), Sequence labeling, Probability graph model, Machine learning model