Research on Information Extraction of Municipal Solid Waste Crisis using BERT-LSTM-CRF

There is much research on the phenomenon of municipal solid waste (MSW) and its improvement measures, and the method of information extraction be adopted to obtain the potential knowledge of MSW from the existing relevant research literature. Due to the complexity and diversity of the MSW, unsupervised training of target texts can be achieved through information data based on manual annotation. According to the characteristics of the BERT language model, a common method in natural language processing(NLP), the pre-trained BERT(Bidirectional Encoder Representation from Transformers) model with LSTM-CRF(Long Short Term Memory-Conditional Random Field) architecture is used in the information extraction of MSW crisis to extract entities and relationships between entities from natural language texts. By the method of calculating and evaluating the extraction effect, it provided technical support for further study of its crisis conversion.