Research and application of Chinese Entity Relation Extraction Based on Cyberspace Security

摘要

Entity relation extraction has played an important role in the construction of semantic knowledge and knowledge graph, Chinese domain entity relation extraction has become more and more important as well. In order to improve the accuracy and practicality, we propose a Chinese entity relation extraction model based on deep neural network, which named BBCM(Bert-BiLSTM-CRF Model). The model uses the Bert language pre-training model to embed words in the corpus data, combines self-labeled "cyberspace security" field data to fine- tune parameters, which improves the model's ability to extract semantic features. The BiLSTM layer in the model can effectively memorize the contextual semantic information of the data, and combine the CRF loss function to obtain the optimal prediction relation. On the experiment of the standard dataset, the BBCM model has a significant improvement(F1 value reached 0.9544) than the baseline model. 7749 items are valid data after verification, and the application effect is obvious.