Construction Of Brucellosis Related Knowledge Base Based On Literature Mining

Brucellosis caused by Brucella is one of the important zoonotic infectious diseases.It is widespread in the world and poses a serious threat to the breeding industry and human health.Its pathogenic and immune mechanisms are still unclear.Brucella virulence genes and immune regulatory genes and their relationships are often submerged in massive scientific literature data,making it difficult to fully understand and systematically study their pathogenic and immune molecular mechanisms.This thesis provides new ideas and clues for the systematic study of the pathogenic and immune mechanisms of Brucella and the development of therapeutic drugs and vaccines by constructing a knowledge base of brucellosis associations.Based on literature mining,it takes brucellosis-related literature abstracts as the research object,takes named entity recognition and relation extraction as research methods,and combines knowledge graphs to build a brucellosis-related knowledge base in this article.The main research contents and conclusions are as follows:(1)The brucellosis literature abstracts retrieved from PubMed by reptile technology were preprocessed,and the retained literature abstract information was screen.Finally,the literature abstracts on brucellosis in cattle and sheep needed for the study were obtained and as a dataset for experiments.(2)Based on the application of the Glove character joint representation-BLSTM-CRF model of the biomedical named entity recognition.An F1 value of 75.62% was obtained on the JNLPBA 2004 general data set,and identified different types of entities such as diseases,genes,proteins,compounds,etc.in the literature abstract.(3)Based on the application of Attention+BiGRU+Gumbel Tree gating unit model of the biomedical relationship extraction.An F1 value of 73.1% was achieved on the general data set DDI.The relationships between entities such as genes,proteins,compounds,etc.in documents was extracted through the model.(4)The brucellosis association knowledge base constructed based on Neo4j graph database.The base integrated brucellosis-related information,performed association analysis on brucellosis-related entities,and realized brucellosis gene association data query and visualization operations,which provided a more intuitive and convenient visual analysis.(5)Search PubMed for articles on the incidence and prevalence of brucellosis and analyze the data on livestock brucellosis epidemics in mainland China from 2016 to 2020.At the same time,To map the national distribution of the incidence of brucellosis,collect information on livestock brucellosis epidemics in the Veterinary Bulletin of the Ministry of Agriculture of the Public Health Science Data Center.