Research On Risk Grading Early Warning Method Based On Security Inspection Text Data

Nowadays,with the rapid development of business intelligence and artificial intelligence,it will be a great breakthrough in the history of security development if machine intelligence and security can be organically combined to form security intelligence.The in-depth research of natural language has made great progress in the fields of emotion analysis,artificial intelligence,text classification and so on,which also provides a feasible basis for the use of natural language related technologies to process security check text data.Many times of hidden danger investigation or safety inspection accumulated a large number of hidden danger text data.The current problem is that most of these text data are only used for storage and query,and the valuable information that can really reflect the law of building materials industry has not been fully mined.Therefore,this paper aims to apply NLP technology to the processing of security related text information,through the analysis of common hidden danger text expression statements,mining valuable hidden information,and optimizing enterprise security management.Taking the text data obtained from the safety inspection of building materials industry as the research object,on the basis of summarizing the previous research results,this paper focuses on how to maximize the use of massive and complex hidden danger data and how to mine the manifestation of hidden danger.Firstly,the paper analyzes the mining process of text data for safety inspection.Firstly,it does the data preprocessing work,including the description of text length distribution,data enhancement,construction of thesaurus in the field of building materials safety and so on.Then,it introduces the identification idea of hidden danger entity and more detailed processing analysis.Then,how to present these processed and analyzed information to safety decision-makers in a more visual way is discussed.The NLP technology is applied to analyze these data in detail,and the final analysis results are visualized.The main contents of the analysis include word2 vec word vector model training,hidden danger text keyword association analysis and hidden danger risk clustering analysis based on K-means algorithm.The main contents of the exhibition include hidden danger high frequency word cloud,common hidden danger distribution tree,common hidden danger description,category analysis,and related sentence cloud of different topics in the clustering results.Finally,it discusses the construction idea of safety production early warning system,and how to optimize the enterprise hidden danger investigation and governance.For different types of hidden dangers,personalized early warning and prevention measures are put forward,such as:enterprises should pay attention to the development of safety education and training activities,pay attention to the formulation of corresponding systems and emergency plans,and avoid the occurrence of accidents due to the weak self-help ability of employees.To provide some theoretical support for safety management personnel in decision-making,comprehensively improve the level of intrinsic safety,and promote the development of safety.