Research On The Method Of Characterizing The Portraits For Scientific And Technical Personnel Based On Data Mining

As the digital construction of enterprises continues to deepen,the construction of numerical portrait models based on employee behavior data has gradually become a hot topic and tends to be a rigid demand for the digital construction of enterprise employees.As the enterprise activity behavior of employees is more complex,the behavior data generated by the enterprise activity behavior is large in volume,structured in many ways,and stored in various ways,etc.,the data utilization rate is low and does not play its value as an asset.Therefore,more and more scholars have begun to study intelligent portrait portrayal and behavior analysis methods,with the aim of improving the efficiency of data analysis,increasing data utilization,and thus improving the accuracy of portrait portrayal as well as behavior analysis.After a research on an oilfield research institute found that the current employee portrait portrayal process,mostly using statistical analysis or supplemented by simple machine learning methods,there are simple analysis means,low analysis efficiency,a single intelligent way to solve the problem is limited,making the portrait portrayal results are more one-sided.Therefore,the establishment of data mining-based portrait characterization model of scientific and technical personnel,design portrait characterization system,comprehensive and accurate portrayal of scientific and technical personnel portrait characteristics become the urgent need for the construction of digital portrait of the enterprise employees.In order to solve the above problems,this paper provides a profound analysis of the business process of portrait characterization of technology personnel in enterprises,proposes a method of characterizing the portraits of scientists and technicians based on data mining and designs a relevant model,analyzes the composition structure of the model in detail and implements it.Firstly,we deeply analyze the content,criteria and process of portrait feature portrayal of scientific and technical personnel,study the data model related to behavior analysis,design the general framework of portrait feature portrayal model,and clarify that the research focuses on diligence analysis of scientific and technical personnel and mining of temporal behavior patterns;secondly,in order to make data collection more accurate and perfect,we design a comprehensive data collection architecture for the characteristics of scientific and technical personnel enterprise behavior data,namely We use Zabbix,ELK,DB acquisition and network port interception to achieve multi-source heterogeneous behavioral data collection;again,we build a self-learning diligence analysis model based on improved GAN model,adopt random forest,statistical analysis and assignment system to analyze the behavioral characteristics of scientific and technical personnel,specify the evaluation criteria of scientific and technical personnel diligence and the training and validation data set of the model,and use the selected data set to train and optimize the model.Then,we propose the method of mining the behavior law of science and technology personnel in time sequence,and use random forest,outlier detection LOF and isolated forest to analyze the behavior of science and technology personnel in different time granularity,so as to mine the behavior law in time sequence,and use the LSTM model based on self-attentive mechanism to realize the rough prediction of science and technology personnel’s Finally,we design and implement the portrait characterization system based on the portrait characterization model based on data mining in the context of enterprise employee portrait characterization business,and verify the feasibility and usefulness of the proposed method through the actual application effect of the model.The research results show that the method of characterizing the portraits of scientists and technicians based on data mining can better solve the problem of digital image construction of enterprise employees,and the application of such enterprise employee behavior data can more accurately realize the portrait feature portrayal of technology employees.At the same time,through extended research and analysis,the proposed portrait feature portrayal method provides a new way of thinking for enterprise level employee digital portrait model construction and employee behavior analysis.