Research And Development Of A Training System Based On Hybrid Intelligent Algorithm And Collaborative Filtering Technology

Nowadays,the country’s economy is in a booming stage of development and various fields are developing rapidly.At the same time,there is a surge in demand for related skilled personnel.The application of electronic information technology in the training field has overturned traditional enterprise training models.Taking urban rail transit training as an example,the use of information technology to precisely control important aspects such as teaching,testing,grading,and evaluation,which can effectively enhance employees’ skill levels and indirectly reduce the occurrence of dangerous accidents during metro operations.Currently,traditional training modalities mainly face the following challenges: firstly,production employees mostly use comprehensive working hours to complete their work,and traditional centralized training and testing models will bring great inconvenience to employees;secondly,the current manual test paper composition or randomly selecting test questions for exams,result in poor test quality and are difficult to meet the diversified talent selection of companies;thirdly,traditional training systems rely solely on annual training plans and lack completeness in improving employee knowledge systems.This thesis takes the urban rail enterprise training system as the background and develops a training management system based on hybrid intelligent algorithms and collaborative filtering technology.The system integrates genetic algorithms and ant colony algorithms as intelligent automatic test paper composition methods,and uses training material feature extraction and collaborative filtering technology to realize training material recommendation functionality.The main work completed includes:(1)Based on enterprise training scenario,summarized the principles of test paper composition,researched the composition strategy,constructed a mathematical model for the goal function to measure exam quality,and designed a set of composition problem-solving models.This solves previous test paper composition issues such as test question difficulty being easily overlooked,poor targeting of tests,incomplete knowledge points,and improper distribution of question types.(2)After comparing the advantages and disadvantages of various test paper composition algorithms,we ultimately chose to construct a test paper composition algorithm by combining directed mutation genetic algorithm with ant colony algorithm.Compared with the current manual test paper composition or the random selection of test questions to compose a test paper,the interface of our system is concise and easy for employees to operate,the way of composing a test paper is efficient and effective,and the system functions are reasonable and comprehensive.(3)By using Chinese word segmentation technology and TF-IDF to complete the feature extraction step,and combined with the fact that the training materials in the system are much greater than the number of users,we have used a project-centered collaborative filtering method to implement the recommendation system.Moreover,we have adopted the idea of using similar projects to expand the score to improve the problem of data sparsity.Through recommending training materials,we can indirectly supplement and improve the employees’ knowledge system,thus meeting the needs of enterprise talents.(4)The system adopts a modular programming idea,dividing the system according to function modules and user roles.The overall system adopts the B/S architecture,and we use Spring Boot and Vue’s front-end and back-end separation development technology to complete the project construction.