Research On Recommendation Method Of Safety Learning Materials Based On Collaborative Filtering

In enterprise production,automation,scale,and regularization have become inevitable for the further development of enterprises.As the company's revenue increases,its staff expands,and the complexity of production equipment increases,it raises higher requirements for enterprise management,and it requires more and more formalization of enterprise production,especially as a production safety issue.It has become the lifeline of the enterprise.Therefore,enterprises have put forward higher requirements for the safety knowledge of employees' learning reserves,requiring employees to understand and comply with more knowledge of safety regulations,including operating safety specifications,operating safety specifications,management safety specifications,and so on.With the increasing content of new knowledge,it is imperative to train and evaluate employees against this knowledge.In order to improve the learning efficiency in unit time,implement innovative learning in a networked manner,save learning time,and solve the problem of overloading learning data and information,a more efficient and intelligent learning recommendation system is required to complete it.This paper first uses the Chinese word segmentation method to process the word segmentation of the learning material,convert the unstructured data in the learning material into structured data,and extract feature information,and then use the results of the word segmentation as the characteristics of the learning material.Then use the feature extraction algorithm to extract features for each learning material.Considering the limitations of the TFIDF algorithm,in order to improve the accuracy of the feature extraction,this article improves on the basis of the TFIDF algorithm,adding a number of more important influence factors.Clustering algorithms are used to classify the learning materials after feature extraction,laying the foundation for the next recommendation.Secondly,there are many existing recommendation algorithms.After a variety of comparative studies,the latter is selected among content-based recommendation algorithms and collaborative filtering recommendation algorithms.Considering that the number of personnel in the safety training assessment system is greater than that of the learning materials,On the order of magnitude,this article selects the item-based collaborative filtering algorithm as the core algorithm of the recommendation system.This article proposes an idea of extending the user's score on items based on the approximation of the items to solve the problem of data sparseness.Can effectively improve the problem of data sparseness.Finally,in order to achieve content recommendation,it is nec essary to obtain the user's score on the learning materials.For this data,there are data tables in the database that specifically record the scoring situation and update it in real time,record the user's score on different learning materials,and record their neighbors The score is also recorded,and a similarity matrix is constructed based on the user score,and the final recommendation result is obtained according to the score of the matrix.The results show that the improved item-based collaborative filtering algorithm can effectively provide suitable learning materials for power plant employees,and give priority to the content materials that employees urgently need to learn,thereby achieving the purpose of intelligent enterprise safety training.