Aiming at the problem of inefficient processing of a large number of work orders in a communication company

this paper proposes an optimisation scheme combining text classification and collaborative filtering recommendation system based on term frequency-inverse document frequency (TF-IDF) algorithm to improve the accuracy of work order classification and the reasonableness of recommendation. The data preprocessing of customer problems in work orders

including word splitting

deactivation and text cleaning

combined with multidimensional features such as the channels through which customers raise problems

the TF-IDF method is used to extract features from the problem text

and the optimal classification hyperplane processing is performed on the contents of work orders through the support vector machine (SVM) model to maximize the spacing between work orders of different categories and to maximize the spacing between work order categories and the spacing between work order categories. The optimal classification hyperplane is applied to the contents of the work order through the SVM model

so as to maximize the spacing between different categories of work orders

and to classify new work orders through the decision function. A telecommunication company can use this to efficiently classify user work orders and designate special processing procedures for different categories of work orders to improve work efficiency and customer satisfaction.