Multiclass Job Recommendation System in the IT Field between Classification and Prediction Method

In general, the process when recruiting workers today takes a lot of energy, money, and time. But with various job recommendation systems that utilize Artificial Intelligence, it is very beneficial for applicants and companies. In this paper, we will provide a comparison of the methods used in the job recommendation system which has more accurate performance, especially in the IT field. There are two methods used, namely the classification method with the KNN algorithm and the prediction method with the SVM algorithm. Experimental begins with processing a dataset containing job vacancies information to carrying out an accuracy training/testing process using Google Collab. The result is that the classification method using the KNN algorithm gives an accuracy value of 69.24%, an F1 score of 68.19%, training time of 0.389 seconds, and test time of 2.278 seconds. While the SVM is divided into two models, namely the RBF model has an accuracy value of 69.08%, an F1 score of 67.94%, a training time of 31.025 seconds, a test time of 7.273 seconds and the Poly model has an accuracy value of 68.09%, an F1 score of 67.20%, a training time of 13.393 seconds, and a test time of 3.209 seconds. With the results obtained from the experimental, the model using KKN has a better performance than the model using SVM in the job recommendation system in the IT field.