Design and Analysis of News Category Predictor

摘要

Recent technological advancements have changedsignificantly the way news is produced, consumed, anddisseminated. Frequent and on-spot news reporting has beenenabled, which smartphones can access anywhere and anytime.News categorization or classification can significantly help in itsproper and timely dissemination. This study evaluates andcompares news category predictors' performance based on foursupervised machine learning models. We choose a standarddataset of British Broadcasting Corporation (BBC) newsconsisting of five categories: business, sports, technology, politics,and entertainment. Four multi-class news category predictorshave been developed and trained on the same dataset: Na?veBayes, Random Forest, K-Nearest Neighbors (KNN), andSupport Vector Machine (SVM). Each category predictor'sperformance was evaluated by analyzing the confusion matrixand quantifying the test dataset's precision, recall, and overallaccuracy. In the end, the performance of all category predictorswas studied and compared. The results show that all categorypredictors have achieved satisfactory accuracy grades. However,the SVM model performed better than the four supervisedlearning models, categorizing news articles with 98.3% accuracy.In contrast, the lowest accuracy was obtained by the KNN model.However, the KNN model's performance can be enhanced byinvestigating the optimal number of neighbors (K) value.