GIẢI PHÁP PHÂN LOẠI CHỦ ĐỀ TỰ ĐỘNG CHO BẢN TIN THỜI SỰ TRUYỀN HÌNH BẰNG KỸ THUẬT HỌC MÁY

Abstract

Video classification using machine learning has become a promising field, aiding in the automatic recognition and categorization into corresponding groups. This process begins with preprocessing video data to extract and convert information into numerical features. Specifically, machine learning algorithms such as KNN, SVM, CNN, and PhoBERT are employed to process and analyze the video content as well as language information within the video. In the experiment, data was collected from the internal storage system of the Can Tho City Radio and Television Station, with each video averaging about 3 minutes in length. These algorithms were deployed and evaluated on this dataset to measure and compare performance. The results of the PhoBERT algorithm achieved an accuracy rate of up to 98%. These results demonstrate the outstanding capability of PhoBERT in processing and recognizing video content, paving the way for the development of an automatic video classification system.