期刊名称：International Journal of Education and Management Engineering(IJEME)
印刷版ISSN：2305-3623
电子版ISSN：2305-8463
出版年度：2021
卷号：11
期号：6
页码：39-48
DOI：10.5815/ijeme.2021.06.05
语种：English
出版社：MECS Publisher
摘要：News categorization from various newspapers is important as readers want to read the news by category. But, the readers face difficulty if the news from different categories is presented without any order. This study aims to determine the category of news from online Bangla newspapers. In this context Bangla news headlines data, along with its categories, were collected from various online newspapers through scrapping. Eight categories of news are considered for this work and the headlines of the news are used for categorization. The input data is modeled by the LSTM and GRU neural networks, and the predicted category is compared with the actual category. For LSTM model, the result gives an accuracy of 82.74% and GRU model, The result gives an accuracy of 87.48%. GRU accuracy is higher than LSTM. Because, GRU training performance is faster than that of LSTM. In GRU 64 units used and in LSTM 128 units used for this research. For this reason, it also suggests that the GRU model gives better results than that of LSTM.
