Explainable Detection: A Transformer-Based Language Modeling Approach for Bengali News Title Classification with Comparative Explainability Analysis Using ML & DL

This dataset is a Bangla dataset. There are total of 4 columns. The data is collected from different newspaper and more importantly, the dataset is raw. Total number of data is 6150. There are 3 classes in the Title Category. They are: National, International, and Sports—number of data from different categories, National-2089, Sports-2008, International-2053. We aim to find the title category from the title using ML, DL, and transformer-based models. As we said the data is in Bangla, after downloading the data might look corrupted. If the data is uploaded on the drive the data will look okay or the Microsoft Excel is up to date, the data will look OK.

本数据集为孟加拉语（Bangla）数据集，共包含4列数据。数据采集自各类报纸，且尤为关键的是，该数据集为原始未处理数据集，总样本量达6150条。标题类别共设有3类，分别为国内（National）、国际（International）与体育（Sports），各类别样本量依次为：国内类2089条、体育类2008条、国际类2053条。本数据集旨在通过机器学习（Machine Learning）、深度学习（Deep Learning）以及基于Transformer（Transformer）的模型，从文本标题中识别其所属类别。如前所述，本数据集为孟加拉语格式，下载后可能出现显示损坏的情况。若将数据集上传至云端存储，或使用最新版本的Microsoft Excel打开，则可正常显示。