‘MONOVAB’: An Annotated Corpus for Bangla Multi-label Emotion Detection

The most popular news portal's Facebook pages such as Prothom Alo, BBC Bangla, BD News 24, Bangla Tribune, Kaler Kantho, Daily Jugantor are picked to build the dataset. Following a manual collection of posts, a total of 130 posts for 11 news topics were obtained and converted into a CSV file. The dataset is annotated in Ekman's seven universal emotions and they are collected using a self-developed scraper algorithm.

本数据集选取当前最具人气的新闻门户网站的Facebook页面作为数据源以完成构建，涵盖Prothom Alo、BBC Bangla、BD News 24、Bangla Tribune、Kaler Kantho及Daily Jugantor等平台。经人工采集帖文后，共获取到覆盖11个新闻主题的130条帖文，并将其转换为CSV（Comma-Separated Values）格式文件。本数据集采用埃克曼（Ekman）提出的七种通用情绪体系进行标注，且所有帖文均通过自研爬虫算法完成采集。