A corpus of 300,000 (full dataset) Bangla sentences labeled according to their sentiment (positive, negative). Currently, only 4000 sentences are available in this repository. If you need the full version, please don't hesitate to drop us an email. The sentences were collected from social media sites, blogs and news portals. It can be used to train Sentiment Analysis systems. It has a separate training set (200,000 sentences) and test set (100,000 sentences). This dataset is large enough to apply Deep Learning in Bangla Sentiment Analysis, as Deep Learning models inherently require more data to be useful in practice.
The corpus is released in plain text format. It has two seperate files (bangla.pos, bangla.neg) for positive and negative labeled sentences
If you need the full version, we can arrange a way to send the dataset to you. Please email at contact@socian.ai
The corpus is licensed under GNU GPLv3, making it very easy to anyone to use the data for any purpose.