Tasks কাজসমূহ
Every resource in the catalog is organized under a task. Each task page collects its datasets, leaderboard, papers, and model checkpoints.
Polarity and emotion classification of Bangla text from social media, reviews, and news comments.
Detecting person, location, organization and other entities in Bangla text.
Bangla↔English and multilingual translation, parallel corpora and paraphrase generation.
Extractive and generative QA over Bangla passages and knowledge.
Abstractive and extractive summarization of Bangla news and documents.
Identifying hateful, offensive, and abusive Bangla content online.
Bangla speech recognition and text-to-speech corpora and models.
Recognizing emotion (anger, happiness, sadness, and more) from the acoustic speech signal — distinct from text-based emotion classification under Sentiment Analysis.
Identifying and distinguishing speakers in Bangla audio — speaker identification, verification, and diarization (who spoke when).
News categorization, fake news detection, and document classification.
Part-of-speech tagging, treebanks, and morphological analysis for Bangla.
Evaluating large language models on Bangla knowledge, reasoning, and generation.
Answering questions about images in Bangla, and evaluating vision-language models on Bengali visual and cultural content.
Recognising printed and handwritten Bangla text at word, line, document and scene level, and extracting information from document images. Isolated character and digit classification is out of scope — see TODO-data.md.
Detecting, correcting and explaining spelling and grammatical errors in Bangla text.
Resolving pronouns and linking mentions that refer to the same entity in Bangla text.
Deciding whether one Bangla sentence entails, contradicts or is neutral towards another.
Choosing the intended sense of an ambiguous Bangla word in context, and inducing sense inventories.
Restoring punctuation and sentence boundaries in unpunctuated Bangla text, typically ASR output.
Rewriting Bangla text into a different style or register while preserving its meaning.
Identifying the position a Bangla text takes towards a target claim, person or topic.
Measuring how hard Bangla text is to read, including readability formulas and lexical complexity prediction.
Generating Bangla descriptions of images and video.
Open-domain and task-oriented Bangla conversational systems, and the corpora behind them.
Translating Bangla speech into text or speech in another language, cascaded or end-to-end.
Linking Bangla entity mentions to entries in a knowledge base. No Bangla-specific work was found when this task was added — see TODO-data.md.