BanglaTOD Dataset

This repository contains datasets collected and organized for research on Bangla Task Oriented Dataset. It is a A multi-domain, multilingual dataset for intent classification, and slot filling in Bangla. Every utterance is written in parallel across three linguistic forms: Bangla (Bengali script), English, and Banglish (Romanized Bangla) to support cross-lingual and code-mixed NLU research for a widely spoken but low-resource language.

Dataset Structure

The datasets are organized into the following domains:
BNLU Dataset/
├── Healthcare.json
├── Restaurant,Tourist,Travel,Hotel.json
├── E-commerce.json
├── Education.json
├── Mobile_Financial_Services.json
└── Government_Services.json

Overview

Property	Value
Total Dialogues	209
Languages	3 (Bangla, English, Banglish)
Domains	9
Intents	64

Domain coverage

Domain	Dialogues	Intents
Education	50	9
Government Service	32	19
Healthcare	25	5
Mobile Financial Service	25	5
E-commerce	25	10
Restaurant	16	5
Tourist	14	5
Travel	12	4
Hotel	10	2
Total	209	64

Data Format

Each file is a JSON array of dialogue objects. Every dialogue has:

Field	Description
dialogue_id	Unique identifier for the dialogue
domains	List of domain(s) the dialogue belongs to
intent	The single dominant user goal for the dialogue
turns	Ordered list of conversational turns