README.md

---
tags:
- codemixed
- dataset
- roman
- bengali
- english
- transaltion
- transliteration
- bengali pattern
- codemixed english
- pattern
---
---
RBE_Sent Dataset Description:

The RBE_Sent (Roman Bengali-English Sentiment) dataset is a synthetic, gold-standard code-mixed dataset developed for sentiment analysis tasks involving Romanized Bengali and English. It captures real-world bilingual usage by blending Roman Bengali with English tokens within the same textual instances. The dataset is designed to support research in multilingual natural language processing, especially in the context of low-resource, code-mixed languages. Each entry in RBE_Sent is annotated for sentiment, enabling supervised learning and evaluation. By providing a structured and labeled resource in this underexplored linguistic domain, RBE_Sent contributes to advancing computational methods for understanding Bengali-English code-mixed communication.

This Dataset contains both Bengali pattern and English pattern codemixed product review text. Dataset has 36,172 roman Bengali-English codemixed product reviews which are combination of 18,086 Bengali pattern and 18,086 English pattern codemixed data.

license: mit
language:
- bn
- en
tags:
- codemixed
- product
- review
- ecommerce
- bengali
- dataset
pretty_name: 


RBE_Sent codemixed online product reviews yield around 49 product categories.  Each of these product categories contains unique products with many reviews.