README.md

metadata

license: mit
task_categories:
  - text-classification
language:
  - bn

Sentiment Analysis Data for the Bengali Language

Dataset Description: This dataset contains a sentiment analysis dataset from Sazzed et al. (2020).

Data Structure: The data was used for the project on improving word embeddings with graph knowledge for Low Resource Languages.

Citation:

@inproceedings{sazzed-2020-cross,
    title = "Cross-lingual sentiment classification in low-resource {B}engali language",
    author = "Sazzed, Salim",
    editor = "Xu, Wei  and
      Ritter, Alan  and
      Baldwin, Tim  and
      Rahimi, Afshin",
    booktitle = "Proceedings of the Sixth Workshop on Noisy User-generated Text (W-NUT 2020)",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2020.wnut-1.8",
    doi = "10.18653/v1/2020.wnut-1.8",
    pages = "50--60",
    abstract = "Sentiment analysis research in low-resource languages such as Bengali is still unexplored due to the scarcity of annotated data and the lack of text processing tools. Therefore, in this work, we focus on generating resources and showing the applicability of the cross-lingual sentiment analysis approach in Bengali. For benchmarking, we created and annotated a comprehensive corpus of around 12000 Bengali reviews. To address the lack of standard text-processing tools in Bengali, we leverage resources from English utilizing machine translation. We determine the performance of supervised machine learning (ML) classifiers in machine-translated English corpus and compare it with the original Bengali corpus. Besides, we examine sentiment preservation in the machine-translated corpus utilizing Cohen{'}s Kappa and Gwet{'}s AC1. To circumvent the laborious data labeling process, we explore lexicon-based methods and study the applicability of utilizing cross-domain labeled data from the resource-rich language. We find that supervised ML classifiers show comparable performances in Bengali and machine-translated English corpus. By utilizing labeled data, they achieve 15{\%}-20{\%} higher F1 scores compared to both lexicon-based and transfer learning-based methods. Besides, we observe that machine translation does not alter the sentiment polarity of the review for most of the cases. Our experimental results demonstrate that the machine translation based cross-lingual approach can be an effective way for sentiment classification in Bengali.",
}