用Transformer分析孟加拉语电影剧集评论，先判相关性再测情感。
CineXDrama: Relevance Detection and Sentiment Analysis of Bangla YouTube Comments on Movie-Drama using Transformers: Insights from Interpretability Tool
- 先过滤无关评论，再对相关评论做情感分类。
- 孟加拉语BERT在相关性检测上准确率达83.99%，情感分析达93.3%。
- 结合LIME工具提升模型可解释性，适合本地化内容分析研究者。
近年来，YouTube已成为孟加拉语电影与剧集的主要传播平台，观众通过评论表达对内容的看法。然而，并非所有评论都适合作为情感分析的依据，因此需要筛选机制。本文提出一个系统：首先评估评论的相关性，再对相关评论进行情感分析。我们构建了一个包含14,000条人工收集并预处理的评论数据集，标注了相关性（相关/无关）和情感（正面/负面）。采用八种Transformer模型（包括BanglaBERT）完成分类任务，其中BanglaBERT在相关性检测中达到83.99%的准确率，在情感分析中达到93.3%。研究还集成LIME以解释模型决策过程，增强透明度。
原文摘要 · Abstract (English)
In recent years, YouTube has become the leading platform for Bangla movies and dramas, where viewers express their opinions in comments that convey their sentiments about the content. However, not all comments are relevant for sentiment analysis, necessitating a filtering mechanism. We propose a system that first assesses the relevance of comments and then analyzes the sentiment of those deemed relevant. We introduce a dataset of 14,000 manually collected and preprocessed comments, annotated for relevance (relevant or irrelevant) and sentiment (positive or negative). Eight transformer models, including BanglaBERT, were used for classification tasks, with BanglaBERT achieving the highest accuracy (83.99% for relevance detection and 93.3% for sentiment analysis). The study also integrates LIME to interpret model decisions, enhancing transparency.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。