CSR Report Sentences and Additional Datasets

Corporate Social Responsibility (CSR) Dataset: This dataset comprises 21,782 distinct sentences extracted from sustainability reports of 100 publicly listed companies across 26 diverse industry sectors. The data was collected from 2024 CSR disclosure documents, with comprehensive metadata including source identifiers, company information, industry sector, and page numbers to enable thematic pattern analysis of corporate sustainability communication strategies. Yahoo Answers Dataset: This community-driven dataset contains question-answer pairs from two knowledge-intensive domains - "Science & Mathematics" and "Computers & Internet" categories. The researchers extracted 2,000 balanced instances from the original dataset of 140,000 training and 6,000 testing samples per category, focusing on content with overlapping vocabularies but distinct conceptual boundaries relevant to technical support and knowledge management systems. IMDB Movie Review Dataset: This sentiment analysis dataset consists of 50,000 movie reviews equally distributed between positive and negative sentiment categories, from which 2,000 instances were extracted through balanced stratified sampling. The dataset represents emotionally-charged textual content with colloquial language patterns, designed to validate opinion mining applications requiring sophisticated analysis of subjective narrative text containing complex emotional nuances and evaluative language.企业社会责任（Corporate Social Responsibility, CSR）数据集：该数据集涵盖从覆盖26个多元化行业板块的100家公开上市企业的可持续发展报告中提取的21782条独立语句。数据采集自2024年CSR披露文件，附带完整元数据，包括来源标识符、企业信息、所属行业板块及页码，可用于开展企业可持续传播策略的主题模式分析。 雅虎问答（Yahoo Answers）数据集：该社区驱动型数据集包含来自“科学与数学（Science & Mathematics）”与“计算机与互联网（Computers & Internet）”两个知识密集型领域的问答对。研究人员从每个类别的原始数据集（含140000条训练样本与6000条测试样本）中提取了2000条均衡实例，重点关注词汇重叠但概念边界清晰的内容，此类内容与技术支持及知识管理系统相关。 IMDB电影评论（IMDB Movie Review）数据集：该情感分析数据集包含50000条电影评论，按积极与消极情感类别平均划分。研究人员通过均衡分层抽样从中提取了2000条实例。本数据集包含带有口语化表达的情绪化文本内容，旨在验证需要对包含复杂情感细微差别与评价性语言的主观叙事文本开展精细分析的观点挖掘应用。