sahilmaniyar888/Indian_Climate_Disaster_data

BharatCRIC印度气候灾害数据集（热浪建议+诈骗对）是一个多语言数据集，包含印度多个地区的热浪建议和诈骗对。数据集支持印地语、英语、博杰普尔语、迈蒂利语和桑塔利语（Ol Chiki脚本），覆盖比哈尔邦、北方邦和贾坎德邦等地区。数据集包含结构化指令上传、烟雾测试和真实的与诈骗的偏好对。数据集适用于文本生成和文本分类任务，特别是语言建模和多类分类。数据集的文件包括完整的结构化指令上传、50行的烟雾测试、以及真实的与诈骗的偏好对。数据集严格遵守验证热线仅在真实行中的规则，诈骗行明确标记为`variant_type=heat_scam`并包含故意的假号码/支付/OTP标记。桑塔利行仅通过手动/本地验证，没有自动扩展。

BharatCRIC: Indian Climate Disaster Data (Heatwave Advisories + Scam Pairs) is a multilingual dataset containing heatwave advisories and scam pairs from various regions in India. The dataset supports Hindi, English, Bhojpuri, Maithili, and Santali (Ol Chiki script), covering states like Bihar, Uttar Pradesh, and Jharkhand. It includes structured instruction uploads, smoke tests, and genuine-vs-scam preference pairs. The dataset is designed for text-generation and text-classification tasks, specifically language modeling and multi-class classification. The files include full structured instruction uploads, a 50-row smoke test, and mirrored genuine-vs-scam preference pairs. The dataset adheres to strict rules: validated helplines only in genuine rows; scam rows are explicitly tagged `variant_type=heat_scam` and carry intentional fake-number/payment/OTP markers. Santali rows are manual/native-validated only, with no automatic Adaption expansion.