standd/saas-taglines-rated

该数据集名为saas-taglines-rated，包含158个SaaS（软件即服务）英雄标签行，这些标签行是从大约14个垂直领域（如开发工具、金融技术、人工智能、数据、安全、营销等）的真实产品落地页中抓取的。每个标签行都通过一个大型语言模型（LLM）法官根据一个标签行品质评分标准进行了1到5分的评级。评分标准涵盖多个维度：价值超越类别、差异化、清晰度、记忆性、共鸣/抱负，其中类别标签得分低是原则。数据集以JSONL格式提供，字段包括：公司名称（company）、垂直领域（vertical）、标签行（tagline）、副标题（subtitle）、置信度（confidence）、品质评分（quality，范围1-5）和品质原因（quality_reasons）。该数据集适用于文本分类任务，特别是用于分析营销文案、SaaS标签行的质量评估和相关NLP研究。

The dataset is named saas-taglines-rated, containing 158 SaaS hero taglines scraped from real landing pages across approximately 14 verticals (e.g., dev tools, fintech, AI, data, security, marketing, …). Each tagline is rated 1-5 on a tagline-quality rubric by an LLM judge. The rubric dimensions include value-over-category, differentiation, clarity, memorability, and resonance/ambition, with category labels scoring low by principle. The data is provided in JSONL format with fields: company, vertical, tagline, subtitle, confidence, quality (1-5), and quality_reasons. It is suitable for text-classification tasks, particularly for analyzing marketing copy, SaaS tagline quality assessment, and related NLP research.