English(EN) Turns out you can't just ask an LLM for CSS selectors and ship them. In our scraping system, first-attempt selectors returned nothing 30 to 40% of the time. The

LLM 生成的 CSS 选择器需要验证才能进行网页抓取

作者 PulseAugur 编辑部 · [1 个来源] · 2026-07-22 19:59

尝试使用大型语言模型生成 CSS 选择器进行网页抓取被证明是不可靠的，首次尝试失败的比例高达 30-40%。解决方案包括首先检查 JSON-LD 数据，然后在实际使用前针对实际 DOM 验证每个生成的选择器。 AI

影响强调了在将大型语言模型的输出集成到自动化系统中时，需要验证层。

排序理由该条目描述了大型语言模型在特定工具（网页抓取 CSS 选择器）应用中的实际应用和局限性。

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

报道来源 [1]

Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] · 2026-07-22 19:59

Turns out you can't just ask an LLM for CSS selectors and ship them. In our scraping system, first-attempt selectors returned nothing 30 to 40% of the time. The

Turns out you can't just ask an LLM for CSS selectors and ship them. In our scraping system, first-attempt selectors returned nothing 30 to 40% of the time. The trick that made it work: check JSON-LD first, then run every generated selector through a validation loop against the r…

链接 ombulabs.ai/…/ai-driven-scraping.html

报道来源 [1]

Turns out you can't just ask an LLM for CSS selectors and ship them. In our scraping system, first-attempt selectors returned nothing 30 to 40% of the time. The

相关实体

相关话题