Unverified50% confidenceTradeoffExact time
基于LLM语义理解的Agent通过理解元素语义含义来定位目标，即使页面HTML结构变化只要功能语义不变仍能正确识别，相比依赖CSS选择器或XPath的传统爬虫和RPA工具对页面改版适应性更强
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Related Claims
UnverifiedTraditional RPA tools locate operation targets by identifying fixed attributes of interface elements (such as CSS selectors, XPath, and element IDs), and scripts may break once the page's HTML structure changes.79% similarUnverified页面结构识别背后依托视觉语言模型（VLM）的多模态理解能力67% similarUnverifiedPage Agent can understand webpage structure and automatically perform operations such as clicking buttons, filling in input fields, and selecting dropdown menus based on natural language instructions.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/771827API
curl https://kongchang.com/api/v1/knowledge/claims/771827MCP
get_claim(id=771827)