Unverified50% confidenceFactExact time
上下文压缩主要有抽取式摘要和生成式摘要两种路径，智能体场景中生成式摘要更常见
1
Sources
50%
Confidence
Long-term
Relevance
8/27/2026
First Seen
Sources
Related Entities
Related Claims
Unverified上下文压缩的三种主流实现模式包括滚动摘要、关键信息提取和分层记忆76% similarUnverifiedAutomatic summarization technology is divided into two types: extractive and abstractive; extractive directly selects key sentences from the original text, while abstractive understands the semantics and reorganizes the language.73% similarUnverified相比截图方案，纯文本提取从源头获取结构化信息，处理链路更短更省资源，且更容易做精细化的过滤和脱敏，Markdown 保留标题、列表、链接、代码块等结构信息72% similarUnverifiedCommon Token compression strategies include three approaches: summary extraction, semantic deduplication, and structural compression.70% similarUnverified压缩功能更适合处理冗余日志，而非承载关键推理链的对话历史68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/810055API
curl https://kongchang.com/api/v1/knowledge/claims/810055MCP
get_claim(id=810055)