Event ·Single source

LlamaIndex 用 ExtractBench 解析文档抽取的置信度过滤实践

LlamaIndex 发布了一篇关于文档抽取中置信度分数用法的文章，核心观点是分数的价值在于指导自动化决策。文章覆盖置信度阈值、精确率与召回率、分数覆盖率、分数粒度和人工审核量五个要点。基于 ExtractBench 的对比显示，在 97% 精确率目标下，LlamaParse Agentic Plus 过滤后召回率达到 66.48%。文章强调应把置信度用于控制生产环境的自动化比例和人工审核量。

Current read

做文档抽取的可以看看，LlamaIndex 用 ExtractBench 实测了置信度过滤怎么用，97% 精确率下 LlamaParse 召回 66.48%，教你算该放多少单自动跑。

2 sources58° AI heatLast updated 2026/9/24 18:41:23

Evidence

2 sources

Primary sourceLlamaIndex

2026/9/24 18:41:23View source

Related source 1Jerry Liu

2026/9/24 19:42:47View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story