智能体
x.com ·
@llama_index
LlamaIndex 讨论文档抽取中置信度分数的实用评估维度
LlamaIndex 发文讨论如何判断抽取置信度分数是否有用：关键不在分数本身，而在于能否据此在生产中控制自动化与人工复核。文章覆盖置信度阈值、精确率与召回率、分数覆盖率、分数粒度以及人工复核量，并用 ExtractBench 比较不同抽取系统在置信度过滤后的表现。在 97% 精确率目标下，LlamaParse Agentic Plus 过滤后在期望字段上达到 66.48% 召回率。
AI推荐理由 给出可验证的具体指标：在 97% 精确率目标下，经置信度过滤后 LlamaParse Agentic Plus 在期望字段上的召回率为 66.48%，并列出置信度阈值、精确率/召回率、覆盖率、粒度与人工复核量等评估维度，对文档抽取落地有参考价值。局限是数据来自厂商自家 ExtractBench 与自家产品，属单方基准，且正文未说明测试集规模与构造方式。
68
浏览 1,594
点赞 8
评论 5
转发 2
正文
confidence scores only matter if they help you decide what to automate.
for document extraction, that usually means knowing how much work you can safely accept at a given precision target. in our latest post, we look at confidence scoring through that lens, including:
✅️ confidence cutoffs
✅️ precision vs. recall
✅️ score coverage
✅️ score granularity
✅️ human review volume
using ExtractBench, we compare how different extraction systems perform after confidence filtering. at a 97% precision target, LlamaParse Agentic Plus reached 66.48% recall on expected fields after filtering.
the useful part of a confidence score isn’t the number itself. it’s whether you can use it to control automation and review in production.
👉️ read the full post:[llamaindex.ai/blog/what-make…](https://www.llamaindex.ai/blog/what-makes-an-extraction-confidence-score-useful)f
[Your browser does not support the video tag.](https://video.twimg.com/ext_tw_video/2103193075321282560/pu/vid/avc1/1096x720/R7Lfsv0JSJbZ301B.mp4?tag=12)
[🔗 View on Twitter](https://x.com/llama_index/status/2103193093247750466/video/1)