Document review with Unsiloed and TypeSafe Jev
What happened
On October 13, Hacker News featured a document review workflow that combines Unsiloed with TypeSafe Jev: one check covers how reliably fields were read, the other covers whether the extracted content makes sense, and flagged records go to a human. The example is a six-line medication list. A dose partly covered by a stain scores 0.74 on reading and 0.94 on plausibility. Another line, clearly printed, has an abnormal dosage that scores 0.99 on reading and 0.04 on plausibility. The two checks trigger different review reasons, separating an unreliable read from content that does not make sense.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageDocument Review with Unsiloed and TypeSafe Jev
Unsiloed 与 TypeSafe Jev 可组合成文档复核流程，分别检查字段识读可靠性与提取内容的合理性，再将异常记录交给人工复核。六行用药清单示例中，受污渍遮挡的剂量识读分数为 0.74、合理性概率为 0.94，另一行清晰印出的异常用量识读分数为 0.99、合理性概率为 0.04，两类检查分别触发不同复核原因。
Heat over time
Heat now 9·Comparable peak 10（Oct 13 17:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.