搜索与检索

jev-document-classification

@Charlyhno-eng2TypeScriptMIT更新于 2026-09-19可当场跑

JEV Document Classification 利用 TypeSafe 的 System One 模型,快速且低成本地对文本文件进行分类。

英文原文

JEV Document Classification enables the rapid and cost-effective classification of text-based documents using AI, leveraging TypeSafe's "System One" model.

Charlyhno-eng/jev-document-classification

它在哪儿调用了 Jev

import { createGateway, experimental_evaluate as evaluate } from 'ai';

server/classification.ts:1

链接指向我们抓取当天的那个 commit,行号是准的。

它问 Jev 的问题

下面是从这个项目源码里原样取出来的 question 组合。

取自 server/classification.ts:113

  1. confidentiality选项

    Assess the sensitivity of the document content. Choose the highest applicable confidentiality level.

    • PublicSafe to share publicly; contains no sensitive business or personal information.
    • InternalFor internal use; contains routine business information that should not be public.
    • ConfidentialContains sensitive business, financial, contractual, or personal information.
    • RestrictedContains highly sensitive personal data, credentials, health data, legal privilege, or critical business secrets.

用你自己的内容跑一遍

代码
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    confidentiality: { type: 'choice', instructions: 'Assess the sensitivity of the document content. Choose the highest applicable confidentiality level.',
      criteria: { Public: 'Safe to share publicly; contains no sensitive business or personal information.', Internal: 'For internal use; contains routine business information that should not be public.', Confidential: 'Contains sensitive business, financial, contractual, or personal information.', Restricted: 'Contains highly sensitive personal data, credentials, health data, legal privilege, or critical business secrets.' } },
  },
});

同类的其他项目