Elysia Tools Docs
ZH
语言
ZHESPTRUFRDEEN
返回主站
浏览工具

AI Tools

AI RAG 分块质量评分器

对文档的候选 RAG 分块方案在四个维度上评分——连贯性（干净的分句/分段边界）、覆盖率（主题聚焦 / 关键词集中度）、上下文重叠健康度以及分块大小一致性——并排对比最多三种方案并推荐最优。纯离线启发式，无模型调用。

html· HTML 结果打开工具

快速开始

用三种语言从你的代码中调用此工具。

cURL
curl -X POST 'https://api.elysiatools.com/zh/api/tools/ai-rag-chunk-quality-scorer' \
  -H 'Content-Type: application/json' \
  -d '{"document":"Retrieval-augmented generation combines a retriever and a language model. The retriever finds relevant passages in a knowledge base. Chunking strategy strongly affects retrieval quality. Smaller chunks are precise but lose context. Larger chunks carry context but dilute relevance. Overlap between chunks preserves context across boundaries. Evaluate several schemes before indexing.","unit":"tokens","method":"sentence","sizeA":256,"overlapA":32,"sizeB":512,"overlapB":64,"sizeC":1024,"overlapC":128,"renderMode":"comparison","precision":1}'

© 2026 Elysia Tools. 开发者文档。

返回主站浏览工具

API 参考

以 JSON 形式 POST 提交输入参数。文件类型参数需先单独上传。

请求端点

HTTP
POST https://api.elysiatools.com/zh/api/tools/ai-rag-chunk-quality-scorer

请求参数

参数名类型必填说明
documenttextarea是—
unitselect否—
methodselect否—
sizeAnumber是—
overlapAnumber否—
sizeBnumber是—
overlapBnumber否—
sizeCnumber是—
overlapCnumber否—
renderModeselect否—
precisionnumber是—

响应格式

HTML 结果

JSON
{
  "result": "<div>Processed HTML content</div>",
  "error": "Error message (optional)",
  "message": "Notification message (optional)",
  "metadata": {
    "key": "value"
  }
}

MCP 集成

将此工具加入你的 Model Context Protocol 服务，让 AI 智能体可以列出并调用它。

服务配置

将以下内容加入你的 MCP 客户端配置：

mcp.json
{
  "mcpServers": {
    "elysiatools-ai-rag-chunk-quality-scorer": {
      "name": "ai-rag-chunk-quality-scorer",
      "description": "对文档的候选 RAG 分块方案在四个维度上评分——连贯性（干净的分句/分段边界）、覆盖率（主题聚焦 / 关键词集中度）、上下文重叠健康度以及分块大小一致性——并排对比最多三种方案并推荐最优。纯离线启发式，无模型调用。",
      "baseUrl": "https://api.elysiatools.com/mcp/sse?toolId=ai-rag-chunk-quality-scorer",
      "command": "",
      "args": [],
      "env": {},
      "isActive": true,
      "type": "sse"
    }
  }
}

列出可用工具

连接到 SSE 端点后，列出已开放的工具：

tools/list
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/list"
}

调用此工具

通过工具 id 调用，参数由其参数表构建：

tools/call
{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "ai-rag-chunk-quality-scorer",
    "arguments": {
  "document": "Retrieval-augmented generation combines a retriever and a language model. The retriever finds relevant passages in a knowledge base. Chunking strategy strongly affects retrieval quality. Smaller chunks are precise but lose context. Larger chunks carry context but dilute relevance. Overlap between chunks preserves context across boundaries. Evaluate several schemes before indexing.",
  "unit": "tokens",
  "method": "sentence",
  "sizeA": 256,
  "overlapA": 32,
  "sizeB": 512,
  "overlapB": 64,
  "sizeC": 1024,
  "overlapC": 128,
  "renderMode": "comparison",
  "precision": 1
}
  }
}
可用逗号分隔的 toolId 在一个会话中串联多个工具，例如 /mcp/sse?toolId=png-to-webp,jpg-to-webp,gif-to-webp（最多 20 个）。

有问题或反馈？请联系 [email protected]

浏览工具
打开工具