eval-relevance

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Eval Relevance

相关性评估

Use this skill to evaluate how relevant an assistant response is to the user’s request.

使用此技能来评估助手回复与用户请求的相关性。

Inputs

输入要求

Require:

The assistant response text to evaluate.
(Optional) The user’s original request for comparison.

必填：

待评估的助手回复文本。 -（可选）用于对比的用户原始请求。

Internal Rubric (1–5)

内部评分标准（1–5分）

5 = Directly addresses the user’s request, stays fully on-topic, and prioritizes what the user actually asked
4 = Mostly relevant, minor digressions or small omissions
3 = Partially relevant, addresses the general topic but misses key parts of the request
2 = Weak relevance, significant digressions or failure to address the core request
1 = Not relevant, does not address the user’s request or answers a different question entirely

5分 = 直接回应用户请求，完全紧扣主题，优先处理用户实际询问的内容
4分 = 大部分相关，存在轻微偏离主题或少量遗漏
3分 = 部分相关，涉及大致主题但未覆盖请求的关键部分
2分 = 相关性较弱，存在明显偏离主题或未回应用户核心请求
1分 = 完全不相关，未回应用户请求或答非所问

Workflow

工作流程

Compare the assistant response to the user’s request (if provided).
Score relevance on a 1-5 integer scale using the rubric only.
Write concise rationale tied directly to rubric criteria.
Produce actionable suggestions that improve relevance.

将助手回复与用户请求（若提供）进行对比。
仅依据上述评分标准，以1-5的整数为相关性打分。
撰写与评分标准直接挂钩的简洁理由。
提出可提升相关性的可行建议。

Output Contract

输出规范

Return JSON only. Do not include markdown, backticks, prose, or extra keys.

Use exactly this schema:

{ "dimension": "relevance", "score": 1, "rationale": "...", "improvement_suggestions": [ "..." ] }

仅返回JSON格式内容。请勿包含Markdown、反引号、散文或额外的键值对。

请严格使用以下 schema：

{ "dimension": "relevance", "score": 1, "rationale": "...", "improvement_suggestions": [ "..." ] }

Hard Rules

硬性规则

```
dimension
```
must always equal
```
"relevance"
```
.
```
score
```
must be an integer from 1 to 5.
```
rationale
```
must be concise (max 3 sentences).
Do not include step-by-step reasoning.
```
improvement_suggestions
```
must be a non-empty array of concrete edits.
Never output text outside the JSON object.

```
dimension
```
必须始终等于
```
"relevance"
```
。
```
score
```
必须是1到5之间的整数。
```
rationale
```
必须简洁（最多3句话）。
请勿包含分步推理内容。
```
improvement_suggestions
```
必须是包含具体修改建议的非空数组。
绝不在JSON对象之外输出任何文本。