golden-dataset-validation

majiayu000 的头像作者 majiayu000

验证规则、模式检查、重复检测和黄金数据集完整性覆盖率分析

📖 中文导读:验证规则、模式检查、重复检测和黄金数据集完整性覆盖率分析 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: golden-dataset-validation description: Use when validating golden dataset quality. Runs schema checks, duplicate detection, and coverage analysis to ensure dataset integrity for AI evaluation. context: fork agent: data-pipeline-engineer version: 1.0.0 author: SkillForge AI Agent Hub tags: [golden-dataset, validation, integrity, schema, duplicate-detection, 2025] allowed-tools: - Read - Grep - Glob user-invocable: false

Golden Dataset Validation

Ensure data integrity, prevent duplicates, and maintain quality standards

Overview

This skill provides comprehensive validation patterns for the golden dataset, ensuring every entry meets quality standards before inclusion.

When to use this skill:

  • Validating new documents before adding
  • Running integrity checks on existing dataset
  • Detecting duplicate or similar content
  • Analyzing coverage gaps
  • Pre-commit validation hooks

Schema Validation

Document Schema (v2.0)

{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "type": "object",
  "required": ["id", "title", "source_url", "content_type", "sections"],
  "properties": {
    "id": {
      "type": "string",
      "pattern": "^[a-z0-9-]+$",
      "description": "Unique kebab-case identifier"
    },
    "title": {
      "type": "string",
      "minLength": 10,
      "maxLength": 200
    },
    "source_url": {
      "type": "string",
      "format": "uri",
      "description": "Canonical source URL (NOT placeholder)"
    },
    "content_type": {
      "type": "string",
      "enum": ["article", "tutorial", "research_paper", "documentation", "video_transcript", "code_repository"]
    },
    "bucket": {
      "type": "string",
      "enum": ["short", "long"]
    },
    "tags": {
      "type": "array",
      "items": {"type": "string"},
      "minItems": 2,
      "maxItems": 10
    },
    "sections": {
      "type": "array",
      "minItems": 1,
      "items": {
        "type": "object",
        "required": ["id", "title", "content"],
        "properties": {
          "id": {"type": "string", "pattern": "^[a-z0-9-/]+$"},
          "title": {"type": "string"},
          "content": {"type": "string", "minLength": 50},
          "granularity": {"enum": ["coarse", "fine", "summary"]}
        }
      }
    }
  }
}

Query Schema

{
  "type": "object",
  "required": ["id", "query", "difficulty", "expected_chunks", "min_score"],
  "properties": {
    "id": {"type": "string", "pattern": "^q-[a-z0-9-]+$"},
    "query": {"type": "string", "minLength": 5, "maxLength": 500},
    "modes": {"type": "array", "items": {"enum": ["semantic", "keyword", "hybrid"]}},
    "category": {"enum": ["specific", "broad", "negative", "edge", "coarse-to-fine"]},
    "difficulty": {"enum": ["trivial", "easy", "medium", "hard", "adversarial"]},
    "expected_chunks": {"type": "array", "items": {"type": "string"}, "minItems": 1},
    "min_score": {"type": "number", "minimum": 0, "maximum": 1}
  }
}

Validation Rules Summary

Rule Purpose Severity
No Placeholder URLs Ensure real canonical URLs Error
Unique Identifiers No duplicate doc/query/section IDs Error
Referential Integrity Query chunks reference valid sections Error
Content Quality Title/content length, tag count Warning
Difficulty Distribution Balanced query difficulty levels Warning

Quick Reference

Duplicate Detection Thresholds

Similarity Action
>= 0.90 Block - Content too similar
>= 0.85 Warn - High similarity detected
>= 0.80 Note - Similar content exists
< 0.80 Allow - Sufficiently unique

Coverage Requirements

Metric Minimum
Tutorials >= 15% of documents
Research papers >= 5% of documents
Domain coverage >= 5 docs per expected domain
Hard queries >= 10% of queries
Adversarial queries >= 5% of queries

Difficulty Distribution Requirements

Level Minimum Count
trivial 3
easy 3
medium 5
hard 3

References

For detailed implementation patterns, see:

  • references/validation-rules.md - URL validation, ID uniqueness, referential integrity, content quality, and duplicate detection code
  • references/quality-metrics.md - Coverage analysis, pre-addition validation workflow, full dataset validation, and CLI/hook integration

  • golden-dataset-curation - Quality criteria and workflows
  • golden-dataset-management - Backup/restore operations
  • pgvector-search - Embedding-based duplicate detection

Version: 1.0.0 (December 2025) Issue: #599

Capability Details

schema-validation

Keywords: schema, validation, schema check, format validation Solves:

  • Validate entries against document schema
  • Check required fields are present
  • Verify data types and constraints

duplicate-detection

Keywords: duplicate, detection, deduplication, similarity check Solves:

  • Detect duplicate or near-duplicate entries
  • Use semantic similarity for fuzzy matching
  • Prevent redundant entries in dataset

referential-integrity

Keywords: referential, integrity, foreign key, relationship Solves:

  • Verify relationships between documents and queries
  • Check source URL mappings are valid
  • Ensure cross-references are consistent

coverage-analysis

Keywords: coverage, analysis, distribution, completeness Solves:

  • Analyze dataset coverage across domains
  • Identify gaps in difficulty distribution
  • Report coverage metrics and recommendations