data-quality

Data Quality Management

Progressive Loading: Core Concepts → Quick Start → Detailed Guide

Core Capabilities

Capability	Description	API Endpoint
Quality Rules	Create, read, update, and delete quality rules	`/api/data-view/v1/explore-rule`
Logical Views	Query view lists and field information	`/api/data-view/v1/form-view`
Inspection Tickets	Create and track inspection tickets	`/api/task-center/v1/work-order`
Knowledge Network	Configure rules based on the knowledge network	`/api/ontology-manager/v1`

Skill Input Parameters

The skill accepts the following input parameters. The large model should pass them in the following format when invoking the skill:

{
  "query": "用户提问内容（必须）",
  "business_docs": ["业务知识文档列表（可选）"],
  "business_desc": "业务知识描述文本（可选）",
  "context": "其它可用上下文文本（可选）"
}

Input Parameter Description

Parameter Name	Type	Required	Description
`query`	string	Yes	User question content, describing the data quality management operation the user wants to perform
`business_docs`	array	No	List of business knowledge documents, business knowledge documents provided by the user when configuring quality rules
`business_desc`	string	No	Business knowledge description text, business knowledge text description provided by the user when configuring quality rules
`context`	string	No	Other available context text, necessary context basis provided by the user

Overview of Input Parameter Usage Scenarios

Scenario	query	context	business_desc	business_docs
Knowledge Network Quality Analysis	✅	✅	✅	✅
View Quality Query	✅	✅	✅	✅
Rule Configuration	✅	✅	✅	✅
Quality Inspection	✅	✅	-	-
Problem Diagnosis	✅	✅	-	-

Distinction of Input Parameter Responsibilities

Important: Each input parameter has a clear division of responsibilities and must not be mixed up:

Input Parameter	Responsibility	Description
`query`	Target Identification (First Priority)	Used to determine the knowledge network or object class to be analyzed
`context`	Target Identification (Second Priority)	When there is no clear target in the query, extract the knowledge network or object class information from the context
`business_desc`	Rule Configuration Basis	Only used as the source of business semantics when configuring quality rules
`business_docs`	Rule Configuration Basis	Only used as the source of business documents when configuring quality rules

Target Identification Priority:

1. 首先从 query 中提取知识网络/对象类信息
2. 如果 query 中没有明确目标，才从 context 中提取
3. business_desc 和 business_docs 永远不能用于确定分析目标

Usage Priority

query Always required, used to determine user intent and analysis target
context Used to provide necessary context basis, serving as a supplementary source of analysis targets when the query has no target
business_desc Only used to provide business semantics when configuring quality rules
business_docs Only used to provide business documents when configuring quality rules

Prerequisites

DATA_QUALITY_BASE_URL=https://10.4.134.26
DATA_QUALITY_AUTH_TOKEN=Bearer xxxxxx

Verification: GET {BASE_URL}/api/eacp/v1/user/get

Key Constraints

For detailed constraints, please refer to: Core Constraints Reference

Configuration First: Environment variables must be verified before use
Evidence-Based: Rule configuration must have a clear basis
Configuration Not Empty: When creating a rule, rule_config cannot be empty
Technical name: rule_config The SQL expression in must use the field technical name
No report does not terminate: When querying the quality report, if "probe report does not exist" is returned, uniformly interpret it as "no quality report yet" and continue to the confirmation step of "whether to configure rules and initiate detection"
Authorized can proceed directly: If the user has clearly expressed "continue processing/resolve the issue/initiate detection" in the current turn, then directly enter the rule configuration and quality detection process
ID semantics cannot be mixed: In the knowledge network object class, data_source.id is the unified view ID (used for mdl_id to query the logical view), not the ticket datasource_id
Ticket data source origin: datasource_id、datasource_name、datasource_type must come from the logical view
Success response compatibility: The rule creation success status is considered successful for both 200/201 ; after the ticket is created successfully, prioritize reading id, and be compatible with work_order_id
Score display format: Scores are displayed directly as numeric values, without the "/100" suffix, rounded to two decimal places
Quality report null value handling: When a dimension score is null, display it as "not configured" and do not include it in the overall score calculation
Unified detection strategy: When multiple views need detection, prioritize configuring rules for all views, then uniformly create one quality detection ticket
Business perspective analysis: Quality report analysis must combine business_desc and business_docs to interpret quality metrics from a business perspective

Document structure

data-quality/
├── SKILL.md                    # 本文件 - 主入口
├── CHANGELOG.md                # 版本历史
├── README.md                   # 文档指南
│
├── core/                       # 核心概念层 (L1)
│   └── core.md                 # 核心概念
│
├── guides/                     # 用户指南层 (L2)
│   ├── quickstart.md           # 快速开始指南
│   └── detailed-guide.md       # 详细工作流
│
├── reference/                  # 参考文档
│   ├── core-constraints.md     # 核心约束（共享引用）
│   ├── glossary.md             # 术语表
│   ├── quality-inspection-workflow.md  # 质量检测工作流
│   ├── knowledge-network-workflow.md   # 知识网络工作流
│   ├── batch-processing-guide.md       # 批量配置处理流程
│   ├── error-handling.md       # 错误处理指南
│   ├── quality-report-scoring.md      # 评分处理策略
│   ├── api-overview.md         # API 概览
│   ├── intent-recognition.md   # 意图识别
│   ├── pagination.md          # 分页规范
│   ├── api-path-validation.md  # API 路径验证
│   ├── api-usage-guide.md      # API 使用规范
│   └── api/                    # 详细 API 文档
│       ├── api_data_view.md
│       ├── api_task_center.md
│       ├── api_knowledge_network.md
│       ├── api_eacp.md
│       └── api_std.md
│
└── examples/                   # 代码示例
    └── basic-usage.md          # 基础用法示例

Progressive loading guide

用户请求分析
    │
    ├─ 简单查询（视图/字段） ──▶ 加载核心 + 快速开始
    │
    ├─ 规则配置 ──▶ 加载快速开始 + API 指南
    │
    ├─ 质量检测 ──▶ 加载详细指南 + 质量检测工作流
    │
    ├─ 知识网络 ──▶ 加载详细指南 + 知识网络工作流
    │
    ├─ 错误处理 ──▶ 加载错误处理指南
    │
    └─ 复杂/错误场景 ──▶ 加载全部 + 示例

Quick navigation

Document	Purpose
Core concepts	Core information and quick reference
Quick start	Detailed examples and common operations
Detailed guide	Complete workflow and advanced features
Quality detection workflow	Complete quality detection process
Knowledge network workflow	Knowledge network scenario handling
Batch configuration handling	Paged loading and serial processing
Core constraints	Shared constraints reference
Error handling	Complete error handling guide
Score handling	Score conversion and display
API overview	API reference
Glossary	Term descriptions
Example code	Code samples