SuperAPI AI Directory

data-semantic

Data & analytics

View repository

Updated 10-15 23:48

About this skill

Feature Overview
Operation Description Key Parameters
list Query Logical View List keyword, datasource_id
query Query Semantic Understanding Results form_view_id
understand Trigger Form Understanding form_view_id or datasource_id
batch Batch Understanding (≤100) form_view_ids
match Batch Object Matching kn_id, ot_id, entries
Input Parameters
Parameter Required Description
operation ✅ list/query/understand/batch/match
auth_token ✅ JWT Token
form_view_id query/understand Single View ID
datasource_id Data Source Understanding Data Source UUID
keyword Optional Keyword Search
Operation Examples

1. Query View List

operation: list
keyword: 用户

2. Query Semantic Results (output full report)

operation: query
form_view_id: <uuid>

Output: Field semantics table + Business object table + Attribute table

3. Single View Understanding (output full report)

operation: understand
form_view_id: <uuid>

Output: Field semantics table + Business object table + Attribute table

4. Batch Understanding (statistics report only)

operation: batch
form_view_ids: [<uuid1>, <uuid2>, ...]

Output: Statistics report only, no detailed semantics output

5. Data Source Batch Understanding

operation: understand
datasource_id: <uuid>

≤50 views execute directly, >50 call Python script

6. Batch Object Matching

operation: match
kn_id: <配置值>
ot_id: <配置值>
entries: [{"name": "客户信息"}]
State Machine
Status Code Status Name Processing Action
0 Not Understood Trigger generation → Poll and wait → Submit confirmation after status 2 → Regenerate → Poll → Submit confirmation → Complete
1 Understanding Poll and wait → Status 2/3/4 → Same as status 2 processing
2 Pending Confirmation Submit confirmation → Regenerate → Poll and wait → Submit confirmation after status 2 → Complete
3 Completed Trigger regeneration → Poll and wait → Submit confirmation after status 2 → Complete
4 Pending Confirmation (Re-understanding) Trigger regeneration → Poll and wait → Submit confirmation after status 2 → Complete
5 Understanding Failed Output failure reason, terminate

⚠️ During batch understanding, regardless of whether the current status is Completed (3) or Pending Confirmation (4), re-understanding will be triggered

Large Data Volume Processing

Protection Mechanisms

  • ≤50 → Execute directly
  • >50 → Recommend using Python script
  • >1000 → Remind to execute during non-working hours
  • >5000 → Recommend contacting the technical team

Python Script

# 数据源批量理解
python scripts/data_semantic_batch.py --token <JWT> --datasource-id <UUID>

# 批量视图理解
python scripts/data_semantic_batch.py --token <JWT> --view-ids <id1,id2>

# 断点续传
python scripts/data_semantic_batch.py --token <JWT> --resume
Output Format

Single View Output (including business object table)

📊 Data Semantic Understanding Report

Item Content
View Technical Name <tech_name>
View Business Name <biz_name>
Understanding Status <status_name>

📈 Recognition Statistics

Category Statistics
Field Semantics Total fields: <total>, Completed: <completed>
Business Objects Object: <count> items, Attributes: <attr_count> items

📋 Field Semantic Completion

Completed Fields

Technical Name Business Name Field Role Field Description
user_id User ID Business Primary Key Unique User Identifier
user_name User Name Business Feature User Name
create_time Creation Time Time Field Record Creation Time

Incomplete Fields

Technical Name Field Type Field Description
ext_field1 VARCHAR Extension Field

🏢 Business Object Recognition Results

1️⃣ <Business Object Name>

Attribute Name Field Technical Name Field Business Name Field Role Field Description
Customer ID customer_id Customer Number Business Primary Key Unique Customer Identifier
Customer Name customer_name Customer Name Business Feature Customer Name

Batch Output (Statistics Report Only)

🧠 Batch Understanding Report

Item Content
Total Views <total_count>
Success <success_count>
Failure <fail_count>

Batch understanding outputs only a statistics report; for details, please query separately

Notes
  1. Token Required - All operations require a valid JWT
  2. Encoding Issue - API responses may be GBK and need to be decoded with latin1
  3. Chinese Garbled Text - match operation uses echo | curl -d @-
  4. Data Discrepancy - Field semantics and business objects come from different tables