Asset inventory and metadata completion allow you to trigger automated governance checks on Catalog / Schema through natural language descriptions. Buddy automatically scans metadata gaps, generates completion suggestions, and supports natural language semantic data discovery, waiting for your confirmation at key checkpoints before implementation.
Overview
Asset inventory is the starting point of data governance, helping you quickly understand the metadata quality and usage status of your data assets. This feature covers the following scenarios:
Metadata Completeness Scan: Identify assets with missing table descriptions, missing field descriptions, or missing owners, and sort them by governance gap severity.
Metadata Auto-completion: Based on field names and contextual semantics, AI automatically generates suggestions for table descriptions and field descriptions. Users can view the batch write results in a dialog box and save them with one-click confirmation.
Natural Language Data Discovery: Users describe their data requirements in natural language, and AI performs semantic matching based on Catalog metadata (table names, field names, table descriptions, Tags, and so on) to return the most relevant table list. The discovery results include "Usage Frequency" and "Recent Activity" information, sourced from query logs and Catalog statistics.
Buddy automatically scans assets within the target scope, generates inventory reports, and waits for your confirmation at key checkpoints before implementation. A single inventory scan has no limit on the number of tables, and the system automatically handles schemas with a large number of tables.
Prerequisites
Buddy has been enabled for the workspace.
The current account has Buddy development mode permissions.
You have view permissions on the Catalog / Schema to be inventoried.
To perform metadata writes (such as completing descriptions or setting owners), you need the metadata edit permission on the corresponding Catalog.
Information You Must Provide
Before performing an asset inventory, Buddy will ask you to provide the following minimum information set, while the rest is automatically analyzed by AI:
Information Item | Description |
Inventory scope | Target Catalog, Schema, or specific table name |
Inventory purpose | Check metadata integrity / Complete descriptions / Find data (can be implicitly expressed through natural language) |
If your input is incomplete, Buddy will ask for the missing information through an "Ask User card" before continuing. If your input is complete, Buddy will execute directly and display the results.
Content That AI Can Intelligently Generate
Content Item | AI generation method |
Metadata Gap Report | Scan descriptions, owners, and tags of all tables and fields, and sort them by missing severity. |
Table Description Suggestion | Infer business meanings based on table names and field names, and generate descriptions in both Chinese and English. |
Field Description Suggestion | Analyze field name patterns (such as create_time and user_id) and data types, and generate semantic descriptions. |
Owner Recommendation | Recommend Owner candidates based on table creators, recent modifiers, and usage frequency. |
Semantic Data Search Result | Perform semantic matching in Catalog metadata based on natural language descriptions, and sort results by usage frequency and activity. |
Operation Steps
Step 1: Expressing Inventory Requirements
Describe your requirements in natural language in the Buddy input box, and Buddy will automatically identify the inventory type and target scope.
Example Input:
Check which tables in the ods_catalog.dw_order Schema are missing descriptions.
Automatically complete descriptions for all fields in the dwd_order_detail table.
Is there a table that records user product browsing behavior? Preferably one that has been used recently.
Please help me find the table that stores order refund information.
Which tables in this Schema do not have an Owner?
Step 2: Completing Missing Information
If the input is incomplete, Buddy will ask you for the missing information through an Ask User card:
If no Catalog / Schema is specified, Buddy will list the available Catalogs in the current Workspace for you to choose from.
If the completion scope is unclear (single table or entire Schema), you will be asked to confirm.
After you answer, Buddy will continue the execution.
Step 3: Reviewing AI-Generated Results
Depending on the scenario, Buddy displays results in different formats:
Metadata Inventory Scenario: Displays scan results in a table, including:
Table name, missing item type (table description, field description / Owner), number of missing items, severity rating
Overall missing rate statistics and ranking
Metadata Completion Scenario: Displays AI-generated description suggestions in a dialog box, including:
Original table and field names
AI-suggested description text
Confidence indicator
Natural Language Data Discovery Scenario: Displays matching results in a list, including:
Table name and its Catalog / Schema path
Table description and core field summary
Semantic match score
Recent usage frequency and activity information (sourced from query logs and Catalog statistics)
Note:
You can modify the AI-generated description in the completion suggestion before confirming the write operation.
Step 4: Confirming and Applying Changes
Metadata Completion: View the batch write results in a dialog box. After one-click confirmation, Buddy calls the DataBuddy API to write descriptions in batches and displays the write results, including the number of successes and failures.
Owner Setting: After confirming the Owner, Buddy updates the table's Owner information in batches.
Data Discovery Results: You can directly continue to operate on the found tables, for example, "Help me configure quality rules for this table" or "View the field details of this table".
Key Milestone Confirmation
Confirmation Item | Confirmation Timing | Description |
Inventory scope | Before Scanning | Catalogs / Schemas / Tables to scan. |
Content Completion | Before Write | AI-generated description suggestions, modifiable one by one |
Owner Candidate | Before Setting | AI-recommended Owner, replaceable |
Batch Operations | Before Batch Write | A dialog box displays the batch completion results, and they are written after one-click confirmation. |
FAQs
Q: Will Metadata Completion Overwrite Existing Descriptions?
A: No. By default, Buddy only fills in missing description fields, and existing descriptions will not be overwritten. If you need to regenerate an existing description, you can explicitly instruct it to "regenerate the description of table xx".
Q: What Is the Matching Scope of Natural Language Data Query?
A: Buddy performs semantic search across all Catalogs you have permission to access in the current Workspace. The search results are semantically matched based on Catalog metadata (table names, field names, table descriptions, Tags, and so on) and comprehensively ranked by combining usage frequency and activity, which are sourced from query logs and Catalog statistics.
Q: How Accurate Are AI-Generated Field Descriptions?
A: AI primarily infers based on field name patterns (for example, create_time → "Creation Time") and contextual semantics, achieving higher accuracy for fields with standardized naming. For business-specific abbreviations or custom naming, it is recommended that you review and modify them as needed.
Q: How to Confirm During Batch Completion?
A: Buddy displays all description suggestions to be written in a dialog box. You can review and modify each one, and then confirm the batch write with one click.
References