Knowledge Base
Rosetta Stone Normalized Datasets
Overview
The Normalized Datasets page provides visibility into how Rosetta Stone has mapped and normalized your data. This interface surfaces confidence scores for AI-generated mappings, helping you understand the quality of your data normalization at a glance.
Location: Rosetta Stone → Normalized Datasets
Understanding the Normalized Datasets Table
The main table displays all datasets that have been processed through Rosetta Stone normalization.
Table Columns
| Column | Description |
|---|---|
| Mapping Type | Icons indicating the type of normalization applied (AI-generated, manual, column-level, row-level) |
| Dataset Name | The name of the normalized dataset. Click to view detailed mapping information. |
| Columns | Number of columns in the dataset |
| Rows | Number of rows in the dataset |
| Confidence | Overall confidence score for the dataset's mappings, color-coded by quality |
| Created | When the normalization was created |
Confidence Score Indicators
Confidence scores are color-coded to help you quickly identify datasets that need attention:
- Green (90%+): High confidence. These mappings are ready to use. Rosetta Stone is highly confident the transformations are correct.
- Yellow (70-89%): Medium confidence. Review recommended. These mappings may benefit from validation.
- Red (Below 70%): Low confidence. These mappings may require review.
Filtering and Searching
Search
Use the search field to find specific datasets by name.
Filter
Use the filter dropdown to narrow results by:
- Confidence level (high, medium, low)
- Mapping type
- Date range
Data Plane
The data plane selector in the header filters datasets to show only those in your current data plane context.
Sorting
Click any column header to sort the table:
- Columns: Sort by number of columns
- Rows: Sort by number of rows
- Confidence: Sort by confidence percentage
- Created: Sort by creation date
Reviewing Mapping Suggestions
When you initiate normalization or use "Ask Rosetta" on a dataset, you can review AI-suggested mappings grouped by confidence level.
High Confidence Mappings
- Displayed with a green indicator
- Can be bulk-accepted using "Accept all high confidence"
- Ready to use without review
Medium Confidence Mappings
- Displayed with a yellow indicator
- Review individually recommended
- Each mapping shows:
- The Rosetta Stone attribute being mapped
- Source columns from your dataset
- Individual confidence percentage
- Accept/Reject buttons
Low Confidence Mappings
- Displayed with a red indicator
- Require user input
- Expand to see details and provide feedback
Providing Feedback
When rejecting a mapping, you can provide feedback to help train Rosetta Stone:
- Click Reject on the mapping
- Enter the reason the mapping is incorrect in the feedback field
- Click Train Rosetta to submit your feedback
This feedback helps improve future mapping accuracy for similar datasets.
Best Practices
- Monitor confidence scores: Use this page to identify datasets with lower confidence that may need attention.
- Start with high-confidence mappings: When reviewing suggestions, accept bulk high-confidence mappings first to save time.
- Review medium-confidence mappings: These often need minor adjustments or validation.
- Provide feedback on incorrect mappings: Your feedback trains the AI to make better suggestions.
- Use confidence as a guide: Low confidence doesn't always mean incorrect—it indicates areas where Rosetta Stone has less certainty about the mappings.