Skip to main content
In this tutorial, you’ll use Rosetta Stone to normalize your data. Normalization maps your unique schema to Narrative’s standard format, enabling queries across datasets from different sources.
This tutorial is part of the First Steps series. Complete Upload Your Data before starting.

​
Prerequisites

  • A Narrative I/O account
  • At least one uploaded dataset (from the previous tutorial)

​
What you’ll learn

  • How to view automatic mapping suggestions for your data
  • How to accept, modify, and reject suggested mappings
  • How to write a custom transformation expression
  • How to test and validate your mappings

​
What you’ll build

By the end of this tutorial, your dataset will be normalized to Rosetta Stone attributes. This means:
  • Your data is queryable through the standard narrative.rosetta_stone table
  • Other users can discover and access your data using standard attribute names
  • Your data can be combined with other normalized datasets

​
Steps

1

Open your dataset

Navigate to Datasets in the left sidebar and select the dataset you uploaded in the previous tutorial.[Screenshot placeholder: Dataset list view with a dataset selected]Click on the dataset name to open its detail view.
2

Navigate to Rosetta Stone mappings

In the dataset detail view, click the Rosetta Stone tab.[Screenshot placeholder: Dataset detail view with Rosetta Stone tab highlighted]This opens the mapping interface where you can see:
  • Your dataset’s columns on the left
  • Suggested Rosetta Stone attributes on the right
  • Confidence indicators for each suggestion
3

Review mapping suggestions

Narrative automatically analyzes your data and suggests mappings. Each suggestion shows:
IndicatorMeaning
Green checkmarkHigh confidence—likely correct
Yellow warningMedium confidence—review recommended
Red XLow confidence or no suggestion—manual mapping needed
[Screenshot placeholder: Mapping suggestions with different confidence levels]Click on any suggestion to see details about why the system made that recommendation.
4

Accept high-confidence mappings

For suggestions with a green checkmark, click Accept to confirm the mapping.[Screenshot placeholder: Accepting a mapping]For example, if you have a column named email_hash and the system suggests the email_sha256 attribute with high confidence, accepting creates a mapping that links them.
You can select multiple high-confidence mappings and accept them in bulk using the Accept All High Confidence button.
5

Customize a mapping with a transformation

Let’s customize a mapping to see how transformations work. Find a date column in your dataset—the system may have suggested a timestamp attribute.Click Edit on that mapping to open the transformation editor.[Screenshot placeholder: Transformation editor]If your dates are in a format like MM/DD/YYYY, you might need a transformation:
TO_TIMESTAMP(your_date_column, 'MM/DD/YYYY')
This converts your date format to the standard timestamp format.Common transformation patterns include:
Source formatTransformation
MM/DD/YYYYTO_TIMESTAMP(col, 'MM/DD/YYYY')
M or F for genderCASE col WHEN 'M' THEN 'male' WHEN 'F' THEN 'female' ELSE 'unknown' END
Uppercase stringsLOWER(col)
Click Save when you’re done editing.
6

Handle unmapped columns

Some columns may not have suggested mappings. You have two options:Map manually: Click Add Mapping and select the appropriate attribute from the list. Use the search to find attributes by name or description.Leave unmapped: Not every column needs to map to a Rosetta Stone attribute. Columns that are specific to your use case or don’t have corresponding attributes can remain unmapped.[Screenshot placeholder: Adding a manual mapping]
7

Test your mappings

Before activating, test your mappings to ensure they work correctly.Click Test Mappings to run a sample transformation.[Screenshot placeholder: Test results view]The test shows:
  • Sample rows from your data
  • The transformed output for each mapping
  • Any validation errors or warnings
Review the output to verify:
  • Values convert correctly (dates parse properly, enums map to valid values)
  • No unexpected nulls appear
  • Transformations produce the expected format
If you see validation errors, click on the error to see details. Common issues include:
  • Date format mismatches
  • Values that don’t match enum options
  • Type conversion failures (e.g., non-numeric strings when expecting integers)
8

Activate your mappings

Once testing passes, click Activate All to enable your mappings.[Screenshot placeholder: Activate confirmation]Your mappings are now live. Data from your dataset is normalized and available through Rosetta Stone.

​
Verify it worked

Run this query to see your normalized data: 
SELECT *
FROM narrative.rosetta_stone
WHERE _nio_source_dataset_id = 'YOUR_DATASET_ID'
LIMIT 10
Replace YOUR_DATASET_ID with your actual dataset ID (visible in the dataset detail view). You should see your data with standardized column names matching the attributes you mapped.

​
What you accomplished

  • Reviewed automatic mapping suggestions
  • Accepted high-confidence mappings
  • Created a custom transformation for a date field
  • Tested mappings before activation
  • Activated mappings to make your data available through Rosetta Stone

​
Troubleshooting

IssueSolution
No mapping suggestions appearEnsure your dataset has uploaded data, not just a schema
Transformation errorsCheck the Transformation Functions reference for correct syntax
Test shows unexpected nullsYour transformation may not handle all source values—add ELSE clauses to CASE statements

​
Next steps

Write Your First Query

Query your normalized data with NQL

How Rosetta Stone Works

Understand the normalization model in depth

Mapping Schemas Guide

Advanced mapping techniques and API usage

Edge Cases

Handle complex mapping scenarios