Confirming / Training Classifications
Transcend leverages machine learning techniques to quickly determine exactly where each and every personal datapoint lives within your individual data silo. With Structured Discovery, Transcend eliminates the need to derive queries for an internal database and maintain them through inevitable database schema changes.
Confirming and correcting guesses in the train tab provides us with data to help tune the next version of our classifier. If you notice many incorrect guesses, please reach out to your Transcned account team to discuss.
Note: If you would like to try the newest classifier leveraging a Large Language Model (LLM), please check with your Transcend account manager.
From Structured Discovery, navigate to a specific data silo, then click on the “Train” tab. Answer our prompted question, Is the NAME datapoint a CATEGORY|SUBCATEGORY?” by either:
- Confirming by clicking on the button "Confirm Category" or pressing c on the keyboard.
- Selecting a different category from the dropdown, and then confirming it
- Skipping the category by clicking on the button "Skip Category" or pressing s on the keyboard.
![Train Transcend on classifying your personal data through a series of questions](/_next/image?url=%2Fdocs%2Fscreenshot%2F2024-04-16-train.png&w=3840&q=75)
As you answer these questions, Transcend will improve our classifier for the various data categories in your data silos.
You can also confirm classifications in bulk for data points with the same name as the one presented. You can see all instances of datapoints with the same name by clicking on the “View all instances of 'name'” button, which will open a new tab with all instances of datapoints with the same name. You can then confirm the category for all instances of the datapoint by clicking on the “Bulk Confirm” button.