Blueprint

Introduction

Built in partnership with NVIDIA and Microsoft, Blueprint accelerates data labeling, helping to find clusters of unidentified, inconsistently, and confusingly named tags. This is the first step in contextualizing data and creating an AI/ML-ready data format.

• Create order from the chaos of unstructured data by leveraging multiple AI/ML/NLP techniques, and automating tag-to-asset mapping.  
• Automate and accelerate the building of data dictionaries, particularly when users have more than 1,000 tags to map.  

  
  

Quick Start

To start your tag-to-asset mapping, access the developer page and select TagBERT. Name your mapping and determine what data source you would like to use as a sample for the  process.  You can either select a topic or upload a sample dataset directly.  The dataset must contain at least 1000 records and be organized with time as rows/key and tags as columns/value.

The mapping process will automatically start running with the default parameters after the validation of the dataset is performed. You can change any of the parameters and re-run as needed.  You will be notified via email when the process is complete.

Once complete, you can view the results or download the output file containing the mapping of tags to clusters.