Skip to main content
Shovels verifies data accuracy through multiple external sources to ensure reliability.

​
Address Verification

For address standardization, we cross-reference details using four different address sources:
  • National Address Dataset from the US Census
  • Open Address dataset
  • Simple Maps
  • ESRI
This multi-source approach ensures:
  • Consistent formatting
  • Accurate geocoding
  • Reliable location identification

​
Contractor Verification

We match contractor information against:
  • Publicly available state license files
  • Business registration records
This ensures contractors in our database are:
  • Properly licensed
  • Legitimate operators in their respective fields

​
Data Labeling Process

For permit classification, we employ a rigorous annotation process:
  • Multiple independent annotators label each record
  • Manual review resolves divergent responses
  • Validation sample size is 1-5% of overall data
Our approach has achieved 98% accuracy in classifications, validated by construction industry experts.

​
Golden Dataset Methodology

Annotators independently solve tasks rather than validating model outputs. This:
  • Prevents annotator bias
  • Creates a “golden dataset” of correct answers
  • Enables benchmarking of new model outputs without fresh validation