Identity Recognition

From Glitchdata
Jump to navigation Jump to search

The ability to identify objects based on their relationships. Uses Deep Learning, Neural Networks to postulate identity based on inputs.

  • Acquire Identity markers as inputs
  • Acquire identity datasets like passport, drivers license, social profiles, phone numbers, address, jobs, friends etc..
  • Tokenise identity data
    • Use NLP, Tokenisation, Topic Modelling
    • NER for phone numbers, DoB, name, identity markers.
  • Graph address, jobs, friends
  • Scoring
    • Score identity for diversity in data (i.e. fraud)
    • Score diversity in address
    • Score diversity in jobs
    • Score diversity in friends
    • Flag for Identity fraud if exceeds threshold
  • Recommendation
    • Propose who this person actually is
    • Use Centrality in identity markers.
  • Add Recognition capabilities
    • Image Recognition
    • Voice Recognition