CPC H04L 63/0407 (2013.01) [H04L 9/3213 (2013.01)] | 18 Claims |
1. A method comprising:
receiving, by a computing system, text data containing customer personal information, wherein the customer personal information originates from a customer-facing service associated with an externally-hosted application that is external to the computing system;
detecting, by the computing system, the customer personal information in the text data received from the customer-facing service using multiple detection layers to detect different types of customer personal information having different structures, wherein detecting the customer personal information in the text data comprises:
detecting, by a first detection layer comprising a rule-based model, a first type of customer personal information that is structured with a deterministic pattern,
detecting, by a second detection layer comprising a heuristic rule-based model, a second type of customer personal information that is structured with a learned pattern, and
detecting, by a third detection layer comprising a machine learning-based model, a third type of customer personal information that is unstructured;
generating, by the computing system, tokenized data based on output of the multiple detection layers, wherein generating the tokenized data comprises replacing each instance of the customer personal information detected in the text data with a respective token; and
sending, by the computing system, a request including the tokenized data to the externally-hosted application.
|