What technology is used to perform redaction and what languages are supported?
NER (Named Entity Recognition) detects entities in the transcript that contain sensitive information. NER systems use machine learning and statistical techniques to automatically analyze and tag named entities. They are trained on large, annotated datasets where human annotators manually label named entities in text. In addition to NER, Genesys Cloud also looks at the context of surrounding words and phrases to flag sensitive information; for example, when the user says, ‘My account number is xxx’ or ‘My password is xxx.’
Currently, this technology only supports English dialects.