Actually, what is appealing about this solution to me is that it can allow for changes in redaction mechanisms over time. With NLP becoming more powerful it is not unrealistic to think that might become palatable over (or in parallel with) more explicit mechanisms. This would also allow for the flexibility of determining which method(s) to apply at run time.
b
On Sep 23, 2015, at 11:00 PM, Steven Legg <> wrote:
> The redaction solution I sketched out was the result of looking for a faster way
> to do redaction. One request determines whether access to the document is permitted
> and simultaneously returns the instructions for redacting the document, as
> obligations, if access is permitted.
