Semantic analysis belongs to the third category in "lies, damn lies, and statistics." I'm sure it's slightly better for Metacortex's real use case - which is a

Semantic analysis belongs to the third category in "lies, damn lies, and statistics." I'm sure it's slightly better for Metacortex's real use case - which is an employee surveillance system - because people won't use strong language, hyperbole, share their real opinions, or talk about certain topics as much in professional communication. But I still find the whole domain to be a fraud with just enough signal in the noise to trick people.