"Veripol works by identifying words and phrases that linguistic studies have shown to be identifiers of possible lies, Ms Queralt said. For example, using many adjectives or refusing to describe a scene are considered hints that a complainant is lying."

The article is about a sentiment analysis application used by Spanish police, and how trusting it as a lie detector could be a bit of a stretch. It could be argued that sentiment analysis and related techniques are not 100% bonkers, but it seems to be unclear at the moment.

From my experience humans currently perform much better at sentiment analysis and related techniques than computers. Computers have the advantage of scale, but this seems like a field where training actual humans on similar techniques would create much better results.

In recent years, I have been conditioned so that every time I hear the words "Believe me" I know the persons is lying.