So if I'm understanding, raw text is fed into the geotext python library, which parses out geographical locations and adds to the count.

I'm very skeptical that there are 18 new jobs on the island of Jinja, and that instead the analysis is picking up the Jinja templating language

Edit: same thing for "cambridge" in the same location as Jinja Island. Obviously probably refers to Cambridge MA, where Tufts/Harvard/MIT are all nearby, so tons of companies will be there.