Here's a working definition for you: the construction of interlinked models that are proven/disproven by repeatable independent experiments. Science is tentative and falsifiable. The normal day-to-day scientist usually deals with somewhat boring measurements and experiments inside of a paradigm. Every now and then the paradigm is shown to be broken and a new paradigm takes its place.

Usually to call something "not science" is to say it is not based on 1) testable, falsifiable hypothesis, 2) accurate data, or 3) an inability to replicate/verify