This document presents an empiricist mentalist semantic theory that could be implemented computationally. It addresses three major objections to the original theory. The intersubjectivity objection is met with the proposal of webs of similar ideas within subjects. The self-applicability objection is addressed by conceiving of ideas as self-applying like coin slots. The abstract ideas objection is dealt with by suggesting ideas are pattern-like rather than image-like, analogous to regular expressions. Further research is needed to develop these proposals.