The distinction between absolute and relative terms was introduced by Peter Unger in his 1971 paper A Defense of Skepticism and differentiates between terms that, in their most literal sense, don't admit of degrees (absolute terms) and those that do (relative terms).[1] According to his account, the term "flat", for example, is an absolute term because a surface is either perfectly (or absolutely) flat or isn't flat at all. The terms "bumpy" or "curved", on the other hand, are relative terms because there is no such thing as "absolute bumpiness" or "absolute curvedness" (although in analytic geometry curvedness is quantified). A bumpy surface can always be made bumpier. A truly flat surface, however, can never be made flatter. Colloquially, he acknowledges, we do say things like "surface A is flatter than surface B", but this is just a shorter way of saying "surface A is closer to being flat than surface B". This paraphrasing, however, doesn't work for relative terms. Another important aspect of absolute terms, one that motivated this choice of terminology, is that they can always be modified by the term "absolutely". For example, it is quite natural to say "this surface is absolutely flat", but it would be very strange and barely even meaningful to say "this surface is absolutely bumpy".
Once the distinction is made, it becomes apparent that the application of absolute terms to describe the real-world objects is doubtful. Absolute terms describe properties that are ideal in a Platonic sense, but that are not present in any concrete, real-world object.
The distinction sets up the foundation for the final argument of the paper: that knowledge requires certainty and that, certainty being an absolute term, it follows that it can never be achieved in reality. It is a Platonic ideal that we can get closer and closer to, but never truly reach. In Unger's own words, "every human being knows, at best, hardly anything to be so".