ABSTRACT
Computational treatments of humor attract attention from lay audiences practically every day. Computational humor roots are positioned within an intersection of artificial intelligence and humor studies, starting with papers on a restricted set of humor, mostly puns, in the early 1990s. A computer had to generate a joke that would be appreciated by a human, or a human had to generate a joke that would be recognized by a computer as non-serious information. If a joke was to be understood, a computer would need to find the scripts, parts of the joke that result in incongruity, and then detect logic that could lead to partial resolution. The clustering approach is interesting as it could show computational usefulness of a humor theory, namely General Theory of Verbal Humor (GTVH), for labeling training sets as well as tagging data in the machine-learning endeavors. Templated humors are becoming more sophisticated from the natural language processing perspective.
