Approximate joke similarity

Some jokes are funnier than others. How to algorithmically recognize the appropriate degree of funniness of jokes is not clear. We propose to establish a funniness level of a new joke based on a set of central jokes that contain jokes that are somewhat similar to the new one. Each joke of the central set can be associated with some funniness value. Funniness of a new joke can then be computed based on its similarity value with a known joke and a funniness value of a known joke. The similarity value of two jokes can be established from one of humor theories.