Automatic Search-and-Replace From Examples With Coevolutionary Genetic Programming

We describe the design and implementation of a system for executing search-and-replace text processing tasks automatically, based only on examples desired behavior. The consist pairs describing original string modified string. Their construction, thus, does not require any specific technical skill. constructs solution to specified task that can be used unchanged popular existing software processing. consists search pattern coupled with replacement expression: former is regular expression which describes both strings replaced their portions reused in latter, how build strings. Our proposed internally genetic programming implements form cooperative coevolution two separate populations are evolved independently, one patterns other expressions. assess our proposal six realistic complexity obtaining very good results, terms absolute quality solutions respect challenging baselines considered.