Stemming is a widely accepted practice in Document Information Retrieval Systems (DIRs), because it is more benefical than harmful [3]as well as having the virtue of improving retrieval effciency by reducing the size of the term index. We will present a technique of semi-automatic stemming that is fine designed for JAVA environment. The method works without deep knowledge of grammar rules of a language in contradistinction to well-known Porter’s algorithm [8]. From that point of view, we can call our method universal for more languages. We will also present tests to show quality of the method and its error- rate.