GRAMMATICAL INFERENCE FOR EVEN LINEAR LANGUAGES BASED ON CONTROL SETS

It is shown that the grammatical inference problem for even linear languages is reduced to the problem for regular sets. It is first shown that there exists a fixed even linear grammar over an alphabet which generates any even linear language over the alphabet with a regular control set. Next, it is shown that for any linear language there exists a unique regular control set. From these, the problem of identifying an unknown even linear language is reduced to the problem of identifying an unknown regular control set for the fixed even linear grammar. An algorithm is given which reduces the inference problem for even linear languages to that for regular sets. With it, any algorithm for regular sets is available for inferring even linear languages. The correctness and time complexity of the algorithm are immediately obtained. 6 Refs.