A Sentence Generation Algorithm for Testing Grammars
Lixiao Zheng, Duanyi Wu · 2009
One simple approach to testing grammars is to derive a number of sentences from the grammar under test and to validate whether they comply with the intended language. Purdom has proposed a by-now folklore sentence generation algorithm which however suffers from the fact that the sentences generated are usually too few and some of them are long with complicated derivation structures. In this paper, we present an improved algorithm which incorporates a length control mechanism into the generation process to produce more and simpler sentences. Experiments demonstrate that besides the benefits for user validation and error location, our algorithm is also helpful in highlighting more errors in some cases.