Extraction of Syntactical Patterns from Parsing Trees
Résumé
This paper presents experiments with a new algorithm designed to detect recurrent syntactical patterns in natural language texts. It first describes the pattern extraction algorithm, which is based on an edit model generalized to SOT (Stratified Ordered Trees). Then it focuses on experiments with French classical literature of the 18 th and 19 th centuries. It goes on to evaluate efficiency before providing some examples of recurrent patterns that are typical of an 18 th century author, Madame de Lafayette.