A Transducer Model for Web Information Extraction

In recent years, many authors have paid attention to web information extractors. They usually build on an algorithm that interprets extraction rules that are inferred from examples. Several rule learning techniques are based on transducers, but none of them proposed a transducer generic model for web information extraction. In this paper, we propose a new transducer model that is specifically tailored to web information extraction. The model has proven quite flexible since we have adapted three techniques in the literature to infer state transitions, and the results prove that it can achieve high precision and recall rates.