Web Information Extraction Based on Hierarchical Model

A hierarchical extraction model based on hidden Markov model is proposed after analyzing some existing algorithms used in the field of Web information extraction. We firstly annotate atom information items and compound information items in HTML documents and then use a bottomup clustering method to build a DOM+ tree. At last, we make use of the annotated information of atom information items and compound information items with the compound information items´ paths in DOM+ tree to build the hierarchical extraction model. Experiments show that we may get better performance by using hierarchical extraction model.