A web table extraction method based on structure and ontology

Abstract

The table extraction is an important issue of Webpage information analysis. At present, there are three mainly methods, which is how to construct the wrapper, how to construct the ontology and directly analysis the structure of a table on the webpage. In the process of analysis, usually these methods are applied independently. Aiming at the shortcomings of single method, this paper presents a synthetic method based on the ontology and structure. In this paper, we firstly locates the tables based on heuristic rules, and then analysis the table structure according to the label and the title ontology, at last extract and save the table data on the basis of the obtained characteristics. The experiments show that the introduction of the ontology greatly improved the accuracy of table structure recognition, and the precision and recall of the methods are better.