A hybrid method for Web data extraction

Abstract

Web data extraction refers to the technology that helps people find wanted information from the Web. We first classify existing data extraction algorithms into two classes: top-down and bottom-up, and then analyze their strengths and weaknesses in terms of extraction accuracy. On the basis of this analysis, we present a hybrid algorithm: bi-direction data extraction (BiDDE for short), which takes the full strengths of both top-down and bottom-up algorithms and yet avoid their weaknesses. The experimental results show that BiDDE has not only higher accuracy than top-down algorithm and bottom-up algorithm, but satisfactory performance.