Information Extraction Framework

Abstract

The literature provides many techniques to infer rules that can be used to configure web information extractors. Unfortunately, these techniques have been developed independently, which makes it very difficult to compare the results: there is not even a collection of datasets on which these techniques can be assessed. Furthermore, there is not a common infrastructure to implement these techniques, which makes implementing them costly. In this paper, we propose a framework that helps software engineers implement their techniques and compare the results. Having such a framework allows comparing techniques side by side and our experiments prove that it helps reduce development costs.