A Clustering-Based Shifting Technique To Align Data Units In Web Data Bases

Abstract

An increasing number of databases have become web available through HTML form based search interfaces. The data units return from the underlying database are repeatedly encoded into the result pages dynamically for human browsing. In this paper we present an automatic annotation approach that first line up the data units on a result page into different groups such that the data in the same group have the same semantic. Then for each group we annotate it from unlike aspects and increasing the different annotations to expect a final annotation label for it. Each basic annotator is used to create a label for the units within their group holistically and a probability model is accepted to conclude the most suitable label for each group. The rules for all aligned groups jointly form the annotation wrapper for the corresponding WDB which can be used to directly annotate the data recovered from the same WDB in reply to new queries without the need to execute the alignment and annotation phases again. As such annotation wrappers can execute annotation rapidly which is important for online applications.