An XML approach to semantically extract data from HTML tables

Abstract

Data intensive information is often published on the internet in the format of HTML tables. Extracting some of the information that is of users' interest from the internet, especially when large number of web pages need to be accessed, is time consuming. To automate the processes of information extraction, this paper proposes an XML way of semantically analyzing HTML tables for the data od interest. It firstly introduces a mini language in XML syntax for specifying ontologies that represent the data of interest. Then it defines algorithms that parse HTML tables to a specially defined type of XML trees. The XML trees are then compared with the ontologies to semantically analyze and locate the part of table or nested tables that have the interesting data. Finally, interesting data, once identified, is output as XML documents.