Robust pointing by XPath language: Authoring support and empirical evaluation

Abstract

In spite of the increasing prevalence of XPath language including its use with XSLT little attention has been paid to empirical study of robust pointing by XPath expressions. The goal of this study is to draw practical implications of the robustness of XPath expressions, taking account of the four kinds of real-life HTML pages. For each DOM node in the sample pages, we created three types of XPath expressions, and investigated to what extent those expressions were able to continue pointing at the same node in the modified pages during the observation period of four months. The types of XPath expressions we used for the investigation include not only absolute expressions that simply follow the hierarchy of a document tree from the root to the target element, but also relative expressions that point to a target element in relation to some stable anchor position. As anchor nodes for the relative addressing expressions, we used elements with an href attribute, because a URL assigned to an href value should not be change too often, and so an href reference should be relatively stable as a semantic descriptor In this paper we briefly introduce an XPath authoring support for an annotation editor, and explain the types of XPath expressions. An empirical evaluation of the XPath expressions is then presented. Finally, we discuss the advantages and limitations of the XPath expressions, taking account of the actual page modifications, and investigate possibilities for further improvement of the addressing method.