Title: An empirical approach to evaluating web application compliance across diverse client platform configurations

Authors: Cyntrica Eaton, Atif M. Memon

Addresses: Department of Computer Science, University of Maryland, 4115 A.V. Williams Building, College Park, MD 20742, USA. ' Department of Computer Science, University of Maryland, 4115 A.V. Williams Building, College Park, MD 20742, USA

Abstract: Web applications are the most widely used class of software today. Increased diversity of web-client platform configurations causes execution of web applications to vary unpredictably, creating a myriad of challenges for quality assurance during development. This paper presents a novel technique and an inductive model that leverages empirical data from fielded systems to evaluate web application correctness across multiple client configurations. The inductive model is based on HTML tags and represents how web applications are expected to execute in each client configuration based on the fielded systems observed. End-users and developers update this model by providing empirical data in the form of positive (correctly executing) and negative (incorrectly executing) instances of fielded web applications. The results of an empirical study show that the approach is useful and that popular web applications have serious client-configuration-specific flaws.

Keywords: web testing; inductive learning; cross-platform compatibility; HTML tags; web applications; web engineering; compliance; client configurations.

DOI: 10.1504/IJWET.2007.012055

International Journal of Web Engineering and Technology, 2007 Vol.3 No.3, pp.227 - 253

Available online: 16 Jan 2007 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article