Adaptable Algorithm for Designed Web Process Sequence Data Analysis
Hai Wang, Shouhong Wang · 2009
ABSTRACT A significant interest for Web process design is to discover the discrepancies between the users' transaction process sequences and the desired process sequence for the Web transaction process. Sequence data analysis has been an important approach to analyzing Web log data in the e-commerce field. There have been many methods for sequence data analysis; however, few existing methods can be applied to analyzing designed Web transaction process sequence data for improving Web process design. This paper proposes an adaptable sequence matching algorithm for analyzing designed Web process sequence data for discovering knowledge about the Web process design. An application of this algorithm to a case of online shopping cart abandonment analysis is presented. Keywords: Web design, Web process sequence data analysis, designed Web process sequences, shopping cart abandonment. (ProQuest: ... denotes formulae omitted.) 1. Introduction A Web process is a business process carried out on the World Wide Web. Traditionally, Web process design has been studied in the field of workflow analysis [Cardoso & Sheth 2003; Cardoso 2006]. Recent research has suggested that Web page design has significant impact on online consumers' attitude and behavior towards Web processes [Chatterjee 2008]. A right Web page design must meet specific needs of right consumers for right Web processes [Zhou et al. 2004; Shergill & Chen 2005; Singh et al. 2008]. Web log sequences (i.e., user click-stream sequences) represent users' Web access behaviors in carrying out Web processes. Massive Web log sequences can be used to discover the general patterns of these process sequences [Wen et al. 2007; Greco & Guzzo 2007]. These patterns are useful for us to understand the online consumers' behaviors as well as problems in Web process design. For instance, by analyzing online shopping Web log sequences of an online store, one might be able to understand more about why online shoppers abandon shopping carts so often, and how the design of the Web site and transaction processes can be improved to reduce shopping cart abandonment. This paper presents a Web log sequence analysis method for improving Web process design. In the literature, various data analysis methods have been proposed to analyze Web log data [Cadez et al. 2000; Ester et al. 2002; Jiang & Tuzhilin 2006; Mobasher et al.2002; Manavoglu et al. 2003; Yang & Padmanabhan 2005]. In general, these methods aim at discovering interesting patterns of sets of sequences [Fayyad et al. 1996; Pei et al. 2004]. Common approaches to analyzing sequence data include time series analysis (e.g., [LeBaron & Weigend 1998]), association rules induction (e.g., [Lee et al. 2003]), and sequences pattern discovery (e.g., [Dutta et al. 2007]). While many research reports have been emphasizing on the performance of algorithms, exploring applications of process sequence data analysis results directly to e-commerce is imperative [Wu et al. 2000]. In this paper, we present a new method of analyzing Web log sequence data for the diagnosis of Web process design. This algorithm can be used to reveal useful information and develop knowledge for improving Web process design. The remainder of this paper is organized as follows. First, we provide a brief overview and discussion of major methods used for analyzing sequence data. Next, we develop an adaptable process sequence matching algorithm of analysis of Web process sequence data. Then, we present an application of the proposed algorithm to a case of shopping cart abandonment. Finally, we conclude with a summary of the study. 2. Related Work: Methods of Process Sequence Data Analysis Web process design is a topic in the field of business process design, and has been extensively studied through workflow analysis and management [Ould 1995; Cardoso & Sheth 2003; Cardoso 2006; Rozinat & van der Aalst 2008; van der Aalst et al. …