Application of neural networks and Kano’s method to content recommendation in web personalization

Abstract

As customers become more skilled in the use of internet, many companies have gradually established their websites with more and more enormous information to get future competition in electronic commerce (EC). However, the miscellaneous information often brings the users at a loss. Web personalization provides a solution to improvement of information overloading on websites. The objective of web personalization is to give users a website they want or need, and thus knowing the needs of users is an important task for content recommendation in web personalization. In this article, we propose a hybrid approach for this task. The proposed approach trains the artificial neural networks to group users into different clusters, and applies the well-established Kano’s method to extracting the implicit needs from users in different clusters. Finally, a real case of tour and travel websites applying the approach is presented to demonstrate the improvement of information overloading.

Introduction

The continuous growth in the size and the use of the World Wide Web imposes new methods of design and development of online information service for most potential market. In order to gain the strategic advantage of further competition in electronic commerce (EC) on the web, many companies have established their websites as a business frontier. However, most of these websites are loaded with a large amount of information about the company, product, or service, and further support transaction functions for online trading to serve users. The overloading information would make users feel lost and frustrated when they surf on the websites. In general, users prefer and are more comfortable with websites that present the right content in ways that their linings with preference (Aragonees & Hart-Davidson, 2002). Consequently, the content provided on the website, the layout of the individual webpage, and the structure of the entire website affect the utility of a website in providing the intended service to its users (Mulvenna, Anand, & Buchner, 2000).

One way to solve the problem of information overloading is to provide a personalized website for each user. Web personalization is defined as any action that adapts the information or service provided by a website to the needs of a particular user or a set of users, using the knowledge gained from the users’ surfing behavior and interests in combination with the content and the structure of the website (Anand and Mobasher, 2007, Eirinaki and Vazirgiannis, 2003, Kim and Cho, 2007). The steps of website personalization process starts with user data collection which can be achieved using different approaches. The user profiles, interests, desires, and needs are captured and processed through direct and indirect engagements (Braynov, 2003). Some applications directly involve user data through surveys, questionnaires, submitting personal information during registration, and so on. In this case, the type of content may be provided for users according to their choices and preferences. Some other applications, building user profiles in accordance with log files, are engaged without the user direct involvement (Liu & Kešelj, 2007). In this case, recommendations of content are based on statistics of page access frequency or traversal path of click on a website.

Once collected, the user profiles and/or interests would be categorized by some classification or association rules, and analyzed to infer the needs and preferences of user with techniques like the content-based filtering (Herlocker et al., 1999, Yue, 1999), collaborative filtering (Goldberg et al., 1992, Resnick et al., 1994, Schafer et al., 2007), and rule-based filtering (Allen, 2000, Dean, 1998, Kim et al., 2006). Content-based filtering is based on individual user preferences. It tracks each user’s behavior and individually recommends the contents the users liked to them. Collaborative filtering is based on comparing a user’s taste with those of the users in order to build up groups of like-minded users. In rule-based filtering the users are asked to answer a set of questions derived from a decision in order, to look for a pattern (e.g. list of products) relative to webpage visit.

Analysis results in user needs and preferences are contributive to determination of the actions that should be performed. However, it is not easy in practical development of web personalization applications. The main difficulty results from the excessive server loading due to the large number of client request/server response pairs (Sedayao, 1995) and a large number of user data that must be stored on the server. Additionally, the server loading enlarges with a high cost as the number of users increases.

In this paper, an approach integrating the artificial neural network (ANN) with Kano’s method (namely Kano–ANN approach) is proposed to create a content recommendation tool for development of web personalization. The Kano–ANN approach uses the adaptive resonance theory (ART) network as the parallel distributed computing platform for unsupervised cluster discovery in user needs and preferences. In addition, the Kano–ANN approach emerges from the concept of collaborative filtering technique, and consequently the data loading would be moderate for servers. By the fusion of Kano’s method (Kano, Seraku, Takahashi, & Tsuji, 1984) that is a well-established psychology-based customer satisfaction methodology, this approach paves the way to seize the preferences and needs of a particular user or a set of users, and categorize them into three groups: the attractive, must-be, and one-dimensional in light of how well they are able to satisfy users. The remaining part of this paper is organized as follows. In next section, the concepts of ART and Kano’s method are reviewed. The Kano–ANN approach is then described in Section 3. A case study on a tour and traveling websites is implemented by proposed approach in Section 4. Finally, concluding remarks are given in Section 5.