A Personalized Recommendation Model in E Commerce Based on TOPSIS Algorithm

A Personalized Recommendation Model in E Commerce Based on TOPSIS Algorithm

Liang Wang (School of Economics and Management, Beijing Institute of Graphic Communication, Beijing, China & Institute of Information System, Beijing Jiaotong University, Haidian, Beijing, China), Runtong Zhang (Institute of Information System, Beijing Jiaotong University, Haidian, Beijing, China) and Huan Ruan (School of Economics and Management, Beijing Institute of Graphic Communication, Beijing, China)
Copyright: © 2014 |Pages: 12
DOI: 10.4018/jeco.2014040107


From the perspective of performance and universality, this paper analyzed the characteristics of typical technologies for personalized recommendation system, and then made a basic architecture for the improved model. With the architecture, this paper introduced a personalized recommendation model in e-commerce system. The model is based on an n-tiers structure and the TOPSIS algorithm, first standardize the user evaluation indexes, and then determine the indexes weights according to user's needs, and finally calculate the personalized recommendation results. This model can be applied to a variety of e-commerce applications, especially for the e-commerce application with structured or semi-structured products such as digital books, journals and other publications.
Article Preview

1. Introduction

As the rapid development of information technology and the Internet, e-commerce is quickly developing as a new and efficient business pattern. Now people can click the mouse in front of the computer to complete the complex behavior of commodity trading. E-commerce brings not only convenient, but also the issue of “information overload and redundancy” (Shardanand & Maes, 1995). Customers in a large number of commodity spaces cannot successfully find the products or services they need. So we are introducing a personalized recommendation model for e-commerce. With this model, website owners can collect customers' information by data mining and analyze customer behavior, finally predict the customers' preferences and make a personalized recommendation for them.

The first formal definition of recommendation system in e-commerce is given by Resnick and Varian in 1997: A recommender system in e-commerce provide users with product information and recommendations using e-commerce websites. The aim of recommender system is to help users make decisions of buying products, or simulate sales representatives to help customers complete shopping process. (Resnick & Varian, 1997) Now this definition has been widely accepted and used.

Most Recommended system is mainly commercially used in e-commerce and other web systems, especially in the fields of publication, movies, music and other media industry. This means that if the recommendation system wants to be successfully utilized in the business, it must provide the users a true and valuable recommendation function, and eventually make profits for the commercial enterprises. The e-commerce personalized recommendation system can be divided into three main modules (Yu, 2007) which have been listed below.

  • 1.

    Input Functional Module: Input functional module is mainly responsible for the collection and updating of user information. In accordance with time, information sources can be divided into current behavior inputs and historical behavior inputs. In accordance with the object, information sources can be divided into targeted customer inputs and community inputs. Targeted customer inputs mainly refer to the target users making purposive evaluation in order to get accurate results from the recommendation system, and the information reflect user’s personal behavior. Community inputs mean the information on behalf of a group of users. Targeted customer inputs include user registration information import, implicit browse input, explicit browse input, keyword input and purchase history input. Community inputs involve project property input, evaluation grade input and evaluation text input.

  • 2.

    Recommendation Approach Module: Recommendation approach module is the core of the entire recommendation system. It directly determines the system capacity and performance. In this module, technical architecture and algorithm are foundations. Details will be introduced in part 3.

  • 3.

    Output Functional Module: Output functional module outputs the recommendation results to the corresponding user. The output forms of recommendation system in e-commerce mainly include related products output, total and average evaluation grade output, evaluation text output, e-mail output and artificial recommended output.

TOPSIS algorism is a multi-objective decision making method (Hwang & Yoon, 1981). It is a sorting method introduced by Hwang C. L. and Yong K. S. This algorithm is identified as an approximation of “ideal solution” in decision analysis. This method is widely applied in many fields including construction project bidding, corporate staff performance appraisal, service agency performance appraisal assessment, comprehensive assessment of regional economic instances and logistics solutions decisions. Personalized recommendation in e-commerce is a sorting method to meet the individual needs, and it has many in common with the application we have mentioned above. Therefore, TOPSIS algorithm can also be used in e-commerce. Through user surveys, we get the “ideal solution” and sort the products or service order by the closeness between them and the “ideal solution”.

Complete Article List

Search this Journal:
Open Access Articles
Volume 18: 4 Issues (2020): Forthcoming, Available for Pre-Order
Volume 17: 4 Issues (2019): 3 Released, 1 Forthcoming
Volume 16: 4 Issues (2018)
Volume 15: 4 Issues (2017)
Volume 14: 4 Issues (2016)
Volume 13: 4 Issues (2015)
Volume 12: 4 Issues (2014)
Volume 11: 4 Issues (2013)
Volume 10: 4 Issues (2012)
Volume 9: 4 Issues (2011)
Volume 8: 4 Issues (2010)
Volume 7: 4 Issues (2009)
Volume 6: 4 Issues (2008)
Volume 5: 4 Issues (2007)
Volume 4: 4 Issues (2006)
Volume 3: 4 Issues (2005)
Volume 2: 4 Issues (2004)
Volume 1: 4 Issues (2003)
View Complete Journal Contents Listing