Using Genetic Algorithms for Personalized Recommendation Chein-Shung hwang, Yi-Ching Su, and Kuo-Cheng tseng Dept of Information Management Chinese Culture University, Taipei, Taiwan Dept of Information Management Chinmin Institute of Technology, Miaoli, Taiwan cshwangafaculty. pccu. edu. tw, poohhh@ms. chinmin ed tw. tikicams. chinmin. edu. tw Abstract. with the high-speed development of customer service orientation, it is essential that the enterprises must find and understand customers'interests and preferences and then provide for suitable products or services. Recom- mender systems provide one way of circumventing this problem. This paper de- scribes a new recommender system, which employs a genetic algorithm to learn personal preferences of customers and provide tailored suggestions. Keywords: Recommender systems: generic algorithm; collaborative filterin 1 Introduction The explosive growth of the world-wide-web has led to an influx of users and conse- quently, a huge increase in the volume of available on-line data. The volume of things is considerably more than any person can possibly filter through to find the ones that he/she will like. Recommender systems have emerged in response to the information overloaded problem. Most Personalized Recommender systems adopt two types of techniques: collaborative filtering approach and content-based approach. Collabora tive filtering approach finds other users that have shown similar tastes to the given user and recommends what they have liked to that user. But it is not well-suited to locating information for a specific content information need. On the other hand, con- tent-based approach recommends items based on the item contents that the user has liked in the past. Combining with content-based approach can eliminate the shortcom- ings of collaborative filtering approach and provide better recommendations [1] Many hybrid recommender systems have been developed for e-commerce applica tions. The typical steps of recommender systems can be described as follows. First customer preference profiles in terms of product features are analyzed and extracted from transaction file and product file. Second, a data mining technique is used to find similar customers who have shown similar interests as on-line customers. Finally, a list of recommendations is provided and can be further adjusted by the subsequent customers' feedbacks. However, for different application strategies, each preference J -S. Pan, S.M. Chen, and N.T. Nguyen(Eds ) ICCCI 2010, Part IL, LNAI 6422, Pp. 104-112, 2010 o Springer-Verlag Berlin Heidelberg 2010J.-S. Pan, S.-M. Chen, and N.T. Nguyen (Eds.): ICCCI 2010, Part II, LNAI 6422, pp. 104–112, 2010. © Springer-Verlag Berlin Heidelberg 2010 Using Genetic Algorithms for Personalized Recommendation Chein-Shung Hwang1 , Yi-Ching Su2 , and Kuo-Cheng Tseng2 1 Dept. of Information Management, Chinese Culture University, Taipei, Taiwan 2 Dept. of Information Management, Chinmin Institute of Technology, Miaoli, Taiwan cshwang@faculty.pccu.edu.tw, poohhh@ms.chinmin.ed.tw, tikic@ms.chinmin.edu.tw Abstract. With the high-speed development of customer service orientation, it is essential that the enterprises must find and understand customers' interests and preferences and then provide for suitable products or services. Recom￾mender systems provide one way of circumventing this problem. This paper de￾scribes a new recommender system, which employs a genetic algorithm to learn personal preferences of customers and provide tailored suggestions. Keywords: Recommender systems; generic algorithm; collaborative filtering. 1 Introduction The explosive growth of the world-wide-web has led to an influx of users and conse￾quently, a huge increase in the volume of available on-line data. The volume of things is considerably more than any person can possibly filter through to find the ones that he/she will like. Recommender systems have emerged in response to the information overloaded problem. Most Personalized Recommender systems adopt two types of techniques: collaborative filtering approach and content-based approach. Collabora￾tive filtering approach finds other users that have shown similar tastes to the given user and recommends what they have liked to that user. But it is not well-suited to locating information for a specific content information need. On the other hand, con￾tent-based approach recommends items based on the item contents that the user has liked in the past. Combining with content-based approach can eliminate the shortcom￾ings of collaborative filtering approach and provide better recommendations [1]. Many hybrid recommender systems have been developed for e-commerce applica￾tions. The typical steps of recommender systems can be described as follows. First, customer preference profiles in terms of product features are analyzed and extracted from transaction file and product file. Second, a data mining technique is used to find similar customers who have shown similar interests as on-line customers. Finally, a list of recommendations is provided and can be further adjusted by the subsequent customers' feedbacks. However, for different application strategies, each preference