Clustering Method using Item Preference based on RFM for Recommendation System in u-commerce

Transcription

1 Clustering Method using Item Preference based on RFM for Recommendation System in u-commerce Young Sung Cho 1, Song Chul Moon 2, Seon-phil Jeong 3, In-Bae Oh 4, Keun Ho Ryu 1 1 Department of Computer Science, Chungbuk National University, Cheongju, Korea 2 Department of Computer Science, Namseoul University, Cheonan-city, Korea, Korea 3 Computer Science and Technology, DST, BNU-HKBU United International College 4 Juseong University, Chungbuk, Korea { youngscho@empal.com, moon@nsu.ac.kr, spjeong@uic.edu.hk, iboh@jsc.ac.kr, khryu@dblab.chungbuk.ac.kr} Abstract. This paper proposes a new method using clustering of item preference based on RFM(Recency, Frequency, Monetary) for recommendation system in u-commerce under fixed mobile convergence service environment which is required by real time accessibility and agility. In this paper, using a implicit method without onerous question and answer to the users, not used user's profile for rating to reduce customers search effort, it is necessary for us to keep the scoring of RFM to be able to reflect the attributes of the item and clustering in order to improve the accuracy of recommendation with high purchasability. To verify improved better performance of proposing system than the previous systems, we carry out the experiments in the same dataset collected in a cosmetic internet shopping mall. Keywords: RFM, Collaborative filtering, Clustering 1. Introduction Along with the advent of ubiquitous networking environment, it is becoming a part of our common life style of enjoying the wireless internet using intelligent portable device such as smart phone, PDA and smart pads, are increasing anytime or anyplace without any restriction of time and place. In these trends, the personalization becomes a very important technology. The customers want the recommendation system to recommend the items which they really wants on behalf of them. The possession of intelligent recommendation system is becoming the company's business strategy. A personalized recommendation system using RFM segmentation analysis technique to meet the needs of customers has been actually processed the research[1,2,3,4,5,6]. We can improve the accuracy of recommendation using clustering of item preference based on RFM so as to be able to reflect the attributes of items. As a result of that, we can propose the personalized recommendation system using clustering of item preference based on RFM. The next chapter briefly reviews the literature related to studies. The chapter 3 is described a new method for personalized recommendation system in detail, such as system architecture with sub modules, the procedure of processing the recommendation, the algorithm for proposing system. The chapterc 4 describes the evaluation of this system in order to prove the criteria of logicality and efficiency through the implementation and the experiment. In chapter 5, finally it is described the conclusion of paper and further research direction. 2. Relative works 2.1. RFM RFM(recency requency monetary) is generally used in database marketing and direct marketing and also is easy for us to analyze the purchased data to recommend the item with high purchasability by RFM scoring in this system. The customer's score and the item's score will be based on the analysis of the situation on the recommendation system. The RFM score will be shown how to determine the customer as follows. RFM = A R + B F + C M (1) The RFM score is correlated to the interest of e-commerce[4]. The RFM score can be a basis factor how to determine purchasing behavior on the internet shopping mall, is helpful to buy the item which they really want by the personalized recommendation. It is necessary for us to keep the analysis of RFM method to be able to reflect the attributes of the item in order to find the items with high purchasability. In this paper, we can use the customers data and purchased data with 60.98% in the rate of portion for the purchasing counts.

2 2.2. Collaborative Filtering Collaborative filtering means that the method of filtering is associated with the interests of a user by collecting preferences or taste information from many users. The terms of collaborative filtering comes from the method based on other users' preferences. There are two types of the method. One is the explicit method which is used user's profile for rating. The other is the implicit method which is not used user's profile for rating, The implicit method is not used user's profile for rating but is used user's web log patterns or purchased history data to show user's buying patterns so as to reflect the user's preferences. There are some kinds of the method of recommendation, such as collaborative filtering, demographic filtering, rule-base filtering, contents based filtering, the hybrid filtering which put such a technique together and association rule and so on in data mining technique currently. The explicit method can not only reflect exact attributes of item, but also still has the problem of sparsity and scalability, though it has been practically used to improve these defects Clustering Clustering can be defined as the process of grouping phisical or abstract objects into classes of similar objects. Clustering involves classifying or segmenting the data into groups based on the natural structure of the data. Clustering techniques [7,8] fall into a group of undirected data mining tools. Clustering algorithm is a kind of customer segmentation methods commonly used in data mining. In this paper, we can do clustering the customers data using K-means clustering algorithm to segment customers and finally forms groups of customers with different features. Through analyzing different groups of customers, we try to do the recommendation for the target customers of internet shopping mall efficiently. The principle of clustering is maximizing the similarity inside an object group and minimizing the similarity between the object groups. K- means is the most well-known and commonly, used partition methods are the simplest clustering algorithm. In the k-means algorithm, cluster similarity is measured in regard to the mean value of the objects in a cluster, which can be viewed as the cluster's center of gravity. This algorithm uses as input a predefined number of clusters that is the k from its name. Mean stands for an average, an average location of all the members of a particular cluster. The euclidean norm is often chosen as a natural distance which customer a between k measure in the k-means algorithm. The ai means the preference of attribute i for customer a. (2) In this paper, we can use the K-means algorithm[9]. 3. Proposing Recommendation System 3.1. System Architecure In this section, we can depict the system configuration concerning the recommendation system using mining association rules based on RFM under fixed mobile convergence service environment which is required by real time accessibility and agility. This system had four agent modules which have the analytical agent, the recommendation agent, the learning agent, the data mining agent in the internet shopping mall environment. We observed the web standard in the web development, so developed the interface of internet to use full browsing in mobile device. As a matter of course, we can use web browser in wired internet to use our recommendation system. We can use the system under WAP in mobile web environment by using feature phone as well as using the internet browser such as safari browser of iphone and Google chrome browser based on android so as to use our system by using smart phone Clustering Algorithm Using Item Preference In this section, we can depict clustering algorithm of item preference based on purchased data to join the customer information for pre-processing so as to be possible to recommend the item with efficiency. Firstly, the sub system can do the task of clustering the neighborhood of customers data and the purchased data in the learning agent. The sub system can classify the purchased data on the basis of several social variables such as customer s score, demographic variables in the code of classification: age, gender, occupation previously for recommendation efficiently. Thus we can use the cluster based on the item preference after being sorted by the code of item category. In this part, it is necessary for us to use clustering algorithm using item preference been different from the paper[4] using the preference of item category. As a matter of course, in this paper, we can use the purchased data having the RFM score of brand item, with a lot of purchasing counts, between the score is more than 19 points and the score is less than 40 points. The procedural algorithm for clustering of item preference is depicted as the following.

3 Table 1. Procedural algorithm for Clustering of item preference Input : Item Category Code Table(CCT), Customer-Item Category-Preference(UCP) Matrix, Output : Feature Vector, Purchased data neighborhood Group begin 1. Classify the purchased data of the Feature Vector by the basis of several social variables; // the Feature Vector which has score, age, gender, occupation 2. For( each brand Item in whole CCT) Compute the average of brand item preference in item category Endfor; 2.1 Compute the average of item preference(pref_uc(u,c) ) by the function of Group by as a aggregative function, it is normalized.; 2.2 For( each Item Category Compute the average of item preference based on CCT Endfor; 2.3 Create the cluster from purchased data using item preference sorted by CCT, extracted by the Feature Vector; // V is the set of all the item preferences that M in CCT, 3. Create the cluster of neighborhood using K-means clustering algorithm ; // The neighborhood cluster is extracted using by k-means clustering algorithm. End; 3.3. The procedural algorithm for recommendation The login user can read users' information and recognize the code of classification. The system can search the information in the cluster selected by using the code of classification. It can scan the preference as the average of brand item in the cluster, suggest the brand item in item category selected by the highest probability for preference as the average of brand item. This system can create the list of recommendation with TOP-N of the highest preference of item to recommend the item with purchasability efficiently. This system takes the cross comparison with purchased history data in order to avoid the duplicated recommendation which it has ever taken. The following algorithm is the procedure algorithm for a personalized recommendation system using clustering of item preference based on RFM. Table 2. Procedure algorithm for Personalized recommendation System using clustering of item preference based on RFM Step 1 : When the user joins the membership, user s information is created, managed the score of user and item, the code of classification reflected social variable such as age, gender, an occupation and propensity of a customer. Step 2 : The login user reads users' information and recognize the code of classification, classifies the cluster using the code of classification reflected demographic variable and user score. Step 3 : The system applies to the data having the RFM score of brand item with a lot of purchasing counts more than 79 points, searches the preference as the average of brand items in the cluster s data classified. Step 4 : The system can selects the highest preference as the average of brand item based on purchased data sorted by the code of item category, created the items of recommendation ordered by descending the preference of brand item. Step 5 : The system can create the list of recommendation with TOP-N of the highest preference of brand item to recommend the item with purchasability efficiently. Step 6 : The system executes the cross comparison with purchased history data in order to avoid the duplicated recommendation which it has ever taken. 4.. The environment of implementation and experiment & evaluation

4 4.1. Experimental environment We propose a new method using clustering of item preference based on RFM under ubiquitous computing environment. In order to do that, we make the implementation for prototyping of the internet shopping mall which handles the cosmetics professionally and do the experiment. We have carried out the implementation and the experiment for proposing system through system design, we have finished the system implementation about prototyping recommendation system. It could be improved and evaluated to proposing system through the result of experiment with the metrics such as precision, recall, F- measure as comparing the three systems. The 1 st system is proposing system called by proposal, the previous systems, which are the 2 nd system with the method(kccp) using the preference of item category, the third existing system Experimental data for evaluation We used 319 users who have had the experience to buy items in e-shopping mall, 580 cosmetic items used in current industry, 1600 results of purchased data recommended in order to evaluate the proposal system. It could be evaluated in Precision, Recall, F-measure for the recommendation system in clusters. It could be proved by the experiment through the experiment with learning data set for 12 months, testing data set for 3 months in a cosmetic cyber shopping mall[4]. We try to carry out the experiments in the same condition of the previous systems with dataset collected in a cosmetic internet shopping mall Experiment & Evaluation The proposing system's overall performance evaluation is precision, recall and F-measure for proposing system comparing than the previous systems which are existing system and the system(kccp) with the algorithm[4] proposed before. The performance was performed to prove the validity of recommendation and the system's overall performance evaluation. The metrics of evaluation for recommendation system in our system was used in the field of information retrieval commonly[10]. Table36. The result for table of precision, recall, F-measure for recommendation ratio by each cluster Fig. 1. The result of recommending ratio by precision Fig. 2. The result of recommending ratio by recall

5 Fig. 3. The result of recommending ratio by F-measure Fig. 4. The result of recommending items of cosmetics Above Table 3 presents the result of evaluation metrics (precision, recall and F-measure) for recommendation system. The new clustering method is improved better performance of proposing system than the previous systems. Our proposing system with the method using item preference is higher 37.46% in recall, higher 15% in F-measure even if it is lower 10.24% in precision than the system(kccp). As a result, we could have the recommendation system to be able to recommend the items with high purchasability. The following figure 4 is shown in the result of screen on a smart phone. The new clustering method is better performance than the previous method although it is lower in precision. 5. Conclusion Recently u-commerce as a application field under fixed mobile convergence service environment required by real time accessibility and agility, is in the limelight. Searching for wallpaper images with mobile device, such as cell phones, PDA, is inconvenient and complex in this ubiquitous computing environment[4]. We proposed a new clustering method using item preference based on RFM for recommendation system in u-commerce in order to to improve the accuracy of recommendation with high purchasability. We have described that the performance of the proposing system with new clustering method is improved better than the system(kccp) and existing system. To verify improved better performance of proposing, we carried out the experiments in the same dataset collected in a cosmetic internet shopping mall. It is meaningful to present a new clustering method using item preference based on RFM for recommendation system in u-commerce recommendation system in the large data environment. The following research will be looking for ways of a personalized recommendation by SOM clustering approach to increase the efficiency and scalability. Acknowledgements. This work 1) was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(mest) (No ) and this paper 2) was supported by funding of Namseoul University References. 1. Young Sung Cho, Moon Haeng Heo, Keun Ho Ryu, Implementation of Personalized recommendation System using RFM method in Mobile Internet Environment, KSCI, 13th-2 Vol, pp 1-5, Mar, (2008) 2. Young Sung Cho, Keun Ho Ryu, "Implementation of Personalized recommendation System using Demographic data and RFM method in e-commerce", 2008 IEEE International Conference on Management of Innovation & Technology Publication, (2008) 3. Jin Byeong Woon, Young Sung Cho, Keun Ho Ryu, Personalized e-commerce Recommendation System using RFM method and Association Rules, KSCI, 15th-12 Vol, pp , Dec, (2010) 4. Young Sung Cho, Seon-phil Jeong, Keun Ho Ryu, "Implementation of Personalized u-commerce Recommendation System using Preference of Item Category based on RFM", the 6th International Conference on Ubiquitous Information Technologies & Applications, pp , Dec, (2011) 5. Young Sung Cho, Keun Ho Ryu, "Personalized Recommendation System using FP-tree Mining based on RFM, KSCI, 17th-2 Vol, Feb., (2012) 6. Young Sung Cho, Song Chul Moon, Si Choon Noh, Keun Ho Ryu, "Implementation of Personalized recommendation System using k-means Clustering of Item Category based on RFM", 2012 IEEE International Conference on Management of Innovation & Technology Publication, Jun, (2012)

6 7. Collier K., Carey B., Grusy E., Marjaniemi C.,and Sautter D., (1998) A Perspective on Data Mining, Northern Arizona University. 8. Hand D., Mannila H., Smyth P. (2001), Principles of Data Mining. The MIT Press. 9. T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning Data Mining, Inference, and Prediction, Springer, (2001) 10. Jonathan L. Herlocker, Joseph A. Kosran, Al Borchers, and John Riedl, An Algorithm Framework for Performing Collaborative Filtering", Proceedings of the 1999 Conference on Research and Development in Information Retrival, (1999) Corresponding Author. Name: Keun Ho Ryu Address: School of Electrical & Computer Engineering, Chungbuk National University Cheongju, Chungbuk , Korea Affiliation: Chungbuk National University khryu@dblab.chungbuk.ac.kr