It has been increasing the number and types of products on the u-commerce market and promoting interest in associated items rapidly. Therefore, agility and real time accessibility are crucial element in u-commerce. Existing collaborative filtering(CF) adopts evaluation methods based on personal profiles. However, these methods have been identified with difficulties in accurately analyzing the customers' tendencies and level of interest, as well as the problems of cost, consequently leaving customers unsatisfied. This paper proposes a new method of recommender system in u-commerce using cluster analysis of segmented merchandise to have different weights based on FRAT(Frequency, Recency, Amount and Type) method. Instead of using user's profiles for rating, we used an implicit method. It is necessary for us to make the task of preprocessing such as keeping FRAT method and clustering of merchandise category in order to recommend items with the profit/weight/importance of segmented merchandise. To verify improved performance of our proposing system compared to the previous system. We have conducted experiments with the same dataset, which was collected from a cosmetic web shopping mall.
[1]
John Riedl,et al.
An algorithmic framework for performing collaborative filtering
,
1999,
SIGIR '99.
[2]
Junho Ahn,et al.
An indoor augmented-reality evacuation system for the Smartphone using personalized Pedometry
,
2012,
Human-centric Computing and Information Sciences.
[3]
Keun Ho Ryu,et al.
Incremental Weighted Mining based on RFM Analysis for Recommending Prediction in u-Commerce
,
2013
.
[4]
Heikki Mannila,et al.
Principles of Data Mining
,
2001,
Undergraduate Topics in Computer Science.
[5]
Simon Parsons,et al.
Principles of Data Mining by David J. Hand, Heikki Mannila and Padhraic Smyth, MIT Press, 546 pp., £34.50, ISBN 0-262-08290-X
,
2004,
The Knowledge Engineering Review.