• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04b
• /
• pp.472-474
• /
• 2001

웹 기반의 쇼핑몰 사이트의 수가 많아지고 그 이용량이 증가하면서, 차별화된 고객 서비스를 위해 다양한 데이터마이닝 기술들이 적용되고 있다. 특히 고객의 취향에 부합하며 그의 필요를 만족하는 상품을 고객에게 제안하는 추천 시스템을 위해 정보 필터링(information filtering) 알고리즘들이 사용되고 있다. 많은 추천 시스템들은 고객들이 상품에 대해 부여한 선호도 정보를 기반으로, 현재 사용중인 고객에게 그와 취향이 비슷한 고객들이 선택했으며, 아직 그가 선택한 적이 없는 상품을 추천하는 협력적 필터링(collaborative filtering) 방법을 사용하고 있다. 본 연구에서는 보통의 협력적 필터링 방법에 내용기반 필터링(content-based filtering) 방법을 적용하고, 고객의 상품에 대한 선호도 점수를 자동으로 계산할 수 있도록 하는 방법을 제안하여 적용함으로써 협력적 필터링 방법을 개선하였다.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2002.05a
• /
• pp.312-319
• /
• 2002

A recommendation system tracks past action of a group of users to make a recommendation to individual members of the group. The computer-mediated marking and commerce have grown rapidly nowadays so the concerns about various recommendation procedure are increasing. We introduce a recommendation methodology by which Korean department store suggests products and services to their customers. The suggested methodology is based on decision tree, product taxonomy, and association rule mining. Decision tree is to select target customers, who have high purchase possibility of recommended products. Product taxonomy and association rule mining are used to select proper products. The validity of our recommendation methodology is discussed with the analysis of a real Korean department store.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2005.04a
• /
• pp.181-184
• /
• 2005

무수히 많은 정보 중에서 특정 사용자에게 가장 유용할 것으로 판단되는 정보를 추천하여 제공함으로써 특정 사용자의 편의를 돕는 시스템이 추천시스템이다. 이러한 추천시스템에 성공적으로 적용된 알고리즘이 협력적 필터링이며 이것은 다른 사용자로부터 먼저 평가된 웹 문서를 제공받아 이를 축적하고 다시 사용자에게 환원하는 알고리즘이다. 하지만 이 알고리즘은 초기평가, 희소성, 확장성 둥의 문제점을 내포하고 있다. 따라서 본 논문은 이러한 문제점을 해결하고 성능 향상을 하기 위해 적용된 개인화 추천시스템 관련 최신 알고리즘들을 비교하고 분석한 결과를 제시한다. 이를 위해 먼저 최근에 발표된 협력적 필터링과 최근접 이웃 알고리즘, 인공 지능기술을 이용한 알고리즘, 군집화 알고리즘 둥 각각에 대한 기술적 분석 결과를 수행한다. 그런 후 이들 다양한 알고리즘들의 조합을 통한 성능 향상 결과에 대한 비교분석과 각각의 조합에 대한 장단점 분석 결과도 또한 제시한다.

• Journal of Information Processing Systems
• /
• v.8 no.2
• /
• pp.191-212
• /
• 2012

Since a social network by definition is so diverse, the problem of estimating the preferences of its users is becoming increasingly essential for personalized applications, which range from service recommender systems to the targeted advertising of services. However, unlike traditional estimation problems where the underlying target distribution is stationary; estimating a user's interests typically involves non-stationary distributions. The consequent time varying nature of the distribution to be tracked imposes stringent constraints on the "unlearning" capabilities of the estimator used. Therefore, resorting to strong estimators that converge with a probability of 1 is inefficient since they rely on the assumption that the distribution of the user's preferences is stationary. In this vein, we propose to use a family of stochastic-learning based Weak estimators for learning and tracking a user's time varying interests. Experimental results demonstrate that our proposed paradigm outperforms some of the traditional legacy approaches that represent the state-of-the-art technology.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2014.11a
• /
• pp.1042-1045
• /
• 2014

Recently several recommender systems have been developed in a variety of applications, but providing accurate recommendations that match the preferences and constraints of various users is quite challenging. This paper presents a method of recommending digital products based on the past preference of a user on the price, rating and selling volume of a product. Experimental results of the method with actual data of Amazon showed that the average accuracy of the recommendations made by the method is 85%. Although the results are preliminary, the method is potentially capable of making more accurate personalized recommendations than existing methods.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.11a
• /
• pp.1135-1136
• /
• 2013

인터넷의 발달함에 따라 데이터가 기존에 비해 기하급수적으로 늘어나게 되는 이른바 빅데이터 시대를 맞이하게 되었다. 이러한 빅데이터는 기존의 시스템으로 처리하기가 쉽지 않아 이를 처리하기 위해 하둡이 개발되었다. 하둡은 분산파일 시스템으로 기존의 시스템에 비해 빅데이터를 처리하는데 적합하며 이를 이용한 다양한 오픈 소스들이 등장하게 된다. 그중 기계학습 알고리즘을 구현한 오픈소스 Mahout은 추천 시스템을 구현하는데 적합하다. 이를 이용하여 기존에 구현한 개인화 영화 추천 시스템을 하둡 시스템으로 구현하고 기존의 XLMiner로 구현한 시스템과 결과를 비교해 본다.

• The KIPS Transactions:PartB
• /
• v.19B no.1
• /
• pp.31-36
• /
• 2012

In recent years, there has been increasing interest in recommender systems which provide users with personalized suggestions for products or services. In particular, researches of collaborative filtering analyzing relations between users and items has become more active because of the Netflix Prize competition. This paper presents the reinforcement learning approach for collaborative filtering. By applying reinforcement learning techniques to the movie rating, we discovered the connection between a time sequence of past ratings and current ratings. For this, we first formulated the collaborative filtering problem as a Markov Decision Process. And then we trained the learning model which reflects the connection between the time sequence of past ratings and current ratings using Q-learning. The experimental results indicate that there is a significant effect on current ratings by the time sequence of past ratings.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.167-183
• /
• 2023

In the recent field of recommendation systems, various studies have been conducted to model sparse data effectively. Among these, GLocal-K(Global and Local Kernels for Recommender Systems) is a research endeavor combining global and local kernels to provide personalized recommendations by considering global data patterns and individual user characteristics. However, due to its utilization of kernel tricks, GLocal-K exhibits diminished performance on highly sparse data and struggles to offer recommendations for new users or items due to the absence of side information. In this paper, to address these limitations of GLocal-K, we propose the GEase-K (Global and EASE kernels for Recommender Systems) model, incorporating the EASE(Embarrassingly Shallow Autoencoders for Sparse Data) model and leveraging side information. Initially, we substitute EASE for the local kernel in GLocal-K to enhance recommendation performance on highly sparse data. EASE, functioning as a simple linear operational structure, is an autoencoder that performs highly on extremely sparse data through regularization and learning item similarity. Additionally, we utilize side information to alleviate the cold-start problem. We enhance the understanding of user-item similarities by employing a conditional autoencoder structure during the training process to incorporate side information. In conclusion, GEase-K demonstrates resilience in highly sparse data and cold-start situations by combining linear and nonlinear structures and utilizing side information. Experimental results show that GEase-K outperforms GLocal-K based on the RMSE and MAE metrics on the highly sparse GoodReads and ModCloth datasets. Furthermore, in cold-start experiments divided into four groups using the GoodReads and ModCloth datasets, GEase-K denotes superior performance compared to GLocal-K.

• Journal of KIISE:Software and Applications
• /
• v.37 no.3
• /
• pp.185-193
• /
• 2010

With the emergence of ubiquitous computing era, various models for providing personalized service have been proposed, and, especially, several recommendation service schemes have been proposed to give tailored services to users proactively. However, the previous recommendation service schemes utilize a wide range of data without and filtering and consider the limited context-aware information to predict user preferences so that they are not adequate to provide personalized service to users. In this paper, we propose an adaptive recommendation service scheme which proactively provides suitable services based on the current context. We use accumulated interaction contexts (IC) between users and devices for predicting the user's preferences and recommend adaptive service based on the current context by utilizing clustering and collaborative filtering. The clustering algorithm improves efficiency of the recommendation service by focusing and analyzing the data that is collected from the locations nearby the users. Collaborative filtering guarantees an accurate recommendation, even when the data is insufficient. Finally, we evaluate the performance and the reliability of the proposed scheme by simulations.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.9
• /
• pp.835-840
• /
• 2008

This paper proposes an enhanced recommendation technique for personalized e-commerce portal analyzing various attitudes of customer. The attitudes are classifies into three types such as "purchasing product", "adding product to shopping cart", and "viewing the product information". We implicitly track customer attitude to estimate the rating of products for recommending products. We classified user groups which have similar preference for each item using implicit user behavior. The preference similarity is estimated using the Cross Correlation Coefficient. Our recommendation technique shows a high degree of accuracy as we use age and gender to group the customers with similar preference. In the experimental section, we show that our method can provide better performance than other traditional recommender system in terms of accuracy.