• Journal of Computing Science and Engineering
• /
• v.7 no.1
• /
• pp.21-29
• /
• 2013

The introduction of recommender systems to existing online services is now practically inevitable, with the increasing number of items and users on online services. Popular recommender systems have successfully implemented satisfactory systems, which are usually based on collaborative filtering. However, collaborative filtering-based recommenders suffer from well-known problems, such as popularity bias, and the cold-start problem. In this paper, we propose an innovative collaborative-filtering based recommender system, which uses the concepts of Experts and Novices to create fine-grained recommendations that focus on being novel, while being kept relevant. Experts and Novices are defined using pre-made clusters of similar items, and the distribution of users' ratings among these clusters. Thus, in order to generate recommendations, the experts are found dynamically depending on the seed items of the novice. The proposed recommender system was built using the MovieLens 1 M dataset, and evaluated with novelty metrics. Results show that the proposed system outperforms matrix factorization methods according to discovery-based novelty metrics, and can be a solution to popularity bias and the cold-start problem, while still retaining collaborative filtering.

• Journal of Broadcast Engineering
• /
• v.19 no.5
• /
• pp.569-577
• /
• 2014

From the year 2009, 3D Stereoscopic movies and TV have been spotlighted after the huge success of a movie called "AVATAR". Moreover, most of 3D movies & contents are created by mixing real-life shots & virtual animated pictures, such as "Robocop 3", "Transformer 4" as shown in 2014. However, the stereoscopic 3D video film shooting with a traditional stereoscopic rig camera system, takes much more time to set the rig system and adjust the system setting for proper film making which necessarily resulting in bigger cost. In fact, these problems have depreciated the success of Avatar as decreasing demand for 3D stereoscopic video shooting. In this paper, inherent problems of traditional stereoscopic rig camera system are analyzed, and as a solution for the problems, a novel implementations of single-lens optical stereoscopic 3D camera system is suggested. The new system can be implemented to a technology for separating two lights when even those lights passing through in the same optical axis. The system has advantages of adjusting the setting and taking video compared with traditional stereoscopic 3D rig systems. Furthermore, this system can acquire comfortable 3D stereoscopic video because of the good characteristics of geometrical errors. This paper will be discussed the single-lens stereoscopic 3D camera system using rolling shutters, it will be tested geometrical errors of this system. Lastly, other types of single lens stereoscopic 3D camera system are discussed to develop the promising future of this system.

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 2009.05a
• /
• pp.27-31
• /
• 2009

본 연구는 웹상에서 거래되는 아이템을 고객에게 추천하는 추천시스템에서 추천대상 고객의 정보와 이웃 고객의 정보를 이용한 협력적 필터링 추천기법에서 선호도 예측을 위해 필요한 이웃의 수가 선호도 예측 정확도에 영향을 주고 있음을 제시하고 이를 이용한 선호도 예측치의 보정 방법에 대하여 제안한다. 본 연구의 제안을 위하여 이웃 기반의 협력적 필터링 알고리즘과 대응평균 알고리즘을 이용하여 MovieLens 1 million dataset에 대하여 선호도 예측 정확도를 분석하고 분석결과를 토대로 개별 선호도 예측에 소요된 이웃의 수와 예측 정확도의 관계를 분석하였다. 분석결과를 이용하여 이웃 수에 따라 선호도 예측 결과를 다수의 집단으로 구분하여 각 집단에서 이웃의 수를 이용한 선호도 예측 정확도 향상에 대한 방법을 제안한다. 본 연구의 제안을 통하여 기존 선호도 예측 알고리즘으로 생성된 예측 결과에 선호도 예측 과정에서 부가적으로 발생한 정보를 추가하여 최종 예측 결과를 향상시킬 수 있을 것으로 기대한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2011.06b
• /
• pp.216-219
• /
• 2011

인터넷 기술이 빠르게 발전하고 변화하여 현재는 많은 수의 컨텐츠와 프로그램 채널이 IP 네트워크를 통해 제공되면서 컨텐츠 서비스 사업자들은 좀 더 향상된 추천시스템이 필요하게 되었다. 그리고 사용자 참여중심의 인터넷 환경인 Web 2.0 시대가 도래하면서 사용자가 직접 생성한 정보들을 활용하는 다양한 연구가 진행되고 있다. 본 논문에서는 타겟 아이템에 대해 인터넷 상에 수많은 사용자들이 생성한 정보들을 ConceptNet을 활용하여 감정벡터를 추출하고 장르별로 분류하는 방법을 결합한 새로운 형태의 영화 추천시스템을 제안한다. 공개용 영화 데이터인 MovieLens 데이터 셋을 이용하여 실험하였고 성능평가는 RMSE 방법과 다양한 추천평가방법으로 기존 협업 필터링 추천시스템과 비교하였으며 실험 결과 기존방식보다 향상된 성능을 보였다.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2011.11a
• /
• pp.131-132
• /
• 2011

Image-based survey can be performed for a floating structure using the hydraulic model tests and empirical methods. I extracted the frame images from a digital camcoder movies and found the corner points for image matching. In the future, we will try the movie acquisition in the improved lab environment for a precise result.

• Journal of Korea Multimedia Society
• /
• v.20 no.2
• /
• pp.336-343
• /
• 2017

We propose a real time personalized recommendation algorithm on the mobile device. We use a unified collaborative filtering with reduced data. We use Fuzzy C-means clustering to obtain the reduced data and Konohen SOM is applied to get initial values of the cluster centers. The proposed algorithm overcomes data sparsity since it extends data to the similar users and similar items. Also, it enables real time service on the mobile device since it reduces computing time by data clustering. Applying the suggested algorithm to the MovieLens data, we show that the suggested algorithm has reasonable performance in comparison with collaborative filtering. We developed Android-based smart-phone application, which recommends restaurants with coupons and restaurant information.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10b
• /
• pp.169-173
• /
• 2006

유비쿼터스 상거래에서 사용자가 정보를 효율적으로 이용할 수 있도록 제어하고 필터링하는 일을 도와주는 개인화된 추천 시스템이 등장하였다. 더 나아가서는 사용자가 원하는 아이템을 예측하고 추천해주며, 이를 위해 협력적 필터링 기술을 적용하고 있다. 이는 사용자의 성향에 맞는 아이템을 예측하고 추천하기 위하여 비슷한 선호도를 가지는 사용자들 간의 유사도 가중치를 계산한다. 본 논문에서는 속성정보에 대한 사용자의 선호도를 고려하지 않은 문제점을 개선하기 위해서 속성정보를 이용한 연관 사용자의 선호도를 협력적 필터링 기술에 반영함으로써 추천의 정확도를 높이고자 한다. 그리고 협력적 필터링의 {연관 사용자-아이템} 행렬에서 사용자들 간의 연관 관계를 유지하면서 차원 수를 감소시키기 위해 ARHP 알고리즘을 이용하여 연관 사용자 군집을 한다. 제안된 방법의 성능 평가를 하기 위해 사용자가 아이템에 대해서 평가한 MovieLens 데이터 집합을 대상으로 평가되었으며, 기존의 Nearest Neighbor Model과 K-Means 군집보다 그 성능이 우수함을 보인다.

• Management Science and Financial Engineering
• /
• v.11 no.3
• /
• pp.137-153
• /
• 2005

Collaborative filtering-based recommendation systems make personalized recommendations based on users' ratings on products. Recommender systems must collect sufficient rating information from users to provide relevant recommendations because less user rating information results in poorer performance of recommender systems. To learn about new users, recommendation systems must first present users with an initial item list. In this study, we designed and analyzed seven selection strategies including the popularity, favorite, clustering, genre, and entropy methods. We investigated how these strategies performed using MovieLens, a public dataset. While the favorite and popularity methods tended to produce the highest average score and greatest average number of ratings, respectively, a hybrid of both favorite and popularity methods or a hybrid of demographic, favorite, and popularity methods also performed within acceptable ranges for both rating scores and numbers of ratings.

• 정보저장시스템학회:학술대회논문집
• /
• 2005.10a
• /
• pp.217-221
• /
• 2005

Nowadays, HD(High Definition) broadcasting is popular all over the world. Many people want to record the HD level contents. BD(Blu-ray Disc) is developed to satisfy the needs. BD player for home movie system was manufactured already. ODD maker will produce BD Drive for PC(Personal Computer) within few months. Backward Compatibility is very important point in BD Drive. Until now, two OL(Object Lens), two optical pickup and two deck system are proposed. We suggest Single OL system in this paper. Single OL is composed of two optical parts and it's weight is heavy. We introduce the magnetic circuit composed of two 3-Pole magnets. In Spite of heavy OL, We achieve the New Actuator which has high performance.

• Journal of Information Technology Applications and Management
• /
• v.16 no.1
• /
• pp.97-113
• /
• 2009

Collaborative filtering(CF) has been a very successful approach for building recommender system, but its widespread use has exposed to some well-known problems including sparsity and scalability problems. In order to mitigate these problems, we propose two novel models for improving the typical CF algorithm, whose names are ISCF(Item-Selected CF) and USCF(User-Selected CF). The modified models of the conventional CF method that condense the original dataset by reducing a dimension of items or users in the user-item matrix may improve the prediction accuracy as well as the efficiency of the conventional CF algorithm. As a tool to optimize the reduction of a user-item matrix, our study proposes genetic algorithms. We believe that our approach may relieve the sparsity and scalability problems. To validate the applicability of ISCF and USCF, we applied them to the MovieLens dataset. Experimental results showed that both the efficiency and the accuracy were enhanced in our proposed models.