Performance Enhancement of Collaborative Filtering through the Fusion Algorithm of Singular Value Decomposition and Association Rule

이은지(Eunji Lee),진서훈(Seohoon Jin),배준성(Jun Sung Bae) 한국자료분석학회 2018 Journal of the Korean Data Analysis Society Vol.20 No.1

Collaborative filtering (CF) is the leading algorithm as a recommendation system. However, the algorithm falls short due to the limitations of sparsity and scalability. Sparsity problem comes when the user has few items purchased or reviewed. Scalability problem occurs under too many users and items. In this study, the limitations of CF are enhanced through the usage of singular value decomposition (SVD) and association rule (AR). AR searches for relevance among items, and SVD decreases the dimensionality of the data to be applied on CF. This hybrid CF algorithm, which is combined with SVD and AR, is compared with other recommendation algorithms. Items which are included in computation were chosen through AR. SVD solved the short falls of data sparsity of CF. Both item-based CF and user-based CF are considered. The hybrid algorithm showed better performance than other recommendation systems under sparsity and scalability problem.

모바일 콘텐츠 활용과 모바일 인터넷에 대한 적극적 태도와의 연관성 분석

장세정(Sejung Jang),진서훈(Seohoon Jin) 한국전자거래학회 2016 한국전자거래학회지 Vol.21 No.4

국내 무선인터넷 산업은 지속적으로 성장하여 향후 웹 접속의 대표기기는 PC가 아닌 모바일 디바이스가 될 것으로 전망된다. 이러한 모바일 디바이스 수요의 폭발적인 증가는 우리들의 시간․공간적 제한을 사라지게 하는 등 편리한 생활을 누릴 수 있도록 만들어 주었지만 정보격차 등 여러 가지 부작용도 나타나게 되었다. 본 논문에서는 모바일 기기를 활용한 콘텐츠 활용도와 모바일 인터넷에 대한 적극적 태도를 세대 간, 교육수준, 경제수준에 따라 분석하고 모바일 콘텐츠 활용의 높음이 적극적 태도에 영향을 주는지에 대해 분석하고자 한다. 연구결과, 연령대가 젊을수록, 교육수준이 높을수록, 경제수준이 높을수록 모바일 콘텐츠에 대한 이용도가 높고 모바일 인터넷 이용에 대한 적극적 태도도 높으며, 모바일 콘텐츠에 대한 이용도가 높을수록, 모바일 인터넷 이용에 대한 적극적 태도도 높다는 것을 확인하였다. 이러한 분석 결과는 모바일 콘텐츠의 활용, 적극적 태도에 개인 특성에 따른 격차를 파악하여 이를 완화시켜줄 정책에 활용하고자 한다. As the industry of Mobile Internet has been gradually developed, the major device for accessing the web is predicted to shift from PC to Mobile devices in the near future. This drastic increase of demand in Mobile devices has drawn a more convenient standard of living by transcending time and space in our lives, however, it triggered various side effects such as a digital divide. This study was conducted to analyze how people utilize the contents via Mobile devices and their active attitude toward Mobile Internet by dividing individual characteristics into generations, education level and income level, and also identify whether Mobile contents utilization has an effect on the active attitude. The research results could be used for establishing a policy to narrow the gap in Mobile contents use and active attitude by individual characteristics.

주성분 회귀에서 원변수의 신뢰구간 구축에 관한 연구

김지은(Gee Eun Kim),진서훈(Seohoon Jin) 한국자료분석학회 2020 Journal of the Korean Data Analysis Society Vol.22 No.3

주변에서 일어나는 대부분의 현상들은 독립적으로 존재하기보단 서로 밀접히 상호의존하는 경우가 많다. 이 중 수치형 변수들로 이루어진 다변량 자료는 회귀분석을 통해 종속변수와 독립변수들의 관계를 이해할 수 있다. 그러나 독립변수들 사이에 높은 상관관계가 존재할 경우, 다중 공선성의 문제로 인해 타당한 분석결과가 나오지 않기 때문에 다양한 방법으로 변수들을 제거하거나 축소, 요약하여 분석을 시행한다. 변수들을 축소하는 대표적인 방법이 주성분 분석이며, 이를 이용한 회귀분석이 주성분 회귀분석이다. 그러나 주성분을 이용할 경우 원변수의 의미를 해석하는 것은 쉽지 않다. 회귀모형이 주성분의 식으로 구성되어 주성분의 해석이 명확하지 않을 경우 개별 원변수의 설명을 시도하는 것은 매우 까다롭다. 따라서 본 논문에서는 주성분 회귀분석에서의 결과해석을 보다 직관적인 관점으로 접근해보고자 주성분 회귀분석에서 원변수의 신뢰구간을 구해보고, 유의한 독립변수들을 이용하여 결과해석을 도출해 내었다. 사례 데이터를 이용하여 신뢰구간을 구축한 결과 의미있는 변수의 탐지가 가능하였고 종속변수에 대한 영향의 방향성을 파악할 수 있었다. Most of the phenomena that occur in the surroundings are more often interdependent than each other. Among these, multivariate data composed of numerical variables can understand the relationship between dependent and independent variables through regression analysis. However, if there is a high correlation between the independent variables, due to the problem of multicollinearity, it is difficult to get valid analysis result, so the variables are removed, reduced, or summarized in various ways. The main method of reducing variables is principal component analysis, and regression analysis using this is principal component regression analysis. However, when using the principal component, it is not easy to interpret the meaning of the original variable. If the regression model is composed of the equations of the principal components and the interpretation of the principal components is not clear, it is very difficult to attempt to explain the individual original variables. Therefore, in this paper, in order to approach the result analysis in the principal component regression from a more intuitive perspective, the confidence interval of the original variable is calculated in the principal component regression, and the result analysis is derived using significant independent variables. As a result of establishing a confidence interval using case data, it was possible to detect meaningful variables and to grasp the direction of influence on dependent variable.

시각화를 통한 잔류물질 정보 사이트 활용도 제고 방안 연구

박세준 ( Se-joon Park ),진서훈 ( Seohoon Jin ) 대한설비관리학회 2021 대한설비관리학회지 Vol.26 No.3

Visualization was studied based on data from drug residue limits for foods provided by Ministry of Food and Drug Safety. Residual Substances Information Web Site provides users with information on the residual acceptance criteria of agricultural medicines for foods we consume or supply. However, many agricultural products and residues provide information in the form of text or tables, which makes it difficult to utilize information from a user's perspective. In this paper, we wanted to study ways to increase utilization by visualizing information that users are interested in and residues that are problematic to many users. Similar food and residue information was visualized using a multi-dimensional scaling method, making it easier for users to grasp information. Using the user log data of the Residual Substance Information web site, the search terms of food and residues are visualized and provided based on association using the frequency of simultaneous search for food and residues. Through this visualization, we expect to be able to help those who want to use the residues information web site.

식품별 잔류물질허용기준에 의한 군집화 연구

이지원 ( Ji-won Lee ),진서훈 ( Seohoon Jin ) 대한설비관리학회 2020 대한설비관리학회지 Vol.25 No.4

In the modern society, as interest in health and beauty increases, interest in food are growing. Accordingly, concerns about pesticides in food are also increasing. Pesticides are essential for harvesting agricultural products. It is expected that all pesticides will disappear by harvest time, but there are cases where it is not. Ministry of Food and Drug Safety is providing drug residue limits for foods consumed by humans. However, since it deals with a wide variety of foods and substances, it may be difficult to manage and use the information. For such difficulties, this paper attempts to cluster each food and MRL using food-specific MRL data. It aims to help easy information management and understanding through cluster results. The data used in the analysis has many missing values in the data due to the characteristic of food-specific MRL. However, it is difficult to deal with the missing due to the characteristics of the acceptance criteria. To solve this problem, this paper proceeds with a two-stage clustering. First, the hierarchical clustering is performed based on the existence of the MRL criteria for each food, and the clustering between foods and MRLs is performed using the biclustering technique from the first result.

정압기 압력패턴 군집화를 통한 관리 효율화 방안 연구

이다연(Dayeon Lee),진서훈(Seohoon Jin),샤피 쿠슈부(Khushboo Shafi),김현진(Hyeonjin Kim) 한국가스학회 2023 한국가스학회 학술대회논문집 Vol.2023 No.11

결측치 대체 기법을 활용한 가스 정압기 압력 예측 및 응용 연구

박현우(Hyunwoo Park),김자현(Jahyun Kim),번가원(Jiayuan Fan),김재훈(Jaehoon Kim),진서훈(Seohoon Jin) 한국가스학회 2023 한국가스학회 학술대회논문집 Vol.2023 No.11

정압설비 관리를 위한 기계학습 활용 연구

이다연(Dayeon Lee),박현우(Hyunwoo Park),김현진(Hyeonjin Kim),번가원(Jiayuan Fan),샤피 쿠슈부(Khushboo Shafi),진서훈(Seohoon Jin) 한국가스학회 2023 한국가스학회 학술대회논문집 Vol.2023 No.5