GAN기반의 하이브리드 협업필터링 추천기 연구

송희석 한국데이터전략학회 2022 Journal of information technology applications & m Vol.29 No.6

As deep learning technology in natural language and visual processing has rapidly developed, collaborative filtering-based recommendation systems using deep learning technology are being actively introduced in the recommendation field. In this study, OCF-GAN, a hybrid collaborative filtering model using GAN, was proposed to solve the one-class and cold-start problems, and its usefulness was verified through performance evaluation. OCF-GAN based on conditional GAN consists of a generator that generates a pattern similar to the actual user preference pattern and a discriminator that tries to distinguish the actual preference pattern from the generated preference pattern. When the training is completed, user preference vectors are generated based on the actual distribution of preferred items. In addition, the cold-start problem was solved by using a hybrid collaborative filtering recommendation method that additionally utilizes user and item profiles. As a result of the performance evaluation, it was found that the performance of the OCF-GAN with additional information was superior in all indicators of the Top 5 and Top 20 recommendations compared to the existing GAN-based recommender. This phenomenon was more clearly revealed in experiments with cold-start users and items. .

Piaget 의 조작형식에 따른 고등학교 지구과학 교과서의 분석

송희석 한국지구과학회 1988 韓國地球科學會誌 Vol.9 No.2

The aim of this study is to provide useful advices and basic data for the future Earth Science Education by analyzing the High School Earth Science textbooks. The analysis included 1) whether the contents of the textbook are properly selected suiting with Piaget's Intellectual development stage 2) whether the basic concept and successive concept are properly selected and organized. This study arrived at the following conclusions. 1) The greater part of highschool students are not able to think the formal operation when they are in need of thinking power in formal operation in the study of 48.696 of curriculum. 2) The contents of Earth Science textbook should be developed naturally from the concrete operation stage to the formal operation stage, but the methods of approaching curriculum are scattered through out the textbook. 3) There are many lessons that can't be practiced in laboratory. The textbook should be developed naturally from the simple structures to the complicated ones. Also it is necessary for teachers to conceive those problems and to be able to help the students approach the curriculum easily.

유비쿼터스 비즈니스 모델 저장소를 이용한 경쟁 전략 수립 방법

송희석,한관희,여재현,박광만,이광희 대한산업공학회 2008 산업공학 Vol.21 No.3

It is important to choose and develop right business model in the process of implementing the ubiquitous service concept. In particular, it is also required to build a method which can assist in establishing a business model in a competitive perspective to differentiate business strategy. However, we could not find the existing studies which tried to develop business model including methods on how to differentiate business strategy in a competitive environment. In this study, we propose an ontology for repository of ubiquitous business model and address on how to apply the proposed business model repository to real ubiquitous industry. We also built a repository of unlicensed wireless device industry to check the applicability of proposed ubiquitous business model ontology and to show the various examples. We expect that building the business model repository contributes to reduce the redundancy and inefficiency for industry as well as to increase likelihood of business success for an individual company.

소셜 네트워크 사용자간 신뢰수준 예측 모형: 트위터 적용사례

송희석 에스케이텔레콤 (주) 2013 Telecommunications Review Vol.23 No.1

소셜네트워크에서 사용자간 신뢰는 해당 소셜네트워크에서 제공하는 사회적 자본의 가치수준을 결정하는 중요한 요인이다. 본 연구는 소셜네트워크에서 정보제공자에 대한 신뢰수준을 예측할 수 있는 사용자간 신뢰수준 예측 모형을 개발하는 것을 목표로 한다. 사용자간 신뢰수준 예측모형을 구성하기 위한 방법으로 신뢰에 대한 다양한 사회과학의 연구모형을 조사하여 신뢰에 영향을 미치는 요인을 추출하고 이들을 사용자간 상호작용 행위기록으로부터 측정하여 신뢰를 예측하는 방법을 제안하고 있다. 또한 트위터 신뢰링크 데이터를 수집하여 트위터 사용자간 신뢰를 예측하는 데 본 연구에서 제안한 방법을 적용하고 그 성능을 평가하였다. 본 연구에서 제안한 방법으로 예측된 신뢰수준은 신뢰자와 피신뢰자간 상호작용이 진행되면서 수정이 되는 동적인 특성을 가진다.

탐구적 지구과학 교육을 위한 교수 - 학습 모형연구

송희석 한국지구과학회 1985 한국지구과학회지 Vol.6 No.2

This study aims at giving effective suggestions for future Earth Science Education by making a theoretical teaching-learning model for inquiry science. The main subject of this study is to establish the learning system, educational goal and guide theory of Earth Science Education. Through above mentioned, following suggestions are presented: (1) The experiments and observations should be more emphasized than lectures in Earth Science Education. (2) It is necessary to retrain Earth Science teachers in order to improve their ability to guide the student's experiments. (3) Well designed laboratory equipments and kids of good quality and resonable price should be supplied. (4) Every teacher should work on positive line, extending the range of information through participating in the Korean Earth Science Education society, Earth Science meetings and various science education seminars.

심층신경망 기반의 뷰티제품 추천시스템

송희석 한국데이터전략학회 2019 Journal of information technology applications & m Vol.26 No.6

Many researchers have been focused on designing beauty product recommendation system for a long time because of increased need of customers for personalized and customized recommendation in beauty product domain. In addition, as the application of the deep neural network technique becomes active recently, various collaborative filtering techniques based on the deep neural network have been introduced. In this context, this study proposes a deep neural network model suitable for beauty product recommendation by applying Neural Collaborative Filtering and Generalized Matrix Factorization (NCF + GMF) to beauty product recommendation. This study also provides an implementation of web API system to commercialize the proposed recommendation model. The overall performance of the NCF + GMF model was the best when the beauty product recommendation problem was defined as the estimation rating score problem and the binary classification problem. The NCF + GMF model showed also high performance in the top N recommendation.

소셜네트워크에서 사용자간 신뢰전파의 전이성과 결합성 특성

송희석 한국정보기술응용학회 2011 한국정보기술응용학회학술대회 Vol.2011 No.1

Deep Neural Network Models to Recommend Product Repurchase at the Right Time : A Case Study for Grocery Stores

송희석 한국데이터전략학회 2018 Journal of information technology applications & m Vol.25 No.2

Despite of increasing studies for product recommendation, the recommendation of product repurchase timing has not yet been studied actively. This study aims to propose deep neural network models usingsimple purchase history data to predict the repurchase timing of each customer and compare performances of the models from the perspective of prediction quality, including expected ROI of promotion, variability of precision and recall, and diversity of target selection for promotion. As an experiment result, a recurrent neural network (RNN) model showed higher promotion ROI and the smaller variability compared to MLP and other models. The proposed model can be used to develop a CRM system that can offer SMS or app-based promotionsto the customer at the right time. This model can also be used to increase sales for product repurchase businesses by balancing the level of ordersas well as inducing repurchases by customers.

인재매칭을 위한 내용기반 척도학습모형의 설계

송희석 한국정보기술응용학회 2020 Journal of information technology applications & m Vol.27 No.6

The job mismatch between job seekers and SMEs is becoming more and more intensifying with the serious difficulties in youth employment. In this study, a bi-directional content-based metric learning model is proposed to recommend suitable jobs for job seekers and suitable job seekers for SMEs, respectively. The proposed model not only enables bi-directional recommendation, but also enables HR matching without relearning for new job seekers and new job offers. As a result of the experiment, the proposed model showed superior performance in terms of precision, recall, and f1 than the existing collaborative filtering model named NCF+GMF. The proposed model is also confirmed that it is an evolutionary model that improves performance as training data increases.

시스템다이내믹스 기반의 다세대 확산 수요 예측 : 이동통신 가입자 수요 예측 적용사례

송희석,김재경,Song, Hee Seok,kim, Jae Kyung 한국데이터전략학회 2017 Journal of information technology applications & m Vol.24 No.2

Forecasting long-term mobile service demand is inevitable to establish an effective frequency management policy despite the lack of reliability of forecast results. The statistical forecasting method has limitations in analyzing how the forecasting result changes when the scenario for various drivers such as consumer usage pattern or market structure for mobile communication service is changed. In this study, we propose a dynamic model of the mobile communication service market using system dynamics technique and forecast the future demand for long-term mobile communication subscriber based on the dynamic model, and also experiment on the change pattern of subscriber demand under various scenarios.