• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.418-419
• /
• 2018

본 논문에서는 사용자의 상품에 대한 평점 정보와 상품의 컨텐츠 정보를 모두 이용하는 하이브리드 추천 모델에 대해서 논의한다. 기존 논문들과는 다르게, 본 논문은 추천의 정확도를 높이기 위해 사용자가 상품의 컨텐츠 (예를 들면, 영화의 장르 또는 상품의 카테고리 등) 에 가질 수 있는 선호도를 예측하고, 이를 추가적으로 활용할 수 있는 딥러닝 기반의 추천 모델을 제안한다. 실세계의 데이터를 이용해서 제안하는 방법의 우수성을 보인다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.6
• /
• pp.169-179
• /
• 2017

The objective of this study is to identify the influence of utilization of SNS among consumers who watch performance art on their behavioral intention in the future when there has been an increasing usage of SNS among consumers along with an increasing importance of SNS in the field of performance art and also the influence of SNS characteristics on behavioral intention of performance art consumers in the future according to their preferred genres. According to the results of empirical analysis in this study, usefulness, pleasure, and social influence, as SNS characteristics, turned out to positively influence on behavioral intention in the future. According to the result of analysis in difference of influence on behavioral intention in the future depending on preferred performance art genre, usefulness, pleasure, and availability in order among SNS characteristics turned out to influence on behavioral intention in the future in pop-music/entertainment concert. In addition, only usefulness turned out to influence on behavioral intention in the musical. Usefulness and availability in order turned out to influence on behavioral intention in the act. Lastly, only usefulness turned out to influence on behavioral intention in the music/ballet. According to aforementioned results of the study, it implies that SNS characteristics of performance art audiences are influencing on behavioral intention in the future. In addition, since there is a difference in behavioral intention in the future among art consumers in each genre, it is required to differentiate utilization and strategies of SNS in each genre.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06b
• /
• pp.381-383
• /
• 2012

협업필터링은 지금까지 많은 추천시스템 연구에서 비교대상이 되거나 더 좋은 추천시스템 방법론을 개발하기 위해서 응용되고 있다. 일반적으로 협업필터링 기법은 명시적으로 관찰된 사용자들의 행동을 기반하는 방법이다. 본 연구에서는 LDA(Latent Dirichlet Allocation)을 이용해 사용자와 추천 대상이 되는 아이템의 숨겨진 특성을 추출하고, 이를 협업필터링기법에 응용했다. 영화 추천시스템 구축을 위한 실험에서, 사용자의 선호도는 다양한 영화 장르를 선호하는 비율로 나타난다는 가정(사용자기반)과 영화 또한 장르의 비율로 표현이 된다는 가정(아이템기반)을 했다. 이러한 가정을 토대로 사용자 사이와 영화 사이 간의 유사도를 정의하고, 협업필터링에 적용했을 때, 전통적인 협업필터링 기법보다 뛰어난 결과를 얻을 수 있었다.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.443-448
• /
• 2019

본 논문은 공공 데이터 Open API와 TMDB(The Movie Database) API를 이용하여 사용자의 선호 영화를 Google에서 제공해주는 Tensoflow로 인공신경망 딥러닝 학습하여 사용자가 선호하는 영화를 맞춤 추천하는 애플리케이션의 설계 및 구현에 대하여 서술한다. 본 애플리케이션은 사용자가 쉽게 영화를 추천받을 수 있도록 만들어진 애플리케이션으로 기존의 필터링 방식으로 추천하는 방식의 애플리케이션들과 달리 사용자의 취향을 딥러닝 학습을 통해 최적의 영화 Contents를 추천함과 아울러 기존 영화의 특성을 학습하여 흥행할 신규 영화를 예측하는 기능 또한 제공한다. 본 애플리케이션에 사용된 신규 영화 흥행 예측 모델은 약 85%의 정확도를 보이며 사용자 맞춤추천의 경우 기존 장르 추천이나 협업 필터링 추천보다 딥러닝을 통한 장르, 감독, 배우 등의 보다 세밀한 학습 추천이 가능하다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.1177-1179
• /
• 2015

최근, 수많은 종류의 도서가 출판되고 있다. 또한 도서의 분야와 장르, 종류가 다양해지고 그 양 역시도 방대해지고 있다. 이러한 상황에서 사용자에게 적절한 도서를 고르기란 어려운 일이다. 본 논문에서는 보다 편리하고 적절한 도서 선택을 위해 도서추천시스템을 제안한다. 사용자의 나이와 성별, 국내/외도서, 선호 장르에 가중치를 부여하고 협업필터링을 사용하는 추천 시스템을 제안한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2011.04a
• /
• pp.1285-1286
• /
• 2011

MovieLens를 대상으로 하는 추천 시스템에 대한 연구 중 k-NN 추천 방법은 정확도가 비교적 높지만 평점을 예측할 수 없는 상황이 발생할 수 있다. 본 논문에서는 기존 방법의 문제점을 해결한 장르기반 추천 방법 제안하고, 실험을 통하여 제안하는 방법이 모든 영화에 대한 평점의 예측이 가능함을 검증한다.

• Journal of Convergence for Information Technology
• /
• v.9 no.5
• /
• pp.1-6
• /
• 2019

As the mobile market expands, a variety of platforms are available to provide multimodal media content. Multimodal media content contains heterogeneous data, accordingly, user requires much time and effort to select preferred content. Therefore, in this paper we propose multimodal media content classification using keyword weighting for recommendation. The proposed method extracts keyword that best represent contents through keyword weighting in text data of multimodal media contents. Based on the extracted data, genre class with subclass are generated and classify appropriate multimodal media contents. In addition, the user's preference evaluation is performed for personalized recommendation, and multimodal content is recommended based on the result of the user's content preference analysis. The performance evaluation verifies that it is superiority of recommendation results through the accuracy and satisfaction. The recommendation accuracy is 74.62% and the satisfaction rate is 69.1%, because it is recommended considering the user's favorite the keyword as well as the genre.

• Journal of Digital Contents Society
• /
• v.13 no.2
• /
• pp.197-203
• /
• 2012

Although the increase in number of the programs provides rich entertainment to viewers, it also has caused a negative of making it hard for viewers to find out their favorite programs. To address the problem, several researches have been performed mainly focusing on the technologies to analyze a viewer's TV watching-patterns and to recommend a program (or a channel) based on the analysis when a viewer changes channels. The researches, however, have the trouble of frequently failing to choose proper programs because, in the real-world broadcasting circumstance, the programs are re-broadcast over a number of the channels and a set of programs of a genre are usually playing in the overlapped times. To avoid the trouble, the data service, proposed in the paper, allows a viewer to book "explicitly" his/her favorite programs and provides a set of functions of listing up the booked program's broadcasting schedules and reserving viewing or recording the booked programs.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.3
• /
• pp.305-310
• /
• 2019

Survival horror games aim to seek cathartic psychological escapism from the framed fear or hostile environment. However, if the goal of the game playing is only "survival" against hostile environment, the motivation of such environment adaptive games may not be based on the framed fear. We may call such games as "environment adaptive survival games". In this paper, we analyze the contents of six such survival games that have large user groups in Steam platform based on the pilot survey. We extract 13 almost common fun elements that induce the fun of such games. An online user survey was conducted through multiple survival game cafe to investigate who play such games. There was no gender differences in playing time but females prefer constructing tools as a fun element and play puzzle and simulation games more than males. We found that survival games could be welcomed by female users under 30 years old if 'horror' element was minimized.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.1
• /
• pp.281-290
• /
• 2017

This study aims to determine how demographic characteristics and SNS use traits of audiences are different depending on preferred performance types such as pop music concert, musical; play, dancing and ballet. SNS use traits are as follows: duration of SNS use, average number of access to SNS account, average time of SNS use, SNS activity type, motivation for SNS use, preferable type of SNS, annual total number of performance watched, and method to gain performance information. Also, the study was conducted to get significant insights in designing marketing strategy using social network services. The results are as follows. First, the result of examining audience's demographic factors depending on preferred performance type showed meaningful differences in sex, age, marital status, form of family, academic level, job and monthly income of the audiences. Second, SNS use traits of the audiences according to preferred performance genres vary in duration of SNS use, average number of access to SNS account, average time of SNS use, SNS activity type, motivation for SNS use, method to gain performance information. These findings showed that demographic characteristics and SNS use traits needed to be classified more specifically based on genres. Additionally, marketing strategy using performance information, traits of contents and customers' patterns through SNS should be specifically developed based on specified target.