• Journal of Korea Society of Digital Industry and Information Management
• /
• v.19 no.3
• /
• pp.221-232
• /
• 2023

In this paper, we propose a recommendation system based on the latent factor model using matrix factorization, which is one of the most commonly used collaborative filtering algorithms for recommendation systems. In particular, by introducing the concept of creating a list of recommended content and a list of non-preferred recommended content, and removing the non-preferred recommended content from the list of recommended content, we propose a method to ultimately increase the satisfaction. The experiment confirmed that using a separate list of non-preferred content to find non-preferred content increased precision by 135%, accuracy by 149%, and F1 score by 72% compared to using the existing recommendation list. In addition, assuming that users do not view non-preferred content through the proposed algorithm, the average evaluation score of a specific user used in the experiment increased by about 35%, from 2.55 to 3.44, thereby increasing user satisfaction. It has been confirmed that this algorithm is more effective than the algorithms used in existing recommendation systems.

• Journal of Intelligence and Information Systems
• /
• v.29 no.3
• /
• pp.145-165
• /
• 2023

Recommendation system plays a significant role on relieving difficulties of selecting information among rapidly increasing amount of information caused by the development of the Internet and on efficiently displaying information that fits individual personal interest. In particular, without the help of recommendation system, E-commerce and OTT companies cannot overcome the long-tail phenomenon, a phenomenon in which only popular products are consumed, as the number of products and contents are rapidly increasing. Therefore, the research on recommendation systems is being actively conducted to overcome the phenomenon and to provide information or contents that are aligned with users' individual interests, in order to induce customers to consume various products or contents. Usually, collaborative filtering which utilizes users' historical behavioral data shows better performance than contents-based filtering which utilizes users' preferred contents. However, collaborative filtering can suffer from cold-start problem which occurs when there is lack of users' historical behavioral data. In this paper, hybrid music recommendation system, which can solve cold-start problem, is proposed based on the playlist data of Melon music streaming service that is given by Kakao Arena for music playlist continuation competition. The goal of this research is to use music tracks, that are included in the playlists, and metadata of music tracks and playlists in order to predict other music tracks when the half or whole of the tracks are masked. Therefore, two different recommendation procedures were conducted depending on the two different situations. When music tracks are included in the playlist, LightFM is used in order to utilize the music track list of the playlists and metadata of each music tracks. Then, the result of Item2Vec model, which uses vector embeddings of music tracks, tags and titles for recommendation, is combined with the result of LightFM model to create final recommendation list. When there are no music tracks available in the playlists but only playlists' tags and titles are available, recommendation was made by finding similar playlists based on playlists vectors which was made by the aggregation of FastText pre-trained embedding vectors of tags and titles of each playlists. As a result, not only cold-start problem can be resolved, but also achieved better performance than ALS, BPR and Item2Vec by using the metadata of both music tracks and playlists. In addition, it was found that the LightFM model, which uses only artist information as an item feature, shows the best performance compared to other LightFM models which use other item features of music tracks.

• The Journal of the Korea Contents Association
• /
• v.22 no.1
• /
• pp.138-147
• /
• 2022

Recently, chatbots have been applied in various fields and have shown good results, and many attempts to use chatbots in shopping mall product recommendation services are being conducted on e-commerce platforms. In this paper, for a conversation system that recommends a fashion that a user wants based on conversation between the user and the system and fashion image information, a transformer model that is currently performing well in various AI fields such as natural language processing, voice recognition, and image recognition. We propose a multimodal-based improved transformer model that is improved to increase the accuracy of recommendation by using dialogue (text) and fashion (image) information together for data preprocessing and data representation. We also propose a method to improve accuracy through data improvement by analyzing the data. The proposed system has a recommendation accuracy score of 0.6563 WKT (Weighted Kendall's tau), which significantly improved the existing system's 0.3372 WKT by 0.3191 WKT or more.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.01a
• /
• pp.23-26
• /
• 2022

OTT 서비스의 이용자가 폭발적으로 증가하고 있는 지금, 사용자에게 맞춤형 상품을 추천하는 것은 해당 서비스에서 중요한 사안이다. 본 논문에서는 콘텐츠 기반 추천 시스템의 모델을 제안하고, 영화 데이터를 추가 해가며 예측력을 높일 최종적인 모델을 채택하고자 한다. 이를 위해 GroupLens와 Kaggle에서 영화 데이터를 수집하고 총 1111개의 영화, 943명의 사용자에게서 나온 71026개의 영화 평가 데이터를 이용하였다. 모델 평가 결과, 장르와 키워드만을 이용한 추천 시스템 모델의 RMSE는 1.3076, 단계적으로 데이터를 추가해 최종적으로 장르, 키워드, 배우, 감독, 나라, 제작사를 이용한 추천 시스템 모델의 RMSE는 1.1870으로 모든 데이터를 추가한 모델의 예측력이 더 높았다. 이에 따라 장르, 키워드, 배우, 감독, 나라, 제작사를 이용해 구현한 모델을 최종적인 모델로 채택, 무작위로 추출한 한 명의 사용자에 대한 영화 추천 리스트를 뽑아낸다.

• The Journal of Korean Association of Computer Education
• /
• v.6 no.2
• /
• pp.147-156
• /
• 2003

Collaborative Filtering is a popular technology in electronic commerce, which adapt the opinions of entire communities to provide interesting products or personalized resources and items. It has been applied to many kinds of electronic commerce domain since Collaborative Filtering has proven an accurate and reliable tool. But educational application remain limited yet. We design collaborative filtering recommendation system using user's ratings in educational contents recommendation. Also We propose a method of similarity compensation using user's information for improvement of recommendation accuracy. The proposed method is more efficient than the traditional collaborative filtering method by experimental comparisons of mean absolute error(MAE) and reciever operating characteristics(ROC) values.

• Journal of Digital Convergence
• /
• v.17 no.11
• /
• pp.201-206
• /
• 2019

While many users searched for insurance information online, there were not many cases of contents recommendation researches on insurance companies' websites. Therefore, this study proposed a page recommendation system with high possibility of preference to users by utilizing page visit history of insurance companies' websites. Data was collected by using client-side storage that occurs when using a web browser. Collaborative filtering was applied to research as a recommendation technique. As a result of experiment, we showed good performance in item-based collaborative (IBCF) based on Jaccard index using binary data which means visit or not. In the future, it will be possible to implement a content recommendation system that matches the marketing strategy when used in a company by studying recommendation technology that weights items.

• Journal of Digital Contents Society
• /
• v.13 no.4
• /
• pp.559-565
• /
• 2012

The remarkable development of IT is contributed to popularization of smart phones, which in turn creates a new domain called app store. Smartphone apps have grown fast because they can be easily purchased through an app store. As the volume of apps traded in app stores is so huge that it is extremely hard for users to find the exact app they want. In general, an app store recommends an app to users based on the search words they entered. In terms of recommendation of app, this kind of content-based method is not effective. To increase accuracy in recommending app, this paper proposes a characteristic similarity-based app recommendation method. This method creates attributes on the app based on the related information such as genre, functionality and number of downloads and then compares them with the propensity to use the app. According to diverse simulations, the method proposed in this paper improved the performance of app recommendation by 33% in average, compared to the conventional method.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.7
• /
• pp.1083-1091
• /
• 2017

As big data technologies have been developed and massive data have exploded from users through various channels, CEO of global IT enterprise mentioned core importance of data in next generation business. Therefore various machine learning technologies have been necessary to apply data driven services but especially recommendation has been core technique in viewpoint of directly providing summarized information or exact choice of items to users in information flooding environment. Recently evolved recommendation techniques have been proposed by many researchers and most of service companies with big data tried to apply refined recommendation method on their online business. For example, Amazon used item to item collaborative filtering method on its sales distribution platform. In this paper, we develop a commercial web service for suggesting music contents and implement three representative collaborative filtering methods on the service. We also produce recommendation lists with three methods based on real world sample data and evaluate the usefulness of them by comparison among the produced result. This study is meaningful in terms of suggesting the right direction and practicality when companies and developers want to develop web services by applying big data based recommendation techniques in practical environment.

• Journal of the Korean Society for information Management
• /
• v.23 no.3 s.61
• /
• pp.91-125
• /
• 2006

Recent advancements in information technology and the Internet have caused an explosive increase in the information available and the means to distribute it. However, such information overflow has made the efficient and accurate search of information a difficulty for most users. To solve this problem, an information retrieval and filtering system was developed as an important tool for users. Libraries and information centers have been in the forefront to provide customized services to satisfy the user's information needs under the changing information environment of today. The aim of this study is to propose an efficient information service for libraries and information centers to provide a personalized recommendation system to the user. The proposed method overcomes the weaknesses of existing systems, by providing a personalized hybrid recommendation method for multimedia contents that works in a large-scaled data and user environment. The system based on the proposed hybrid method uses an effective framework to combine Association Rule with Collaborative Filtering Method.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.3
• /
• pp.99-105
• /
• 2021

In addition to the OTT video production service represented by Nexflix and YouTube, a personalized recommendation system for content with artificial intelligence has become common. YouTube's personalized recommendation service system consists of two neural networks, one neural network consisting of a recommendation candidate generation model and the other consisting of a ranking network. Netflix's video recommendation system consists of two data classification systems, divided into content-based filtering and collaborative filtering. As the online platform-led content production is activated by the Corona Pandemic, the field of virtual influencers using artificial intelligence is emerging. Virtual influencers are produced with GAN (Generative Adversarial Networks) artificial intelligence, and are unsupervised learning algorithms in which two opposing systems compete with each other. This study also researched the possibility of developing AI platform based on individual recommendation and virtual influencer (metabus) as a core content of OTT in the future.