• Journal of Korea Multimedia Society
• /
• v.24 no.12
• /
• pp.1673-1681
• /
• 2021

Item cold start is a well studied problem in the research field of recommender systems. Still, many existing collaborative filters cannot recommend items accurately when only a few user-item interaction data are available for newly introduced items (Cold items). We propose a interaction feature prediction method to mitigate item cold start problem. The proposed method predicts the interaction features that collaborative filters can calculate for the cold items. For prediction, in addition to content features of the cold-items used by state-of-the-art methods, our method exploits the interaction features of k-nearest content neighbors of the cold-items. An attention network is adopted to extract appropriate information from the interaction features of the neighbors by examining the contents feature similarity between the cold-item and its neighbors. Our evaluation on a real dataset CiteULike shows that the proposed method outperforms state-of-the-art methods 0.027 in Recall@20 metric and 0.023 in NDCG@20 metric.

• Journal of Information Technology Applications and Management
• /
• v.20 no.3_spc
• /
• pp.219-230
• /
• 2013

Real time accessiblity and agility in Ubiquitous-commerce is required under ubiquitous computing environment. The Research has been actively processed in e-commerce so as to improve the accuracy of recommendation. Existing Collaborative filtering (CF) can not reflect contents of the items and has the problem of the process of selection in the neighborhood user group and the problems of sparsity and scalability as well. Although a system has been practically used to improve these defects, it still does not reflect attributes of the item. In this paper, to solve this problem, We can use a implicit method which is used by customer's data and purchase history data. We propose a new clustering method of weighted preference for customer using k-means clustering and Bayesian network in order to improve the accuracy of recommendation. To verify improved performance of the proposed system, we make experiments with dataset collected in a cosmetic internet shopping mall.

• The KIPS Transactions:PartB
• /
• v.19B no.3
• /
• pp.165-176
• /
• 2012

In recent years, due to rapidly increasing user-created internet contents coupled with the development of community-based websites, the internet resource recommendation systems are attracting attentions of the users. However, most of the systems have failed in properly reflecting users' characteristics and thus they have difficulty in recommending appropriate resources to users. In this paper, we propose an internet resource recommendation method using FOAF and SNA which fully reflects the characteristics of users. In our method, 1) we extract the data about user characteristics and tags using FOAF; 2) we generate graphs representing users, user characteristics and tags after inserting data into 3 matrixes and integrating them; 3) we recommend the appropriate internet resources after selecting common characteristics of the recommended items and Hot tags by analyzing social network. For verification of our proposed method, we implemented our method to establish and analyze an experimental social group. We verified through our experiments that the more users added in the social network, the higher quality of recommendation result we got than the item-based recommendation method. By using the suggested idea in this paper, we can make a more appropriate recommendation of resources to users while effectively retrieving explosively increasing internet resources.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.01a
• /
• pp.437-438
• /
• 2019

스마트폰을 중심으로 한 모바일 기기의 보급과 온라인 소셜 네트워크 서비스의 이용자들이 증가하면서 사용자들은 많은 콘텐츠를 소비하고 공유한다. 이는 콘텐츠 사용자들의 개별적 기호에 맞지 않거나 만족도가 떨어지는 콘텐츠를 소비하게 한다. 이와 같은 문제를 해결하기 위해 소셜 네트워크 사용자에게 적합한 콘텐츠를 추천하기 위한 기법에 대한 연구가 활발하게 진행되고 있다. 본 논문에서는 온라인 상에 존재하는 다양한 정보 중에서 공연과 관련한 콘텐츠들을 중심으로 사용자 성향별로 추천을 해줄 수 있는 협업필터링 방법에 대하여 제안한다.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2004.10a
• /
• pp.430-460
• /
• 2004

This study presented a personalization system that adopted a methodology which is applicable for digital content recommendation and executed by the Internet service providers. The system made a recommendation to the users on the basis of their preferences, while most techniques for recommending digital content have focused on considering the similarity of content. In addition, it developed a method of evaluation to determine the priority of recommendations and adopted measures when selecting a set of recommendations. To experiment the feasibility and effectiveness of the presented methodology, a prototype system was developed and was applied to an English newspaper on the Internet.

• Journal of Information Technology and Architecture
• /
• v.9 no.1
• /
• pp.57-66
• /
• 2012

Recently, with a rapidly growing of the mobile content market, a variety of mobile-based applications are being launched. But mobile devices, compared to the average computer, take a lot of effort and time to get the final contents you want to use due to the restrictions such as screen size and input methods. To solve this inconvenience, a recommender system is required, which provides customized information that users prefer by filtering and forecasting the information.In this study, an tailored multi-information recommendation system utilizing a Personalized information recommendation system on smartphone is proposed. Filtering of information is to predict and recommend the information the individual would prefer to by using the user-based collaborative filtering. At this time, the degree of similarity used for the user-based collaborative filtering process is Euclidean distance method using the Pearson's correlation coefficient as weight value.As a real applying case to evaluate the performance of the recommender system, the scenarios showing the usefulness of recommendation service for the actual restaurant is shown. Through the comparison experiment the augmented reality based multi-recommendation services to the existing single recommendation service, the usefulness of the recommendation services in this study is verified.

• The Journal of Society for e-Business Studies
• /
• v.21 no.2
• /
• pp.1-22
• /
• 2016

News contents has been collected, selected, edited and sometimes distorted by the news recommendation mechanisms of online portals in nowadays. Prior studies had not confirmed the consensus of newsworthiness, and they had not tried to empirically validate the impacts of newsworthiness on audience reactions. This study challenged to summarize the concepts of newsworthiness and validate the impact of representativeness of both editor's and audience's perspective on audience reactions as perceived news quality, trust on news portal, perceived usefulness, service satisfaction, loyalty, continuous usage intention, and word-of-mouth intention by adopting the representativeness heuristics method and information adoption model. 357 valid data had been collected using a scenario survey method. Subjects in each groups are exposed by 3 news recommendation mechanisms: 1) the time-priority news exposure mechanism (control group), 2) the reference-score-based news recommendation mechanism (a single treatment group), and 3) the major-news-priority exposure mechanism sorting by the reference scores made by peer audiences (the mixed treatment group). Data had been analyzed by the MANOVA and PLS method. MANOVA results indicate that only mixed method of both editor and audience recommendation mechanisms impacts on perceived news quality and trust. PLS results indicate that perceived news quality and trust could significantly affect on the perceived usefulness, service satisfaction, loyalty, continuance usage, and word-of-mouth intention. This study would contributions to empathize the role of information technology in media industry, to conceptualize the news value in the balanced views of both editors and audiences, and to empirically validate the benefits of news recommendation mechanisms in academy. For practice, the results of this study suggest that online news portals would be better to make mixed news recommendation mechanisms to attract audiences.

• Journal of Digital Contents Society
• /
• v.19 no.4
• /
• pp.701-709
• /
• 2018

Our research team implemented and operated the system by analyzing the membership information and identifying the different preferences for each group and providing the results of the recommendation based on accumulated membership information and activity log data to the individual. The utilization log was followed up. We analyzed how many people use recommended services and analyzed whether there are any factors other than the personalization service algorithm that affect the service utilization of the system with personalization. In addition, we propose recommendation methods based on behavioral changes when incentives are given through analyzing patterns of users' usage according to methods of recommending services and contents that are often used based on analysis contents.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.613-614
• /
• 2008

A content recommendation according to users preferences comes up in the Internet application due to contents overwhelming. This paper newly proposes a method to predict contents preference using tags in conjunction with Content-Based Filtering. By implementing this method, this paper cleans up the contents sparsity problem in Content-Based Filtering, and shows the outstanding improvements.

• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.27-40
• /
• 2023

With the increase in the spread of smart devices and the impact of COVID-19, the consumption of media contents through smart devices has significantly increased. Along with this trend, the amount of media contents viewed through OTT platforms is increasing, that makes contents recommendations on these platforms more important. Previous contents-based recommendation researches have mostly utilized metadata that describes the characteristics of the contents, with a shortage of researches that utilize the contents' own descriptive metadata. In this paper, various text data including titles and synopses that describe the contents were used to recommend similar contents. KLUE-RoBERTa-large, a Korean language model with excellent performance, was used to train the model on the text data. A dataset of over 20,000 contents metadata including titles, synopses, composite genres, directors, actors, and hash tags information was used as training data. To enter the various text features into the language model, the features were concatenated using special tokens that indicate each feature. The test set was designed to promote the relative and objective nature of the model's similarity classification ability by using the three contents comparison method and applying multiple inspections to label the test set. Genres classification and hash tag classification prediction tasks were used to fine-tune the embeddings for the contents meta text data. As a result, the hash tag classification model showed an accuracy of over 90% based on the similarity test set, which was more than 9% better than the baseline language model. Through hash tag classification training, it was found that the language model's ability to classify similar contents was improved, which demonstrated the value of using a language model for the contents-based filtering.