• Journal of KIISE
• /
• v.44 no.6
• /
• pp.621-627
• /
• 2017

Collaborative filtering is a technique that effectively recommends unrated items for users. Collaborative filtering is based on the similarity of the items evaluated by users. The existing top-N recommendation methods are based on pair-wise and list-wise preference models. However, these methods do not effectively represent the relative preference of items that are evaluated by users, and can not reflect the importance of each item. In this paper, we propose a new method to represent user's latent preference by combining an existing preference model and the notion of inverse user frequency. The proposed method improves the accuracy of existing methods by up to two times.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.11a
• /
• pp.753-756
• /
• 2004

유비쿼터스 컴퓨팅 환경에서 추천시스템은 무수히 많은 정보들에 대하여 사람들이 적절한 선택을 할 수 있도록 도와준다. 사용자에게 필요한 정보를 찾아주고, 정보들의 우선순위를 결정해주는 추천시스템에 있어서 사용자의 위치는 보다 가치있는 정보를 제공할 수 있는 도구가 된다. 위치기반 추천시스템은 사용자가 아이템들로부터 얼마나 멀리 떨어져있는가를 고려하여 상위 리스트들을 제공할 수 있어야 한다. 하지만 일반적인 추천시스템에서 주로 사용되고 있는 기존의 사용자 기반 협업필터링 기법은 사용자의 자발적인 정보 입력에 의존함으로써 일정한 수의 사용자 정보가 축적되어 있지 않으면 정확한 추천이 불가능한 단점이 있다. 본 논문에서는 아이템에 기반한 협업 필터링 기법을 확률적으로 분석하고, 아이템의 위치에따라 랭킹을 부여하는 방법과 사용자의 위치정보를 추천알고리즘에 적용시켜 보다 정확하고 효율적인 추천방법을 제안하였다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.3
• /
• pp.543-551
• /
• 2015

As the use of internet and mobile devices became generalized, users utilizing search and recommendation in order to find the information they want in the midst of various websites have become common. In order to recommend more appropriate item for users, this paper proposes a recommendation technique that reflects the users' preference change following the flow of time by applying users' activity and time information. The proposed technique, after classifying the data in categories including the tag information that is considered at the time of choosing the items, only uses the data that users' preference change following the flow of time is reflected. For the users who prefer the corresponding category, the item that is extracted by applying tag information to collaboration filtering technique is recommended and for general users, items are recommended based on the ranking calculated by using the tag information. The proposed technique was experimented by using hetrec2011-movielens-2k data set. The experiment result indicated that the proposed technique has been more enhanced the accuracy, appropriacy, compared to item-based, user-based method.

• Journal of Digital Convergence
• /
• v.11 no.2
• /
• pp.235-242
• /
• 2013

Reputation systems aggregate users' feedback after the completion of a transaction and compute the "reputation" of products, services, or providers, which can assist other users in decision-making in the future. With the rapid growth of online e-Learning content providing services, a suitable reputation system for more credible e-Learning content delivery has become important and is essential if educational content providers are to remain competitive. Most existing reputation systems focus on generating ratings only for user reputation; they fail to consider the reputations of products or services(item reputation). However, it is essential for B2C e-Learning services to have a reliable reputation rating mechanism for items since they offer guidance for decision-making by presenting the ranks or ratings of e-Learning content items. To overcome this problem, we propose a novel collaborative filtering based reputation rating method. Collaborative filtering, one of the most successful recommendation methods, can be used to improve a reputation system. In this method, dual information sources are formed with groups of co-oriented users and expert users and to adapt it to the reputation rating mechanism. We have evaluated its performance experimentally by comparing various reputation systems.

• Journal of Broadcast Engineering
• /
• v.14 no.2
• /
• pp.238-252
• /
• 2009

Due to the rapid increase of available contents via the convergence of broadcasting and internet, the efficient access to personally preferred contents has become an important issue. In this paper, for recommendation scheme for TV programs using a collaborative filtering technique is studied. For recommendation of user preferred TV programs, our proposed recommendation scheme consists of offline and online computation. About offline computation, we propose reasoning implicitly each user's preference in TV programs in terms of program contents, genres and channels, and propose clustering users based on each user's preferences in terms of genres and channels by dynamic fuzzy clustering method. After an active user logs in, to recommend TV programs to the user with high accuracy, the online computation includes pulling similar users to an active user by similarity measure based on the standard preference list of active user and filtering-out of the watched TV programs of the similar users, which do not exist in EPG and ranking of the remaining TV programs by proposed rank model. Especially, in this paper, the BM (Best Match) algorithm is extended to make the recommended TV programs be ranked by taking into account user's preferences. The experimental results show that the proposed scheme with the extended BM model yields 62.1% of prediction accuracy in top five recommendations for the TV watching history of 2,441 people.

• 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.

• Journal of Korea Multimedia Society
• /
• v.9 no.2
• /
• pp.189-196
• /
• 2006

This paper proposes a priority based AOI management model as a method to effectively process message traffic for collaboration in virtual environments. Where message traffic reaches the maximum capacity of the message replay server in DVE, some of the events might be delayed to be updated at the remote clients. Because existing AOI models depend only on the occurrence time of events, they have a problem that more important events in the context might be processed later than less serious ones. Close Interactions with other participants are relatively more important than simple events which happen at a distance from the participant. This study classifies the priorities of events according to the degree of interest, and proposes a method to reduce processing delay time of events highly interested by a user. The proposed model offers a way to effectively utilize limited capacity of a server using a priority queue mechanism, which is able to handle different kinds of events. To prevent from starvation of simple events and to give fairness to the proposed algorithm, event occurrence time is also considered as well as degree of interest when processing events.