• Journal of Intelligence and Information Systems
• /
• v.29 no.1
• /
• pp.249-263
• /
• 2023

Collaborative Filtering, a representative recommendation system methodology, consists of two approaches: neighbor methods and latent factor models. Among these, the latent factor model using matrix factorization decomposes the user-item interaction matrix into two lower-dimensional rectangular matrices, predicting the item's rating through the product of these matrices. Due to the factor vectors inferred from rating patterns capturing user and item characteristics, this method is superior in scalability, accuracy, and flexibility compared to neighbor-based methods. However, it has a fundamental drawback: the need to reflect the diversity of preferences of different individuals for items with no ratings. This limitation leads to repetitive and inaccurate recommendations. The Adaptive Deep Latent Factor Model (ADLFM) was developed to address this issue. This model adaptively learns the preferences for each item by using the item description, which provides a detailed summary and explanation of the item. ADLFM takes in item description as input, calculates latent vectors of the user and item, and presents a method that can reflect personal diversity using an attention score. However, due to the requirement of a dataset that includes item descriptions, the domain that can apply ADLFM is limited, resulting in generalization limitations. This study proposes a Generalized Adaptive Deep Latent Factor Recommendation Model, G-ADLFRM, to improve the limitations of ADLFM. Firstly, we use item ID, commonly used in recommendation systems, as input instead of the item description. Additionally, we apply improved deep learning model structures such as Self-Attention, Multi-head Attention, and Multi-Conv1D. We conducted experiments on various datasets with input and model structure changes. The results showed that when only the input was changed, MAE increased slightly compared to ADLFM due to accompanying information loss, resulting in decreased recommendation performance. However, the average learning speed per epoch significantly improved as the amount of information to be processed decreased. When both the input and the model structure were changed, the best-performing Multi-Conv1d structure showed similar performance to ADLFM, sufficiently counteracting the information loss caused by the input change. We conclude that G-ADLFRM is a new, lightweight, and generalizable model that maintains the performance of the existing ADLFM while enabling fast learning and inference.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.11a
• /
• pp.229-232
• /
• 2008

웹 정보량이 급속히 늘어나면서 원하는 정보를 효율적으로 찾는 검색 기술의 중요성이 커지고 있다. 검색의 정확성을 높이기 위해서는 검색 질의어와 함께 사용자의 환경, 검색 만족도와 같은 다양한 정보가 필요하다. 사용자의 명시적 피드백을 요구하는 것은 거부감을 줄 수 있으므로 사용자의 잠재적 피드백과 연관 검색어 분석을 통해 검색 질의어를 확장하는 연구가 이뤄지고 있다. 그러나 이러한 검색어 확장과 검색 정확성 사이의 상관관계에 대한 분석이 없어 연관 검색어를 정량적으로 평가할 수 없었다. 본 논문에서는 사용자가 검색 질의어를 변경하면서 검색을 반복하는 과정을 사용자의 잠재적 피드백의 하나로 보고 사용자 만족도를 반영하는 페이지 방문 시간과 함께 분석하여 연속적으로 입력된 검색어가 검색 결과 순위와 사용자 만족도에 미치는 영향을 분석하는 방법을 제안하였다. 마우스 클릭 정보 분석을 통하여 사용자의 검색 만족도를 정량화하였고 특정 주제어에서 관련 검색어가 확장되어 가는 과정은 트리 구조로 표현하였다. 이를 통해 하나의 주제어와 관련해 연속적으로 입력된 검색어 집합으로부터 연관검색어를 추출하고 검색 결과의 정확성을 높일 수 있으며 제안된 트리 구조를 다양한 방향으로 분석하여 검색어, 검색 결과, 사용자 만족도, 배경 지식 등 단순 검색어 분석에서는 나타나지 않는 다양한 정보를 얻을 수 있다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.1
• /
• pp.125-134
• /
• 2021

With the rapid development of the film industry, the number of films is significantly increasing and movie recommendation system can help user to predict the preferences of users based on their past behavior or feedback. This paper proposes a movie recommendation system based on the latent factor model with the adjustment of mean and bias in rating. Singular value decomposition is used to decompose the rating matrix and stochastic gradient descent is used to optimize the parameters for least-square loss function. And root mean square error is used to evaluate the performance of the proposed system. We implement the proposed system with Surprise package. The simulation results shows that root mean square error is 0.671 and the proposed system has good performance compared to other papers.

• Korean journal of applied entomology
• /
• v.62 no.3
• /
• pp.117-124
• /
• 2023

The management of Ricania spp. is difficult because this pest has a wide host range and diverse habitats such as agricultural, suburban, urban, and forested areas. However, the researches for Ricania spp. management have been focused on only agricultural crops. Thus, it is required to determine the arboreal host preference of Ricania spp. at the surrounding areas of the farms to increase its management efficiency. To determine its host preference at arboreal plants, we reviewed the previous studies and investigated the densities of Ricania spp. at woody plants with high ecological importance but insufficiently studied. This study identified 120 species in 53 families of arboreal hosts of Ricania spp. Only Cornus officinalis and Styrax japonicus were preferred by all developmental stages of Ricania spp. The host preference of Ricania spp. was changed according to its developmental stages. This phenomenon would be caused by that each developmental stage of Ricania spp. would prefer different parts of woody plant, and require different nutrients for its survivor and reproduction. These results of this study could be helpful to make a plan of comprehensive management strategies for Ricania spp.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.04b
• /
• pp.304-306
• /
• 2002

디지털 컨텐츠 산업의 성장, 전자상거래의 활성화, 기업의 흠페이지 활용 증가 등으로 온.오프라인에서 컨덴츠의 수요가 증가하면서 컨덴츠를 관리하는 컨텐츠 관리 시스템 시장의 성장 잠재성이 높아가고 있다. 본 논문에서는 이러한 컨텐츠 관리 시스템의 마지막 단계인 컨텐츠 배포 단계에 있어 모든 사용자에게 동일한 컨텐츠를 제공하는 것이 아니라 사용자의 관심에 따라 다른 컨텐츠를 동적으로 제공하는 컨텐츠 주천 시스템을 설계 및 구현한다. 본 시스템은 규칙 기반 추천 방식을 사용하고 있으며 규칙으로는 사용자간의 연관성을 나타내는 사용자 협업적 규칙과 항목간의 연관성을 나타내는 항목 협업적 규칙이 존재한다. 또한 컨덴츠에 대한 사용자의 선호도를 측정하기 위해서 선호범위를 정의하고 있으며 취미, 관심분야와 같이 하나 이상의 값을 가질 수 있는 다중 값을 처리하친 있다. 시스템은 추천을 위한 정보 즉, 선호범위와 사용자 프로파일 그리고 규칙들을 생성하는 오프라인 작업과 이러한 정보를 이용하여 실시간으로 사용자에게 추천해주는 온라인 작업으로 나뉘어 진다.

• Proceedings of the KAIS Fall Conference
• /
• 2003.06a
• /
• pp.222-225
• /
• 2003

웹의 급속한 성장으로 수많은 양의 정보가 매일같이 쏟아져 나오고 있다. 이는 특정 상품정보를 얻으려는 고객들에게 많은 혼란을 야기할 수 있다. 이러한 문제의 해결을 위해 추천시스템이 개발되었고, 추천 시스템은 고객들이E-Commerce 상에서 상품을 구매하는 것을 도와주기 위해서 지속적인 증가추세로 사용되고 있다. 이러한 추천시스템은 다양한 고객들의 선호도에 따라 유사성과 비유사성에 대한 정보의 기초위에서 고객들의 잠재적인 관심 항목들에 대해 개인의 취향에 맞게 추천하는 기술들을 제공한다. 그러나, 추천시스템에 많은 관심을 가짐에도 불구하고 그들의 성능에 대한 공개된 기술이나 정보는 매우 제한적이다. 본 논문에서는, 과거 고객들의 구매행동, 고객정보 데이터마이닝의 연관규칙을 이용한 E-Commerce 추천시스템을 설계하고 구현하였다.

• Journal of the Korea Society of Computer and Information
• /
• v.5 no.3
• /
• pp.201-207
• /
• 2000

The digital economic is increased electronic commerce of firms in business environment This article propose on the study of cyber shopping mall in factor of apprehension. short summaries of this study are as follows, 1) the most important of factor's name is "Advertisement factor" 2) the second factor's name is "product factor" 3) so on.(reliability, price, purchase convenience) This 5 factor's is the most important of cyber shopping mall. The result of this study, cyber shopping mall have to enforced advertisement, sales force. The advertisement increased on apprehension of customer that purchased.omer that purchased.

• Cartoon and Animation Studies
• /
• s.13
• /
• pp.211-221
• /
• 2008

Animation programs are regarded as bad genre in the aspect of advertisement revenues due to relative low viewing ratios. But programming of animation can be considered positively in the aspect of channel loyalty and preference. This study is based on the assumption that if the scheduling of animation genres could bring out the improvement of channel image of broadcasting stations in the long term, the terrestrial broadcasters could reconsider the increase of animation programs. The research questions of this study are as follows: 1. What is the relationship between the programming ratio of animation genres and the concentration of viewing patterns? 2. How has the increase of animation programming influenced the channel image of broadcasters? 3. Why is it necessary to increase of reinforce the programming of animation genre in the new media platforms? The teenagers aged from 8 to 14 can generally make a decision which program they want to watch. Likewise, the adult viewers aged from 30 to 40 can make a selection which commodity they want to buy. The results of this study indicate that the adult viewers have showed the strong preference for the same broadcasting stations which they have been exposed to in their teenages. This result implies that in the new media environment, animation genre can lead the viewers' loyally and preference for the broadcasting channel for a long period.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.6D
• /
• pp.791-799
• /
• 2008

To build the system which has high utilization and usefulness for users, it is necessary to know the information type and use-demand that the use want. The purpose of this study is to forecast the preference and demand of utilization for bus information when bus information is offered through cellular phon. The accomplishments of this research are as follow : Firstly, importance on the level of individual factor and the value of change's figure can be evaluated, using preference analysis on bus information by conjoint analysis. Secondly, by establishing the use-demand model bus information using binary logit model, influence factor on whether or not the use of the user. Finally, ordered probit model was built by use behavior model in payment per call or per month of potential user of bus information. Through call times and sensitive analysis by payment methods, elasticity point, optimal payment fee, and use probability was analyzed. This study make application as basic to efficient bus information policy and to improve use rate of bus information in future because this study make it possible to get preference analysis, use-demand analysis and estimation of optimal payment fee which is reflecting various requirement in use of bus information user.

• Journal of Korean Elementary Science Education
• /
• v.41 no.1
• /
• pp.65-78
• /
• 2022

This study aims to 1) verify the validity of the Science self-efficacy scale and 2) perform a latent mean analysis of the background variables about a pre-service teacher. The study uses pre-tests to analyze data from 187 pre-service teachers, which uses Tark's Science self-efficacy scale (2011). To identify the factor structure, exploratory factor analysis was performed. Based on the results of the pilot test, the expert group council revised the scale for the pre-service teachers to respond to the 3-factor structure. In the main test, 354 data were analyzed through a modified Science self-efficacy scale, and exploratory and confirmatory factor analyses were performed. The results of the study are as follows: First, in the pilot test, the pre-service teacher responded to a 3-factor instrument, but the validity of two items was examined further below. Second, the pre-service teachers responded to a 3-factor instrument on 29 items for the modified Science self-efficacy scale. The total reliability of the instrument was .886 and the reliability of each factor was analyzed as .882-.886. Finally, the latent mean analysis by gender showed that females have a higher self-regulation efficacy factor and males have a higher self-confidence factor (Cohen's d > .3). Furthermore, there is a significant difference in task difficulty preference and self-regulatory efficacy factor (Cohen's d > .8) between the college preparatory and science subject preference. This study provides important insights into and contributions to the accurate scientific self-efficacy diagnosis of pre-service teachers, as well as proposes a curriculum to improve the scientific self-efficacy of prospective teachers.