• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2000.04a
• /
• pp.120-125
• /
• 2000

We propose a semi-automatic and dynamic indexing algorithm for multi-media database(e.g. movie files, audio files), which are difficult to create indexes expressing their emotional or abstract contents, according to user's sensitivity by using user's histories of access to database. In this algorithm, we simply categorize data at first, create a vector space of each user's interest(user model) from the history of which categories the data belong to, and create vector space of each data(title model) from the history of which users the data had been accessed from. By continuing the above method, we could create suitable indexes, which show emotional content of each data. In this paper, we define the recurrence formulas based on the proposed algorithm. We also show the effectiveness of the algorithm by simulation result.

• Proceedings of the KSRS Conference
• /
• 2004.10a
• /
• pp.128-131
• /
• 2004

In this paper, we describe reconstructed 3D-urban modeling techniques for laser scanner and CCD camera system, which are loading on the vehicle. We use two laser scanners, the one is horizon scanner and the other is vertical scanner. Horizon scanner acquires the horizon data of building for localization. Vertical scan data are main information for constructing a building. We compared extraction of edge aerial image with laser scan data. This method is able to correct the cumulative error of self-localization. Then we remove obstacles of 3D-reconstructed building. Real-texture information that is acquired with CCD camera is mapped by 3D-depth information. 3D building of urban is reconstructed to 3D-virtual world. These techniques apply to city plan. 3D-environment game. movie background. unmanned-patrol etc.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.711-712
• /
• 2008

We developed a XMAS (X Multimedia Analysis System) for analyzing the illegal and objectionable multimedia in Internet environment based on Web2.0. XMAS uses the MPEG-7 visual descriptor and multi-class SVM (support vector machine) and its performance (accuracy on precision) is about 91.6% for objectionable multimedia analysis and 99.9% for illegal movie retrieval.

• Journal of Information Science Theory and Practice
• /
• v.1 no.1
• /
• pp.10-26
• /
• 2013

Domain transfer is a widely recognized problem for machine learning algorithms because models built upon one data domain generally do not perform well in another data domain. This is especially a challenge for tasks such as opinion classification, which often has to deal with insufficient quantities of labeled data. This study investigates the feasibility of self-training in dealing with the domain transfer problem in opinion classification via leveraging labeled data in non-target data domain(s) and unlabeled data in the target-domain. Specifically, self-training is evaluated for effectiveness in sparse data situations and feasibility for domain adaptation in opinion classification. Three types of Web content are tested: edited news articles, semi-structured movie reviews, and the informal and unstructured content of the blogosphere. Findings of this study suggest that, when there are limited labeled data, self-training is a promising approach for opinion classification, although the contributions vary across data domains. Significant improvement was demonstrated for the most challenging data domain-the blogosphere-when a domain transfer-based self-training strategy was implemented.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2017.10a
• /
• pp.151-153
• /
• 2017

In recommendation system, collaborative filtering is the most important algorithm. Collaborative filtering is a method of making automatic predictions about the interests of a user by collecting preferences or taste information from many users. In this paper five algorithms were used. Metrics such as Recall-Precision, FPR-TPR,RMSE, MSE, MAE were calculated. From the result of the experiment, the user-based collaborative filtering was the best approach to recommend movies to users.

• International Journal of Internet, Broadcasting and Communication
• /
• v.11 no.3
• /
• pp.71-76
• /
• 2019

In this paper, we propose a simple one-click marketing solution for mobile devices which can advertise a product which is embedded in a mobile video while watching the video on a smartphone. If a specific product of interest appears in the video to the user, one can simply click on the product in the video and a pop-up window with information about the product is proposed. The implementation of the system is expected to enable users to gain real-time information about the product while watching the video without having to search for the product again after watching the movie, and thereby facilitating more mobile commerce. We use a two-fold system to prevent the failure of tracking which often occurs on a single online tracking system, so that the user cannot always get the commercial product information.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.11a
• /
• pp.1176-1178
• /
• 2013

협업 필터링(Collaborative Filtering)이란 많은 사용자들로부터 얻은 기호정보(taste information)에 따라 사용자들의 관심사들을 자동적으로 예측하여, 아이템에 대한 목표 사용자의 선호도와 다른 사용자의 선호도를 비교 분석하여 목표 사용자가 좋아할 만한 아이템을 추천하는 기법이다. 그러나 협업 필터링 기법은 고객 정보와 평가 정보가 충분히 많아야 정확성이 높은 추천 결과가 나타난다. 본 논문에서는 영화를 한 번도 평가하지 않은 사용자들에게 영화를 추천 해주기 위한 즉, 협업 필터링의 희박성 문제(Sparsity Problem)를 해결하기 위한 한 가지 방법으로 개인의 감정 정보를 이용하여 문제를 해결하는 방법을 소개한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.366-369
• /
• 2022

The glut of information aggravated the process of data analysis and other procedures including data mining. Many algorithms were devised in Big Data and Data Mining to solve such an intricate problem. In this paper, we conducted research about the comparison of several similarity measures and community detection algorithms in collaborative filtering for movie recommendation systems. Movielense data set was used to do an empirical experiment. We applied three different similarity measures: Cosine, Euclidean, and Pearson. Moreover, betweenness and eigenvector centrality were used to detect communities from the network. As a result, we elucidated which algorithm is more suitable than its counterpart in terms of recommendation accuracy.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.261-262
• /
• 2023

본 연구에서는 감성 분석 기반의 키워드 도출형 영화 리뷰 웹사이트를 개발하였다. 사용자들은 영화에 대한 리뷰를 작성할 때, 자동으로 키워드를 추출하는 기능을 활용하여 다양하면서도 빠르게 정보를 얻을 수 있다. 사용자가 작성한 리뷰를 시스템에 입력하면, 내부적으로 ChatGPT를 활용하여 텍스트를 분석하고 키워드를 추출한다. 이를 통해 사용자는 별다른 노력 없이도 키워드를 통해 영화의 장르, 감독, 배우, 플롯 요소 등 다양한 정보를 빠르게 확인할 수 있다. 추출된 키워드는 저장되어 시각화에 활용되며, 사용자들은 리뷰에 대한 원하는 정보를 쉽게 얻을 수 있다. 개발된 키워드 도출형 영화 리뷰 웹사이트는 사용자들에게 빠르고 다양한 정보를 제공하며, 영화 관련 결정을 내리는 데에 도움을 줄 것으로 기대된다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.11a
• /
• pp.82-85
• /
• 2008

고객의 기호와 요구에 부응하는 서비스의 제공을 위해 영화 요소 중 정확한 장르의 분류는 고객의 선택에 있어 중요한 문제이다. 기존의 수작업에 의한 장르 분류는 시간과 비용, 신뢰성 등에서 비효율적이다. 이러한 문제의 해결을 위해 영화 시놉시스(Synopsis) 기반의 기계학습 방법은 효율적인 대안이 될 수 있다. 본 논문에서는 대다수 영화서비스 주체가 보유하고 있는 시놉시스 정보를 기반으로 하여 기계학습을 이용한 영화장르 분류에 관한 하나의 정형화된 방법을 제시한다. 기계학습 Algorithm 중 LibSVM, RandomComittee, LMT, NaiveBayes, PART Algorithm 을 이용하여 Algorithm 별, 장르별 분류 정확도를 측정하여 비교한다.