• Journal of the Korea Society of Computer and Information
• /
• v.27 no.9
• /
• pp.77-89
• /
• 2022

In this paper, we propose a system which automatically generates SHACL schemas to describe and validate RDF knowledge graphs constructed by RML mappings. Unlike existing studies, the proposed system generates the schemas based on not only RML mapping rules but also metadata extracted from RML mapping input data in various formats such as CSV, JSON, XML or databases. Therefore, our schemas include the constraints on data type, string length, value range and cardinality, which were not present in the existing schemas. And we solves the problem with "repeated properties" which overlooked in existing studies. Through a conformance test consisting of 297 cases, we show that the proposed system generates correct constraints for the graphs. The proposed system can contribute to automation of the tedious and error-prone existing manual validation processes.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.660-662
• /
• 2022

Due to the prolonged aftermath of COVID-19, the number of modern people suffering from stress and mental illness is increasing. Recently, various methods of color therapy are being studied using LED lighting to improve concentration, relieve stress, manage skin, and improve sleep quality. In this paper, pulse waves are measured and heart rate variability is extracted using a PPG (Photoplethysmogram) sensor to analyze a person's mental and physical state. Using RGBLED and Arduino, we propose a mood lighting system that automatically changes colors according to changes in mental and physical conditions by producing color lighting of various wavelengths, which are mainly used to stabilize the mental state.

• Annual Conference of KIPS
• /
• 2011.11a
• /
• pp.357-360
• /
• 2011

모자이크는 여러 개의 작은 영상들을 모아서 하나의 큰 영상을 만드는 것을 말한다. 본 논문에서는 모자이크 방법 중 하나인 사진을 이용하여 영상을 만드는 포토 모자이크 방법을 컴퓨터 알고리즘으로 구현하는 기술 제안을 한다. 이미지를 원하는 사이즈의 타일로 나눈 다음, 나눠진 타일을 16등분을 한다. 16등분된 이미지 각 요소들에 대하여 RGB 평균값을 계산하여 총 48개 특징값을 추출하여 데이터베이스에 저장해둔다. 그리고 타일과 비교가 될 이미지들은 이미 똑같은 작업을 통하여 데이터베이스에 저장이 되어 있다. 이렇게 저장된 값들을 통하여 유클리드 거리를 통하여 두 이미지의 유사도를 측정하게 된다. 최적의 값을 찾으면 바로 대입하는 것이 아니라 이전에 있던 타일 이미지의 명암값을 새로 삽입되는 타일에 부여를 하여, 부드러운 영상을 만들게 된다. 그리고 타일을 삽입할 때 이전에 사용된 이미지는 배열의 마지막으로 옮겨지게 되며 사용횟수를 체크하여 반복적 사용을 제한하였다.

• Annual Conference of KIPS
• /
• 2020.11a
• /
• pp.644-647
• /
• 2020

최근 교육분야에서는 IT 기술을 활용하여 교육을 혁신하는 것을 의미하는 에듀테크에 대한 관심이 높아지고 있다. 단순한 지식의 전달이 아닌 사용자의 수준에 맞춰진 학습을 하고 자신의 학습 내용을 스스로 모니터링할 수 있는 새로운 교육시스템이 필요하다. 이에 본 논문에서는 빅데이터를 활용한 영어학습 애플리케이션를 제안한다. 제안하는 애플리케이션은 영어뉴스 기사에서 추출한 빅데이터를 활용하여 사용자 수준에 맞춘 유용한 문장을 분석해 자동으로 문제를 생성하고 사용자의 음성데이터를 강세 분석 알고리즘으로 원어민 발음과 비교분석 하여 발음 및 강세를 교정할 수 있도록 설계 및 구현하였다.

• Annual Conference of KIPS
• /
• 2023.11a
• /
• pp.514-517
• /
• 2023

본 연구는 인 컨택스트 러닝 (In-Context Learning)을 오디오-언어 작업에 적용하기 위한 멀티모달 (Multi-Modal) 딥러닝 모델을 다룬다. 해당 모델을 통해 학습 단계에서 오디오와 텍스트의 소통 가능한 형태의 표현 (Representation)을 학습하고 여러가지 오디오-텍스트 작업을 수행할 수 있는 멀티모달 딥러닝 모델을 개발하는 것이 본 연구의 목적이다. 모델은 오디오 인코더와 언어 인코더가 연결된 구조를 가지고 있으며, 언어 모델은 6.7B, 30B 의 파라미터 수를 가진 자동회귀 (Autoregressive) 대형 언어 모델 (Large Language Model)을 사용한다 오디오 인코더는 자기지도학습 (Self-Supervised Learning)을 기반으로 사전학습 된 오디오 특징 추출 모델이다. 언어모델이 상대적으로 대용량이기 언어모델의 파라미터를 고정하고 오디오 인코더의 파라미터만 업데이트하는 프로즌 (Frozen) 방법으로 학습한다. 학습을 위한 과제는 음성인식 (Automatic Speech Recognition)과 요약 (Abstractive Summarization) 이다. 학습을 마친 후 질의응답 (Question Answering) 작업으로 테스트를 진행했다. 그 결과, 정답 문장을 생성하기 위해서는 추가적인 학습이 필요한 것으로 보였으나, 음성인식으로 사전학습 한 모델의 경우 정답과 유사한 키워드를 사용하는 문법적으로 올바른 문장을 생성함을 확인했다.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.881-891
• /
• 2023

Recently, various studies on malicious URL detection using artificial intelligence have been conducted, and most of the research have shown great detection performance. However, not only does classical machine learning require a process of analyzing features, but the detection performance of a trained model also depends on the data analyst's ability. In this paper, we propose a DL-ML Fusion Hybrid Model for malicious web site URL detection based on URL lexical features. the propose model combines the automatic feature extraction layer of deep learning and classical machine learning to improve the feature engineering issue. 60,000 malicious and normal URLs were collected for the experiment and the results showed 23.98%p performance improvement in maximum. In addition, it was possible to train a model in an efficient way with the automation of feature engineering.

• Journal of the Korean earth science society
• /
• v.27 no.1
• /
• pp.83-94
• /
• 2006

The objective of this paper is to extract and analyze the sediment environment change information in the Sachencheon, Gangneung, Korea that was seriously damaged as a result of typhoon Rusa aftermath early in September, 2002 using multi-temporal remote sensing data. For the extraction of change information, an unsupervised approach based on the automatic determination of thresholding values was applied. As the change detection results, turbidity changes right after typhoon Rusa, the decrease of wetlands, the increase of dry sand and channel width and changes of relative level in the stream due to seasonal variation were observed. Sedimentation in the cultivated areas and restoration works also affected the change near the Sacheoncheon. In addition to the change detection analysis, several environmental thematic maps including microtopographic map, distributions of estimated amount of flood deposits and flood hazard landform classification map were generated by using remote sensing and field survey data. In conclusion, multi-temporal remote sensing data can be effectively used for natural hazard analysis and damage information extraction and specific data processing techniques for high-resolution remote sensing data should also be developed.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.9
• /
• pp.69-76
• /
• 2011

According to the related reports, about 60 percents of ship collisions have resulted from operating mistake caused by human factor. Specially, the report said that negligence of observation caused 66.8 percents of the accidents due to a human factor. Hence automatic detection and tracking of an object from an IR images are crucial for safety navigation because it can relieve officer's burden and remedies imperfections of human visual system. In this paper, we present a method to detect an object such as ship, rock and buoy from a sea IR image. Most edge directions of the sea image are horizontal and most vertical edges come out from the object areas. The presented method uses them as a characteristic for the object detection. Vertical edges are extracted from the input image and isolated edges are eliminated. Then morphological closing operation is performed on the vertical edges. This caused vertical edges that actually compose an object be connected and become an object candidate region. Next, reference object regions are extracted using horizontal edges, which appear on the boundaries between surface of the sea and the objects. Finally, object regions are acquired by sequentially integrating reference region and object candidate regions.

• KIISE Transactions on Computing Practices
• /
• v.21 no.2
• /
• pp.154-159
• /
• 2015

Over the past few years, user needs in the smartphone application market have been shifted from diversity toward intelligence. Here, we propose a novel cognitive agent that plans the daily routines of users using the lifelog data collected by the smart phones of individuals. The proposed method first employs DPGMM (Dirichlet Process Gaussian Mixture Model) to automatically extract the users' POI (Point of Interest) from the lifelog data. After extraction, the POI and other meaningful features such as GPS, the user's activity label extracted from the log data is then used to learn the patterns of the user's daily routine by POMDP (Partially Observable Markov Decision Process). To determine the significant patterns within the user's time dependent patterns, collaboration was made with the SNS application Foursquare to record the locations visited by the user and the activities that the user had performed. The method was evaluated by predicting the daily routine of seven users with 3300 feedback data. Experimental results showed that daily routine scheduling can be established after seven days of lifelogged data and feedback data have been collected, demonstrating the potential of the new method of place-time-activity coupled daily routine planning systems in the intelligence application market.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.26 no.1
• /
• pp.56-68
• /
• 2023

A system that extracts seabed topographic information by simultaneously and continuously observing the horizontal position and water depth in the sea by combining a single beam echosounder and GNSS was constructed. By applying the developed system to actual measurements of small-scale sea areas, the effectiveness of bathymetry and sea-floor topographic data acquisition using GNSS and echosounder was examined. By using the developed outdoor program DS-NAV and indoor program DS-CAD and applying the tide level data at the time of actual measurement of the target sea area, it was possible to derive bathymetry results based on the datum level i.e. approximate lowest low water level(A.L.L.W). By using the developed outdoor program DS-NAV and indoor program DS-CAD and applying the tide level data at the time of actual measurement of the target sea area, it was possible to derive the results of bathymetric survey based on the datum level. From database built through the actual measurement. it was possible to create 3D model of the sea-floor topography and extract cross-sections. The results of this study are expected to be economically useful for extracting seabed topographical information from small sea areas or in dredging sites for offshore construction.