Earticle

본 논문은 교양과목 추천 시스템 개발을 목적으로, 학생들의 흥미를 기반으로 교양과목을 추천하는 방법을 제안한다. 사용자로부터 흥 미 키워드를 입력받아 관련 키워드를 제시하고, 사용자가 원하는 키워드를 선택한다. 최종적으로 교양과목 추천 알고리즘을 적용하여 점수가 높은 과목을 추천한다. 시스템 개발을 위해, 218개의 인문학 과목을 수집하고 키워드를 추출한 후 워드 임베딩을 하여 키워드간 의 유사도를 비교한다. 이 시스템은 학생들의 적성에 맞춘 과목을 추천하여 교육적 의의를 높이는 데 목적이 있다. 키워드 기반 추천 시스템은 초기 사용자 데이터가 부족한 상황에서도 개인화된 추천을 제공할 수 있으며, 학생들이 단순히 학점을 잘 받을 수 있는 과목 대신, 흥미와 적성에 맞는 과목을 선택하도록 도와준다. 이는 대학 교육의 본래 목표인 전인 교육을 실현하는 데 중요한 역할을 한다.

This study aims to develop a liberal arts course recommendation system based on students' interests. Users input interest keywords, from which related keywords are prioritized for selection. A course recommendation algorithm is applied to suggest courses with high scores. For system development, we collected 218 liberal arts subjects, extracted keywords, and performed word embedding. This system enhances educational significance by recommending courses tailored to students' aptitudes. The keyword-based recommendation system can provide personalized suggestions even with limited initial user data, guiding students to choose courses aligned with their interests and aptitudes rather than merely focusing on grades. This approach plays a crucial role in realizing the holistic education goal of higher education.

본 연구에서는 female adult mesh (FASH) 팬텀으로부터 획득한 CT 영상에 FNLM 알고리즘을 적용하여 유사도를 분석하고자 한다. 먼저, GATE (geant4 application for tomographic emission) 프로그램을 이용하여 FASH의 전신 CT 스캔 데이터를 모델링한 뒤 그 데이터로부터 MATLAB 프로그램을 통해 442 × 256 사이즈의 CT axial 영상을 획득하였다. 그리고 획득한 원본 이미지에 표준편차 0.002를 가지는 가우시안 노이즈를 부가함으로서 Noisy 이미지를 획득한 뒤 그 이미지에 fast non-local means (FNLM) 알고리즘을 가중치 조절 인자 h를 0.05로 적용하였다. 본 연구에서는 유사도를 평가하기 위하여 평균 제곱근 편차 (Root Mean Square Error, RMSE)와 최대 신호 대 잡음비 (Peak Signal-to-noise ratio, PSNR)를 사용하였다. 원본이미지와 FNLM 알고리즘이 적용된 영상과의 유사도를 평가 한 결과, 원본이미지와 노이즈 이미지의 유사도에 비하여 RMSE 수치는 1.45배 향상되었으며, PSNR 수치는 1.04배 향상되었다. 결론적으로 FNLM 알고리즘은 영상을 복원하는데 효과가 있음이 증명되었다.

In this study, we analyzed the similarity about image applied fast non-local means (FNLM) algorithm to CT images obtained from the female adult mesh (FASH) phantom. First, full-body CT scan data of the FASH was modeled using the genant4 application for tomographic emission (GATE) program, and from that data, a 442 × 256 CT axial image was obtained through the MATLAB program. Subsequently, a noisy image was obtained by adding Gaussian noise with a standard deviation of 0.002 to the acquired original image and the FNLM algorithm which has 0.05 weight was applied to the image. In this study, root mean square error (RMSE) and peak signal-to-noise ratio (PSNR) were used to evaluate the similarity. As a result of evaluating the similarity between the original image and the image with FNLM algorithm, the RMSE level was 1.45 times and PSNR level was 1.04 times higher than the noisy image. In conclusion, we demonstrated that the FNLM algorithm can restore efficiently the noisy images.

The purpose of this study was to investigate the effectiveness of dy-namic time warping (DTW) in gait research. Participants in this study were consist of 10 males and 10 females. Equipment used for collecting the gait data of participants in this study was three-dimensional (3D) motion analysis system consisted of 8 infrared CCD cameras operated with a sampling frequency of 120 frames/sec. DTW program used in this study was made using the MATLAB and the normal operation of the DTW program was verified by comparison of result manually calculated and output by the DTW program. Flexion angle of the knee joint of both feet obtained by 3D motion analysis system was analyzed by the DTW program and symmetry index (SI) equation. Statistical analysis of the values obtained by DTW was performed by one-sample t-test in confi-dence interval (CI) 99%, 95%, 90%, 85%, and 80% each using the SPSS. The subjects’ left and right legs were compared 20 times, and other steps of the same foot were compared 20 times. In this study, DTW showed different results from SI which is generally used to test the sim-ilarity of gait. Compared to that of DTW, the threshold figure for similarity evaluation in SI, 10%, is considered too large/high. When the CI thresh-old figure of 95% was adopted in statistical analysis, DTW demonstrat-ed a lower rate of judging two sequences as similar even in the case of normal gait. This study suggests that DTW can be used for the similarity test of gait research.

유사성행렬은 MDS 분석 등 다양한 다차원적 분석의 기초자료로 활용된다. 유사성행렬을 얻는 방법으로 항목의 수가 상대적으로 적을 때는 쌍대비교법이 사용될 수 있으나 항목의 수가 많을 경우 분류과제가 권장된다. 본 연구에서는 분류과제의 한 형태로서 지정분류과제의 특성과 제약을 두 개의 연구를 통해 알아보았다. 실험 1에서는 분류 순서가 MDS의 차원구조에 영향을 주는지를 알아보기 위해 분류 순서를 달리 한 두 종류의 지시문에서 얻은 MDS 차원구조를 비교하였는데 MDS 평면상에서 항목들의 상대적 위치는 서로 다르지 않았다. 목록 구성을 달리해서 목록구성의 영향을 분석한 실험 2의 결과는 각각의 목록구성에 합당한 차원구조를 산출하였다. 두 연구의 결과는 지정분류과제가 분류순서에 영향을 받지 않으면서 안정된 차원구조를 보여주며, 항목집합에 내재된 차원구조를 잘 드러낸다는 것이다. 이 결과를 근거로 지정분류과제의 몇 가지 특성과 제약에 관하여 논의하였다.

A similarity matrix provides basic data for multidimensional scaling analysis such as MDS and HCA. A similarity matrix can be obtained through paired comparison methods or sorting tasks. A sorting task is preferred when the number of item is large. We investigated the traits and constraints of a bounded sorting task, which is a subcategory of sorting task, in two studies. In Exp. 1, we compared two MDS planes from different sorting orders, but found no difference in the axis values. In Exp. 2, we compared two MDS planes from the different item constructions, and found a difference in the interpretation of axis meaning inherent in item construction. The results of the two studies showed that bounded sorting tasks produce stable dimensions free from sorting order, and sensible dimensions sensitive to the item construction. On the basis of these findings some traits and constraints were discussed.

The rate of scientific literature has been increased in the past few decades; new topics and information is added in the form of articles, papers, text documents, web logs, and patents. The growth of information at rapid rate caused a tremendous amount of additions in the current and past knowledge, during this process, new topics emerged, some topics split into many other sub-topics, on the other hand, many topics merge to formed single topic. The selection and search of a topic manually in such a huge amount of information have been found as an expensive and workforce-intensive task. For the emerging need of an automatic process to locate, organize, connect, and make associations among these sources the researchers have proposed different techniques that automatically extract components of the information presented in various formats and organize or structure them. The targeted data which is going to be processed for component extraction might be in the form of text, video or audio. The addition of different algorithms has structured information and grouped similar information into clusters and on the basis of their importance, weighted them. The organized, structured and weighted data is then compared with other structures to find similarity with the use of various algorithms. The semantic patterns can be found by employing visualization techniques that show similarity or relation between topics over time or related to a specific event. In this paper, we have proposed a model based on Cosine Similarity Algorithm for citation network which will answer the questions like, how to connect documents with the help of citation and content similarity and how to visualize and navigate through the document.

The study on word similarity computation plays an important role in natural language processing (NLP). Recently the algorithm based on HowNet is widely used and proves to work well in Chinese word similarity computation. However, the relationship between the number of brother nodes and the fineness of the hierarchy is not considered. This paper investigates the ratio of two words on the brother nodes’ number called sememe probability density and proposes an improved algorithm based on HowNet. The results indicate that the correlation measure of the algorithm presented by this paper is 75.4%, and it is much better than the major state-of-the-art method (68.1%).

For patient information identification, demographic fields (name and address), and medical information (medical history) are compared in order to measure the similarities. The improvement of algorithm related to Koreans’ Hangul name is necessary in the individual distinguishing part in order to introduce MPI system matching Korea. Since Hangul is a type of combination, we cannot match the names more exactly by the one-dimensional distance comparative function which was used in existing English. This paper suggests the algorithm for the Korean name comparison.

Dynamic time warping algorithm (DTW) is a method of measuring the similarity of time series. Concerning the problem that DTW cannot keep high classification accuracy when the computation speed improved, a FG-DTW method based on the idea of naive granular computing is proposed. In this method, firstly, better temporal granularity is acquired by calculating temporal variance feature and it is used to replace original time series; Secondly, the elastic size of under comparing time series granularity allow dynamic adjustment through DTW algorithm and optimal time series corresponding granularity is obtained; Finally, DTW distance is calculated by optimal corresponding granularity model. At the same time, the early termination strategy of infimum function is introduced to improve the efficiency of FG-DTW algorithm. Experiments show that the proposed algorithm improves the running rate and accuracy effectively.

Searching for similar documents from huge amounts of documents is an important and time consuming problem. Although the numerous precise models have been developed for the task, the traditional search algorithms are unable to meet the needs of users for quick search. Herein, a new document similarity calculation and search method with high efficiency is proposed. The calculation of the similarity is based on the total probability model and the efficient search is achieved via level n nodes and paths of citation graph. A special approach from the branch and bound limits the search scope and provide decision algorithm. With the increase in the number of documents, the efficiency of the proposed algorithm is dramatically promoted.

When Cloud Platform for Computer Aided Engineering Simulation (CPCAES) are developed to use Service-Oriented Architecture (SOA), on the one hand, it’s necessary to accurately express the service requirements of users, on the other hand it’s need to semantic of the existing services in the platform, and then to match the ontologies. The existing Ontology Web Language for Services / Universal Description Discovery and Integration (OWL-S / UDDI) algorithm has weak matching precision, and not meet the Computer Aided Engineering (CAE) simulation applications. The other semantic similarity algorithms also have weakness in judgment factor, or do not meet the requirements of CAE simulation applications for matching to service requirement ontology in cloud platform. Such that, the authors present a kind of ontology of service requirement for CAE simulation, model the content of the ontology, including resource requirements, computing requirements, computing job requirements, input, output, etc. The authors give the ontology mapping relationship between Ontology Web Language for Services (OWL-S) and the ontology, give the matching decision rules for the ontologies, propose a matching algorithm for matching the ontologies, and compare with the classic OWL-S / UDDI matching algorithm and a similarity matching algorithm proposed in other paper in the ability for measure the similarity quantity of services, the ability to suit the applications, as well as algorithms recall rate and precision rate. The results show the matching algorithm proposed in this paper is more suitable for service requirement ontology of CPCAES, can use the quantify value for the similarity analysis to the ontologies, and has higher precision rate and higher recall rate, which can reach about 90%. The research work in this paper is used in the second prototype of CPCAES, and the research team is developing the third prototype based on semantic Web Services and SOA framework.

Semantic similarity is fundamental operation in the field of computational lexical semantics, artificial intelligence and cognitive science. Accurate measurement of semantic similarity between words is crucial. The paper presents an effective algorithm for semantic similarity metric of word pairs. Different from previous work, in the new algorithm not only path length, but also IC values have been taken into account. We evaluate our model on the data set of Rubenstein and Goodenough, which is traditional and widely used. Coefficients of correlation between human ratings of similarity based on seven algorithms are calculated. Experiments show that the coefficient of our proposed algorithm with human judgment is 0.8820, which demonstrate that our new algorithm significantly outperformed others.

A novel digital camouflage generation scheme based on color quantization is proposed. First, a color similarity measurement in RGB color space is presented by exploiting the similarity of two color vectors. Then main colors in the background of the host image are extracted by combining color similarity and color quantization. Finally, the camouflage image is generated by replacing target pixels with main background colors. Simulated experiments prove that the proposed algorithm can achieve full fusion of the target and the background, and the generated camouflage image has satisfactory visual quality.

In this paper, a pixel value prediction algorithm using edge components in three directions is proposed. There are various directional edges and similarity between adjacent pixels in natural images. After detecting the edge components in the x-axis direction, the y-axis direction, and the diagonal axis direction, the pixel value is predicted by applying the detected edge components and similarity between neighboring pixels. In particular, the predicted pixel value is calculated according to the intensity of the edge component in the diagonal axis direction. Experimental results show that the proposed algorithm can effectively predict pixel values. The proposed algorithm can be used for applications such as reversible data hiding, reversible watermarking to increase the number of embedded data.

Traditional methods UBCF have limitations of poor recommendation quality and problems of data sparsity. To alleviate these problems, a novel collaborative filtering algorithm is designed, which firstly get the users’ ratings and time intervals for each attribute from the users’ ratings for items, then produce two methods to calculate the similarity between users, introduce a weighting parameters to control the weight between the two similarity methods in order to get a fusion similarity between two users. The results show that this method is able to improve the accuracy of predicted values, resulting in improving recommendation quality of the collaborative filtering recommendation algorithm.

스마트폰의 보급화에 의해 영상은 쉽게 획득할 수 있지만 야간의 조명 조건의 불균형성 그리고 영상 데이터의 전송과 압축 과정에서 열화와 잡음(Noise)이 생성되기도 한다. 이러한 잡음을 최소화하고 영상 화질을 개선하기 위해 Non-Local Means(NLM)은 주변의 픽셀값이 아닌 이미지 내에서 현재 patch와 유사한 patch를 찾아 잡음을 제거한다. 하지만, 유사도를 측정하는 데 있어서 유사한 pactch가 커질수록 유용성이 떨어지는 단점을 가지고 있다. 본 논문에서는 Sum of Absolute Differences 연산을 이용하여 유사도를 계산하고 유사도에 따라 가중치를 적용하는 잡음 제거 방법을 제안한다. 제안한 알고리즘을 사용하여 Salt and Pepper 잡음 이미지에 PSNR 개선이 평균 7.98dB 향상되며 NLM 대비 1.06dB 개선된다. 제안한 알고리즘을 기존 NLM 최적화 논문에 적용 하였을 때 성능 향상을 기대할 수 있다.

With the widespread adoption of smartphones, acquiring images has become easier. However, challenges arise due to uneven lighting conditions at night and the degradation and noise introduced during image transmission and compression. To minimize this noise and improve image quality, Non-Local Means (NLM) techniques are used, which unlike traditional methods, seek out patches within the image that are similar to the current patch to eliminate noise. However, a drawback of NLM is the diminishing utility as the similar patches become larger. This paper proposes a noise reduction method that utilizes the Sum of Absolute Differences to calculate similarity and applies weights accordingly. The proposed algorithm demonstrates an average improvement of 6.911dB in Peak Signal-to-Noise Ratio (PSNR) on Salt and Pepper noise images, showing a 0.713dB improvement over traditional NLM. When the proposed algorithm is applied to existing NLM optimization papers, performance improvements can be expected.

Semantic-based information retrieval techniques understand the meanings of the concepts that users specify in their queries, but the traditional semantic matching methods based on the ontology tree have three weaknesses which may lead to many false matches, causing the falling precision. In order to improve the matching precision and the recall of the information retrieval, this paper proposes a multi-agent improved semantic similarity matching algorithm based on the ontology tree, which can avoid the considerable computation redundancies and mismatching during the entire matching process. The results of the experiments performed on our algorithm show improvements in precision and recall compared with the information retrieval techniques based on the traditional semantic similarity matching methods.