• The KIPS Transactions:PartD
• /
• v.16D no.3
• /
• pp.317-326
• /
• 2009

Recently, there have been many research efforts proposed on trajectory information. Most of them mainly focus their attention on those objects moving in Euclidean space. Many real-world applications such as telematics, however, deal with objects that move only over road networks, which are highly restricted for movement. Thus, the existing methods targeting Euclidean space cannot be directly applied to the road network space. This paper proposes a new clustering scheme for a large volume of trajectory information of objects moving over road networks. To the end, we first define a trajectory on a road network as a sequence of road segments a moving object has passed by. Next, we propose a similarity measurement scheme that judges the degree of similarity by considering the total length of matched road segments. Based on such similarity measurement, we propose a new clustering algorithm for trajectories by modifying and adjusting the FastMap and hierarchical clustering schemes. To evaluate the performance of the proposed clustering scheme, we also develop a trajectory generator considering the observation that most objects tend to move from the starting point to the destination point along their shortest path, and perform a variety of experiments using the trajectories thus generated. The performance result shows that our scheme has the accuracy of over 95% in comparison with that judged by human beings.

• Spatial Information Research
• /
• v.11 no.3
• /
• pp.203-212
• /
• 2003

An integrated 3D GIS-based approach for understanding underground environment is proposed and applied to a land subsidence in densely populated region. Bedrock and geological discontinues were treated as main factors in this study. Because land subsidence in this study area was caused by cavity owing to dissolved limestone in percolating ground water. Ground was classified according to bedrock types using a clustering method and geological information, N value, and RQD value of boreholes were visualized and integrated by 3D-GIS. Therefore it was possible to recognize underground space easily and analyze the ground information effectively.

• The KIPS Transactions:PartD
• /
• v.14D no.3 s.113
• /
• pp.347-354
• /
• 2007

It is not easy for the customers to search the valuable information on the goods among countless items available in the Internet. In order to save time and efforts in searching the goods the customers want, it is very important for a recommender system to have a capability to predict accurately customers' preferences. In this paper we present a refined neighbor selection algorithm for clustering based collaborative filtering in recommender systems. The algorithm exploits a graph approach and searches more efficiently for set of influential customers with respect to a given customer; it searches with concepts of weighted similarity and ranked clustering. The experimental results show that the recommender systems using the proposed method find the proper neighbors and give a good prediction quality.

• Journal of Korea Game Society
• /
• v.17 no.6
• /
• pp.77-88
• /
• 2017

The purpose of this study is to compare existing player type theories using data clustering. For the study, 235 result data of the gamified class in second semester of A university at 2016 used. This study applied K-means and Silhouette to decide the appropriate number of clusters. The player types applied in this study are Bartle's 2-D and 3-D player types, Ferro's five types, and BrainHex. According to the results, Bartle's 2D player type was found to be the best in perspective of data clustering. This study also analyzed the distribution of characteristics for each player types. The results of this study are expected to have an impact on player analysis, which is used in the application of gamification or in the development process.

• The KIPS Transactions:PartD
• /
• v.16D no.3
• /
• pp.327-338
• /
• 2009

A microarray dataset is 2-dimensional dataset with a set of genes and a set of conditions. A bicluster is a subset of genes that show similar behavior within a subset of conditions. Genes that show similar behavior can be considered to have same cellular functions. Thus, biclustering algorithm is a useful tool to uncover groups of genes involved in the same cellular process and groups of conditions which take place in this process. We are proposing a polynomial time algorithm to identify functionally highly correlated biclusters. Our algorithm identifies 1) the gene set that has hidden patterns even if the level of noise is high, 2) the multiple, possibly overlapped, and diverse gene sets, 3) gene sets whose functional association is strongly high, and 4) deterministic biclustering results. We validated the level of functional association of our method, and compared with current methods using GO.

• Journal of Digital Contents Society
• /
• v.18 no.5
• /
• pp.989-994
• /
• 2017

In this paper, we propose a pothole detection system using 2D LiDAR and a pothole detection algorithm. Conventional pothole detection methods can be divided into vibration-based method, 3D reconstruction method, and vision-based method. Proposed pothole detection system uses two inexpensive 2D LiDARs and improves pothole detection performance. Pothole detection algorithm is divided into preprocessing for noise reduction, clustering and line extraction for visualization, and gradient function for pothole decision. By using gradient of distance data function, we check the existence of a pothole and measure the depth and width of the pothole. The pothole detection system is developed using two LiDARs, and the 3D pothole detection performance is shown by detecting a pothole with moving LiDAR system.

• The KIPS Transactions:PartD
• /
• v.12D no.4 s.100
• /
• pp.507-520
• /
• 2005

The R-trees are usually used for an index of trajectories in moving-objects databases. However, they need to access a number of nodes to trace same trajectories because of considering only a spatial proximity. Overlaps and dead spaces should be minimized to enhance the performance of range queries in moving-objects indexes. Trajectories of moving-objects should be preserved to enhance the performance of the trajectory queries. In this paper, we propose the TP3DR-tree(Trajectory Preserved 3DR-tree) using clusters of trajectories for range and trajectory queries. The TP3DR-tree uses two split policies: one is a spatial splitting that splits the same trajectory by clustering and the other is a time splitting that increases space utilization. In addition, we use connecting information in non-leaf nodes to enhance the performance of combined-queries. Our experiments show that the new index outperforms the others in processing queries on various datasets.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2017.04a
• /
• pp.167-167
• /
• 2017

스마트팜 내부의 3차원 공간의 온도, 습도, 기압, 공기질 분석을 통한 돈사 미세 조절 기술에 대한 연구가 진행 중이다. 해당 특성 중에서 기압을 제외한 환경인자들은 돈사 내의 구조 특성상 위치별로, 시간별로 매우 상이한 변이의 형태를 보인다. 일정 시점을 기준으로 계측 지점 이외의 지점에 대한 환경인자들을 공간적으로 추정하는 기술은 대표적으로 컴퓨터 분석 기술에 의존하고 있다. 시간 복잡도가 매우 높은 CFD(Computer Fluid Dynamics) 방식은 정밀도 측면에서 유리하나, 상응하는 제어 기술/하드웨어 등의 부재로 모델링 결과의 활용도가 낮다고 볼 수 있다. 본 연구에서는 CFD를 수행하는 과정에 있어 실질적으로 유효한 단위로 공간 분해능을 낮추고, 동등한 크기의 세부 공간에 대한 모델링을 병렬적으로 수행하기 위한 방안을 연구하였다. 실험적으로 돈사 환경을 3차원으로 구성하기 위하여, 공기 흡입구, 배출구, 기둥, 덕트 요소를 포함시켰다. 실내 공간을 1차적으로 가로, 세로, 높이방향으로 $3{\times}3{\times}3$ 균등 분배한 후 3차원 행렬로 분할하였다. 각 분할된 행렬에 대한 연산 수행을 위하여 현재까지 대중에 공개된 SBC(Single Board Computer) 중 가장 높은 연산 수행 능력이 있는 Odroid-XU4(Hardkernel, AnYang, Korea) 16식을 병렬 클러스터링 기술로 연동하였다. 하나의 AP당 8개의 코어가 내장되어 있으므로, 총 128개의 코어를 이용하여 동시에 128개의 3D 정방행렬 연산이 가능하도록 구성하였다. 모델링을 위한 수학적 모델로는 실험적으로 Steady turbulent model (Newtonian coefficient)을 이용하였다. 클러스터링을 이용한 병렬 처리의 특성상 균등한 정보량을 동시에 배분해야 하므로 108 ($27{\times}4$)개의 코어를 이용하여 1차적으로 나뉜 공간을 다시 4등분하여 동시에 $12{\times}12{\times}12$에 해당하는 공간 분해능에 대한 처리를 동시에 수행할 수 있도록 하였다. 2단계에 걸쳐 분할한 공간 세그먼트에 대한 클러스터링 연산 수행 결과 초당 15회 정도의 연산을 수행할 수 있었으며, 시간 분해능을 100으로 설정한 경우 약 5초가 수행되었다. 선행적으로 수행하였던 CFD 모델링 (OpenFOAM)과 비교하였을 때 상대적으로 정밀도가 낮은 3차원 모델링 결과를 얻을 수 있었다. 모델링에 소요되는 시간을 비약적으로 경감 시킨 장점을 살려 적정한 공간 분할 기법과 추가로 발생하는 다수의 바운더리 조건을 근사적으로 추정할 수 있는 데이터 마이닝 기술이 보완되어야 할 것이다.

• Journal of the Korean Society for Library and Information Science
• /
• v.40 no.3
• /
• pp.175-190
• /
• 2006

For the recent several years, there has been much interest in lexical semantic network However it seems to be very difficult to evaluate the effectiveness and correctness of it and invent the methods for applying it into various problem domains. In order to offer the fundamental ideas about how to evaluate and utilize lexical semantic networks, we developed two automatic vol·d clustering systems, which are called system A and system B respectively. 68.455.856 words were used to learn both systems. We compared the clustering results of system A to those of system B which is extended by the lexical-semantic network. The system B is extended by reconstructing the feature vectors which are used the elements of the lexical-semantic network of 3.656 '-ha' verbs. The target data is the 'multilingual Word Net-CoroNet'. When we compared the accuracy of the system A and system B, we found that system B showed the accuracy of 46.6% which is better than that of system A. 45.3%.

• Journal of Industrial Convergence
• /
• v.20 no.12
• /
• pp.195-202
• /
• 2022

Underground Geospatial Information Map Management System(UGIMMS) integrates various underground facilities in the underground space into 3D mesh data, and supports to check the 3D image and location of the underground facilities in the mobile app. However, there is a problem that it takes a long time to run in the app because various underground facilities can exist in some areas executed by the app and can be seen layer by layer. In this paper, we propose a deep learning-based K-means vertex clustering algorithm as a method to reduce the execution time in the app by reducing the size of the data by reducing the number of vertices in the 3D mesh data within the range that does not cause a problem in visibility. First, our proposed method obtains refined vertex feature information through a deep learning encoder-decoder based model. And second, the method was simplified by grouping similar vertices through K-means vertex clustering using feature information. As a result of the experiment, when the vertices of various underground facilities were reduced by 30% with the proposed method, the 3D image model was slightly deformed, but there was no missing part, so there was no problem in checking it in the app.