• Annual Conference of KIPS
• /
• 2014.11a
• /
• pp.767-770
• /
• 2014

대용량 데이터를 분석하기 위한 맵리듀스 기반 k-NN 조인 질의처리 알고리즘은 최근 데이터 마이닝 및 분석을 기반으로 하는 응용 분야에서 매우 중요하게 활용되고 있다. 그러나, 대표적인 연구인 보로노이 기반 k-NN 조인 질의처리 알고리즘은 보로노이 인덱스 구축 비용이 매우 크기 때문에 대용량 데이터에 적합하지 못하다. 아울러 보로노이 셀 정보를 저장하기 위해 사용하는 R-트리는 맵리듀스 환경의 분산 병렬 처리에 적합하지 않다. 따라서 본 논문에서는 새로운 그리드 인덱스 기반의 k-NN 조인 질의 처리 알고리즘을 제안한다. 첫째, 높은 인덱스 구축 비용 문제를 해결하기 위해, 데이터 분포를 고려한 동적 그리드 인덱스 생성 기법을 제안한다. 둘째, 맵리듀스 환경에서 효율적으로 k-NN 조인 질의를 수행하기 위해, 인접셀 정보를 시그니처로 활용하는 후보영역 탐색 및 필터링 알고리즘을 제안한다. 마지막으로 성능 평가를 통해 제안하는 기법이 질의 처리 시간 측면에서 기존 기법에 비해 최대 3배 높은 질의 처리 성능을 나타냄을 보인다.

• Journal of the FoodService Safety
• /
• v.2 no.1
• /
• pp.10-21
• /
• 2021

According to the analysis which investigated visitors of HACCP system certified stores and non-visitors of such stores on the awareness of HACCP for livestock products, 77.1% (246 pollees) heard of HACCP certification for livestock products, 67.1% (214 pollees) had seen the HACCP certification mark for livestock products, 62.1% (198 pollees) heard of HACCP certification for livestock products in meet retail shops, and 51.4% (164 pollees) were not aware of the recent TV · subway advertisements regarding HACCP certification for livestock products. For every questionnaire on the awareness of HACCP for livestock products, visitors of HACCP system certified stores showed significantly higher response rate than nonvisitors (p<0.01, p<0.001). The majority of pollees (74.9%, 239 pollees) replied that the word HACCP for livestock products brings up the image of safe livestock products, and 37.0% answered that the term HACCP defines 'Hazard analysis critical control point'. Regarding the questions on HACCP system for livestock products, 38.6% showed that they were most curious in terms of the benefits of such system. The demand analysis on HACCP for livestock products for consumer was also conducted. In the analysis, the demand for support of the policy (4.06 points) was higher than demand for education · public promotion of HACCP (4.03 points) and demand for related application (3.90 points).

• Journal of the Korea Institute of Military Science and Technology
• /
• v.22 no.2
• /
• pp.266-277
• /
• 2019

In modern warfare, securing time for preemptive response is recognized as an important factor of victory. The naval combat system, the core of naval forces, also strives to increase the effectiveness of engagement by improving its real-time information processing capabilities. As part of that, it is considered to use the TPR-tree in the naval combat system's target indexing because spatio-temporal searches can be performed quickly even as the number of target information increases. However, because the TPR-tree is slow to process updates, there is a limitation to handling frequent updates. In this paper, we present a method for improving the update performance of TPR-tree by applying the bottom-up update scheme, previously proposed for R-tree, to the TPR-tree. In particular, we analyze the causes of overlaps occurring when applying the bottom-up updates and propose ways to limit the MBR expansion to solve it. Our experimental results show that the proposed technique improves the update performance of TPR-tree from 3.5 times to 12 times while maintaining search performance.

• Journal of KIISE:Databases
• /
• v.32 no.4
• /
• pp.345-361
• /
• 2005

A transform-space index indexes objects represented as points in the transform space An advantage of a transform-space index is that optimization of join algorithms using these indexes becomes relatively simple. However, the disadvantage is that these algorithms cannot be applied to original-space indexes such as the R-tree. As a way of overcoming this disadvantages, the authors earlier proposed the transform-space view join algorithm that joins two original- space indexes in the transform space through the notion of the transform-space view. A transform-space view is a virtual transform-space index that allows us to perform join in the transform space using original-space indexes. In a transform-space view join algorithm, the order of accessing disk pages -for which various space filling curves could be used -makes a significant impact on the performance of joins. In this paper, we Propose a new space filling curve called the adaptive row major order (ARM order). The ARM order adaptively controls the order of accessing pages and significantly reduces the one-pass buffer size (the minimum buffer size required for guaranteeing one disk access per page) and the number of disk accesses for a given buffer size. Through analysis and experiments, we verify the excellence of the ARM order when used with the transform-space view join. The transform-space view join with the ARM order always outperforms existing ones in terms of both measures used: the one-pass buffer size and the number of disk accesses for a given buffer size. Compared to other conventional space filling curves used with the transform-space view join, it reduces the one-pass buffer size by up to 21.3 times and the number of disk accesses by up to $74.6\%$. In addition, compared to existing spatial join algorithms that use R-trees in the original space, it reduces the one-pass buffer size by up to 15.7 times and the number of disk accesses by up to $65.3\%$.

• Composites Research
• /
• v.31 no.2
• /
• pp.43-50
• /
• 2018

The effects of deposition temperature on chemical vapor deposited silicon carbide (CVD-SiC) were studied to obtain high deposition rates and excellent bending strength characteristics. Silicon carbide prepared at $1250{\sim}1400^{\circ}C$ using methyltrichlorosilane(MTS : $CH_3SiCl_3$) by hot-wall CVD showed deposition rates of $95.7{\sim}117.2{\mu}m/hr$. The rate-limiting reaction showed the surface reaction at less than $1300^{\circ}C$, and the mass transfer dominant region at higher temperature. The activation energies calculated by Arrhenius plot were 11.26 kcal/mole and 4.47 kcal/mole, respectively. The surface morphology by the deposition temperature changed from $1250^{\circ}C$ pebble to $1300^{\circ}C$ facet structure and multi-facet structure at above $1350^{\circ}C$. The cross sectional microstructures were columnar at below $1300^{\circ}C$ and isometric at above $1350^{\circ}C$. The crystal phases were all identified as ${\beta}$-SiC, but (220) peak was observed from $1300^{\circ}C$ or higher at $1250^{\circ}C$ (111) and completely changed to (220) at $1400^{\circ}C$. The bending strength showed the maximum value at $1350^{\circ}C$ as densification increased at high temperatures and the microstructure changed from columnar to isometric. On the other hand, at $1400^{\circ}C$, the increasing of grain size and the direction of crystal growth were completely changed from (111) to (220), which is the closest packing face, so the bending strength value seems to have decreased.

• Journal of the Society of Cosmetic Scientists of Korea
• /
• v.37 no.1
• /
• pp.83-90
• /
• 2011

Priming refers to the incidental activation of knowledge structures, such as trait concepts and stereotypes, by the current situational context. Many studies have shown that the recent use of a trait construct or stereotype, even in an earlier or unrelated situation, carries over for a time to exert an unintended, passive influence on the interpretation of behavior. Recent researches have shown that attitudes and other affective reaction can be triggered automatically by the mere presence of relevant objects and events. So assuming that behavioral responses to fragrance are also represented mentally, they should also be capable of becoming automatically activated, by the same principles that govern the development of automaticity of other representations. On the basis of this theory, this study examined the effect of fragrance on the quality of cosmetic products.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.2 s.46
• /
• pp.231-242
• /
• 2007

The existing pattern mining methods[1,2,3,4,5,6,11,12,13] do not use location generalization method on the set of location history data of moving object, but even so they simply do extract only frequent patterns which have no spatio-temporal constraint in moving patterns on specific space. Therefore, it is difficult for those methods to apply to frequent pattern mining which has spatio-temporal constraint such as optimal moving or scheduling paths among the specific points. And also, those methods are required more large memory space due to using pattern tree on memory for reducing repeated scan database. Therefore, more effective pattern mining technique is required for solving these problems. In this paper, in order to develop more effective pattern mining technique, we propose new location generalization method that converts data of detailed level into meaningful spatial information for reducing the processing time for pattern mining of a massive history data set of moving object and space saving. The proposed method can lead the efficient spatial moving pattern mining of moving object using by creating moving sequences through generalizing the location attributes of moving object into 2D spatial area based on $R^*$-Tree and Area Grid Hash Table(AGHT) in preprocessing stage of pattern mining.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.21 no.1
• /
• pp.123-129
• /
• 2012

Recently, interests on cleaning robots workable in pipes (termed as in-pipe cleaning robot) are increasing because Garbage Automatic Collection Facilities (i.e, GACF) are widely being installed in Seoul metropolitan area of Korea. So far research on in-pipe robot has been focused on inspection rather than cleaning. In GACF, when garbage is moving, the impurities which are stuck to the inner face of the pipe are removed (diameter: 300 mm or 400 mm). Thus, in this paper, by using TRIZ (Inventive Theory of Problem Solving in Russian abbreviation), an in-pipe cleaning robot of GACF with the 6-link sliding mechanism will be proposed, which can be adjusted to fit into the inner face of pipe using pneumatic pressure(not spring). The proposed in-pipe cleaning robot for GACF can have forward/backward movement itself as well as rotation of brush in cleaning. The robot body should have the limited size suitable for the smaller pipe with diameter of 300 mm. In addition, for the pipe with diameter of 400 mm, the links of robot should stretch to fit into the diameter of the pipe by using the sliding mechanism. Based on the conceptual design using TRIZ, we will set up the initial design of the robot in collaboration with a field engineer of Robot Valley, Inc. in Korea. For the optimal design of in-pipe cleaning robot, the maximum impulsive force of collision between the robot and the inner face of pipe is simulated by using RecurDyn(R) when the link of sliding mechanism is stretched to fit into the 400 mm diameter of the pipe. The stresses exerted on the 6 links of sliding mechanism by the maximum impulsive force will be simulated by using ANSYS$^{(R)}$ Workbench based on the Design Of Experiment(in short DOE). Finally the optimal dimensions including thicknesses of 4 links will be decided in order to have the best safety factor as 2 in this paper as well as having the minimum mass of 4 links. It will be verified that the optimal design of 4 links has the best safety factor close to 2 as well as having the minimum mass of 4 links, compared with the initial design performed by the expert of Robot Valley, Inc. In addition, the prototype of in-pipe cleaning robot will be stated with further research.

• Journal of Soil and Groundwater Environment
• /
• v.12 no.2
• /
• pp.37-46
• /
• 2007

This study developed composition-plastic media with woodchips and plastic as main materials, and examined the performance of media. Compared to the existing commercial media, the media had similar performance in removal efficiency and microbes attaching characteristic, and was evaluated that they are distinguished from economic side. Performance test of media was conducted to examine the removal capacity of toluene and hydrogen sulfide in a gas stream by using a lab-scale biotrickling filter systems packed with them. At a volumetric loading of $1.5\;m^3/hr$ with inlet concentration 260 ppm and empty bed residence time (EBRT) 42s, the toluene removal efficiency was shown over 90%, and the maximum elimination capacity of toluene in the biotrickling filter was $77g/m^3{\cdot}hr$. Effective co-treatments of $H_2S$ and Toluene were observed in the lab-scale biotrickling filters. The maximum elimination capacity of $H_2S$ was $100\;g-S/m^3{\cdot}hr$. Up to 100 ppm, the concentration of $H_2S$ did not have an effect on toluene removal efficiency, but the removal efficiency of toluene decreased with increasing inlet $H_2S$ concentration.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.1 s.307
• /
• pp.53-66
• /
• 2006

This paper proposes the combined image retrieval system that gives the same relevance as exhaustive search method while its performance can be considerably improved. This system is combined with two different retrieval methods and each gives the same results that full exhaustive search method does. Both of them are two-stage method. One uses condensation of feature vectors, and the other uses binary-tree clustering. These two methods extract the candidate images that always include correct answers at the first stage, and then filter out the incorrect images at the second stage. Inasmuch as these methods use equal algorithm, they can get the same result as full exhaustive search. The first method condenses the dimension of feature vectors, and it uses these condensed feature vectors to compute similarity of query and images in database. It can be found that there is an optimal condensation ratio which minimizes the overall retrieval time. The optimal ratio is applied to first stage of this method. Binary-tree clustering method, searching with recursive 2-means clustering, classifies each cluster dynamically with the same radius. For preserving relevance, its range of query has to be compensated at first stage. After candidate clusters were selected, final results are retrieved by computing similarities again at second stage. The proposed method is combined with above two methods. Because they are not dependent on each other, combined retrieval system can make a remarkable progress in performance.