• Korean Journal of Environmental Agriculture
• /
• v.38 no.1
• /
• pp.54-60
• /
• 2019

BACKGROUND: Hydrogen fluoride is one of the 97 accident preparedness substances regulated by the Ministry of Environment (Republic of Korea) and chemical accidents should be managed centrally due to continual occurrence. Especially, hydrogen fluoride has a characteristic of rapid diffusion and very toxic when leaking into the environment. Therefore, it is important to predict the impact range quickly and to evaluate the residual contamination immediately to minimize the human and environmental damages. METHODS AND RESULTS: In order to estimate the accident impact range, the off-site consequence analysis (OCA) was performed to the worst and alternative scenarios. Also, in order to evaluate the residual contamination of hydrogen fluoride in crop, the samples in accident site were collected from 15-divided regions (East direction from accident sites based on the main wind direction), and the concentration was measured by fluoride ($F^-$) ion-selective electrode potentiometer (ISE). As a result of the OCA, the affected distance by the worst scenario was estimated to be >10 km from the accident site and the range by the alternative scenario was estimated to be about 1.9 km. The residual contamination of hydrogen fluoride was highest in the samples near the site of the accident (E-1, 276.82 mg/kg) and tended to decrease as it moved eastward. Meanwhile, the concentrations from SE and NE (4.96~28.98 mg/kg) tended to be lower than the samples near the accident site. As a result, the concentration of hydrogen fluoride was reduced to a low concentration within 2 km from the accident site (<5 mg/kg), and the actual damage range was estimated to be around 2.2 km. Therefore, it is suggested that the results are similar to those of alternative accident scenarios calculated by OCA (about 1.9 km). CONCLUSION: It is difficult to estimate the chemical accident-affecting range/region by the OCA evaluation, because it is not possible to input all physicochemical parameters. However simultaneous measurement of the residual contamination in the environment will be very helpful in determining the diffusion range of actual chemical accident.

• Journal of the Korean Geotechnical Society
• /
• v.25 no.12
• /
• pp.87-105
• /
• 2009

Nonlinear dynamic deformation characteristics, expressed in terms of normalized shear modulus reduction curve (G/$G_{max}-\log\gamma$, G/$G_{max}$ curve) and damping curve (D-$\log\gamma$), are important input parameters with shear wave velocity profile ($V_s$-profile) in the seismic analysis of (new or existing) fill dam. In this paper, the reasonable and economical methods to evaluate the nonlinear dynamic deformation characteristics for core zone and rockfill zone respectively are presented. For the core zone, 111 G/$G_{max}$ curves and 98 damping curves which meet the requirements of core material were compiled and representative curves and ranges were proposed for the three ranges of confining pressure (0~100 kPa, 100 kPa~200 kPa, more than 200 kPa). The reliability of the proposed curves for the core zone were verified by comparing with the resonant column test results of two kinds of core materials. For the rockfill zone, 135 G/$G_{max}$ curves and 65 damping curves were compiled from the test results of gravelly materials using large scale testing equipments. The representative curves and ranges for G/$G_{max}$ were proposed for the three ranges of confining pressure (0~50 kPa, 50 kPa~100 kPa, more than 100 kPa) and those for damping were proposed independently of confining pressure. The reliability of the proposed curves for the rockfill zone were verified by comparing with the large scale triaxial test results of rockfill materials in the B-dam which is being constructed.

• Journal of agriculture & life science
• /
• v.50 no.3
• /
• pp.1-11
• /
• 2016

With increasing interest, there have been studies on LiDAR(Light Detection And Ranging)-based DEM(Digital Elevation Model) to acquire three dimensional topographic information. For producing LiDAR DEM with better accuracy, Filtering process is crucial, where only surface reflected LiDAR points are left to construct DEM while non-surface reflected LiDAR points need to be removed from the raw LiDAR data. In particular, the changes of input values for filtering algorithm-constructing parameters are supposed to produce different products. Therefore, this study is aimed to contribute to better understanding the effects of the changes of the levels of GroundFilter Algrothm's Mean parameter(GFmn) embedded in FUSION software on the accuracy of the LiDAR DEM products, using LiDAR data collected for Hwacheon, Yangju, Gyeongsan and Jangheung watershed experimental area. The effect of GFmn level changes on the products' accuracy is estimated by measuring and comparing the residuals between the elevations at the same locations of a field and different GFmn level-produced LiDAR DEM sample points. In order to test whether there are any differences among the five GFmn levels; 1, 3, 5, 7 and 9, One-way ANOVA is conducted. In result of One-way ANOVA test, it is found that the change in GFmn level significantly affects the accuracy (F-value: 4.915, p<0.01). After finding significance of the GFmn level effect, Tukey HSD test is also conducted as a Post hoc test for grouping levels by the significant differences. In result, GFmn levels are divided into two subsets ('7, 5, 9, 3' vs. '1'). From the observation of the residuals of each individual level, it is possible to say that LiDAR DEM is generated most accurately when GFmn is given as 7. Through this study, the most desirable parameter value can be suggested to produce filtered LiDAR DEM data which can provide the most accurate elevation information.

• Korean Journal of Radiology
• /
• v.2 no.1
• /
• pp.28-36
• /
• 2001

Objective: To provide a systematic overview of the effects of various parameters on contrast enhancement within the same population, an animal experiment as well as a computer-aided simulation study was performed. Materials and Methods: In an animal experiment, single-level dynamic CT through the liver was performed at 5-second intervals just after the injection of contrast medium for 3 minutes. Combinations of three different amounts (1, 2, 3 mL/kg), concentrations (150, 200, 300 mgI/mL), and injection rates (0.5, 1, 2 mL/sec) were used. The CT number of the aorta (A), portal vein (P) and liver (L) was measured in each image, and time-attenuation curves for A, P and L were thus obtained. The degree of maximum enhancement (Imax) and time to reach peak enhancement (Tmax) of A, P and L were determined, and times to equilibrium (Teq) were analyzed. In the computed-aided simulation model, a program based on the amount, flow, and diffusion coefficient of body fluid in various compartments of the human body was designed. The input variables were the concentrations, volumes and injection rates of the contrast media used. The program generated the time-attenuation curves of A, P and L, as well as liver-to-hepatocellular carcinoma (HCC) contrast curves. On each curve, we calculated and plotted the optimal temporal window (time period above the lower threshold, which in this experiment was 10 Hounsfield units), the total area under the curve above the lower threshold, and the area within the optimal range. Results: A. Animal Experiment: At a given concentration and injection rate, an increased volume of contrast medium led to increases in Imax A, P and L. In addition, Tmax A, P, L and Teq were prolonged in parallel with increases in injection time The time-attenuation curve shifted upward and to the right. For a given volume and injection rate, an increased concentration of contrast medium increased the degree of aortic, portal and hepatic enhancement, though Tmax A, P and L remained the same. The time-attenuation curve shifted upward. For a given volume and concentration of contrast medium, changes in the injection rate had a prominent effect on aortic enhancement, and that of the portal vein and hepatic parenchyma also showed some increase, though the effect was less prominent. A increased in the rate of contrast injection led to shifting of the time enhancement curve to the left and upward. B. Computer Simulation: At a faster injection rate, there was minimal change in the degree of hepatic attenuation, though the duration of the optimal temporal window decreased. The area between 10 and 30 HU was greatest when contrast media was delivered at a rate of 2 3 mL/sec. Although the total area under the curve increased in proportion to the injection rate, most of this increase was above the upper threshould and thus the temporal window was narrow and the optimal area decreased. Conclusion: Increases in volume, concentration and injection rate all resulted in improved arterial enhancement. If cost was disregarded, increasing the injection volume was the most reliable way of obtaining good quality enhancement. The optimal way of delivering a given amount of contrast medium can be calculated using a computer-based mathematical model.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.167-181
• /
• 2018

Over the past decade, deep learning has been in spotlight among various machine learning algorithms. In particular, CNN(Convolutional Neural Network), which is known as the effective solution for recognizing and classifying images or voices, has been popularly applied to classification and prediction problems. In this study, we investigate the way to apply CNN in business problem solving. Specifically, this study propose to apply CNN to stock market prediction, one of the most challenging tasks in the machine learning research. As mentioned, CNN has strength in interpreting images. Thus, the model proposed in this study adopts CNN as the binary classifier that predicts stock market direction (upward or downward) by using time series graphs as its inputs. That is, our proposal is to build a machine learning algorithm that mimics an experts called 'technical analysts' who examine the graph of past price movement, and predict future financial price movements. Our proposed model named 'CNN-FG(Convolutional Neural Network using Fluctuation Graph)' consists of five steps. In the first step, it divides the dataset into the intervals of 5 days. And then, it creates time series graphs for the divided dataset in step 2. The size of the image in which the graph is drawn is $40(pixels){\times}40(pixels)$, and the graph of each independent variable was drawn using different colors. In step 3, the model converts the images into the matrices. Each image is converted into the combination of three matrices in order to express the value of the color using R(red), G(green), and B(blue) scale. In the next step, it splits the dataset of the graph images into training and validation datasets. We used 80% of the total dataset as the training dataset, and the remaining 20% as the validation dataset. And then, CNN classifiers are trained using the images of training dataset in the final step. Regarding the parameters of CNN-FG, we adopted two convolution filters ($5{\times}5{\times}6$ and $5{\times}5{\times}9$) in the convolution layer. In the pooling layer, $2{\times}2$ max pooling filter was used. The numbers of the nodes in two hidden layers were set to, respectively, 900 and 32, and the number of the nodes in the output layer was set to 2(one is for the prediction of upward trend, and the other one is for downward trend). Activation functions for the convolution layer and the hidden layer were set to ReLU(Rectified Linear Unit), and one for the output layer set to Softmax function. To validate our model - CNN-FG, we applied it to the prediction of KOSPI200 for 2,026 days in eight years (from 2009 to 2016). To match the proportions of the two groups in the independent variable (i.e. tomorrow's stock market movement), we selected 1,950 samples by applying random sampling. Finally, we built the training dataset using 80% of the total dataset (1,560 samples), and the validation dataset using 20% (390 samples). The dependent variables of the experimental dataset included twelve technical indicators popularly been used in the previous studies. They include Stochastic %K, Stochastic %D, Momentum, ROC(rate of change), LW %R(Larry William's %R), A/D oscillator(accumulation/distribution oscillator), OSCP(price oscillator), CCI(commodity channel index), and so on. To confirm the superiority of CNN-FG, we compared its prediction accuracy with the ones of other classification models. Experimental results showed that CNN-FG outperforms LOGIT(logistic regression), ANN(artificial neural network), and SVM(support vector machine) with the statistical significance. These empirical results imply that converting time series business data into graphs and building CNN-based classification models using these graphs can be effective from the perspective of prediction accuracy. Thus, this paper sheds a light on how to apply deep learning techniques to the domain of business problem solving.

• Journal of Intelligence and Information Systems
• /
• v.24 no.2
• /
• pp.221-241
• /
• 2018

Deep learning is getting attention recently. The deep learning technique which had been applied in competitions of the International Conference on Image Recognition Technology(ILSVR) and AlphaGo is Convolution Neural Network(CNN). CNN is characterized in that the input image is divided into small sections to recognize the partial features and combine them to recognize as a whole. Deep learning technologies are expected to bring a lot of changes in our lives, but until now, its applications have been limited to image recognition and natural language processing. The use of deep learning techniques for business problems is still an early research stage. If their performance is proved, they can be applied to traditional business problems such as future marketing response prediction, fraud transaction detection, bankruptcy prediction, and so on. So, it is a very meaningful experiment to diagnose the possibility of solving business problems using deep learning technologies based on the case of online shopping companies which have big data, are relatively easy to identify customer behavior and has high utilization values. Especially, in online shopping companies, the competition environment is rapidly changing and becoming more intense. Therefore, analysis of customer behavior for maximizing profit is becoming more and more important for online shopping companies. In this study, we propose 'CNN model of Heterogeneous Information Integration' using CNN as a way to improve the predictive power of customer behavior in online shopping enterprises. In order to propose a model that optimizes the performance, which is a model that learns from the convolution neural network of the multi-layer perceptron structure by combining structured and unstructured information, this model uses 'heterogeneous information integration', 'unstructured information vector conversion', 'multi-layer perceptron design', and evaluate the performance of each architecture, and confirm the proposed model based on the results. In addition, the target variables for predicting customer behavior are defined as six binary classification problems: re-purchaser, churn, frequent shopper, frequent refund shopper, high amount shopper, high discount shopper. In order to verify the usefulness of the proposed model, we conducted experiments using actual data of domestic specific online shopping company. This experiment uses actual transactions, customers, and VOC data of specific online shopping company in Korea. Data extraction criteria are defined for 47,947 customers who registered at least one VOC in January 2011 (1 month). The customer profiles of these customers, as well as a total of 19 months of trading data from September 2010 to March 2012, and VOCs posted for a month are used. The experiment of this study is divided into two stages. In the first step, we evaluate three architectures that affect the performance of the proposed model and select optimal parameters. We evaluate the performance with the proposed model. Experimental results show that the proposed model, which combines both structured and unstructured information, is superior compared to NBC(Naïve Bayes classification), SVM(Support vector machine), and ANN(Artificial neural network). Therefore, it is significant that the use of unstructured information contributes to predict customer behavior, and that CNN can be applied to solve business problems as well as image recognition and natural language processing problems. It can be confirmed through experiments that CNN is more effective in understanding and interpreting the meaning of context in text VOC data. And it is significant that the empirical research based on the actual data of the e-commerce company can extract very meaningful information from the VOC data written in the text format directly by the customer in the prediction of the customer behavior. Finally, through various experiments, it is possible to say that the proposed model provides useful information for the future research related to the parameter selection and its performance.

• Journal of Intelligence and Information Systems
• /
• v.23 no.4
• /
• pp.147-168
• /
• 2017

There have been many studies on accurate stock market forecasting in academia for a long time, and now there are also various forecasting models using various techniques. Recently, many attempts have been made to predict the stock index using various machine learning methods including Deep Learning. Although the fundamental analysis and the technical analysis method are used for the analysis of the traditional stock investment transaction, the technical analysis method is more useful for the application of the short-term transaction prediction or statistical and mathematical techniques. Most of the studies that have been conducted using these technical indicators have studied the model of predicting stock prices by binary classification - rising or falling - of stock market fluctuations in the future market (usually next trading day). However, it is also true that this binary classification has many unfavorable aspects in predicting trends, identifying trading signals, or signaling portfolio rebalancing. In this study, we try to predict the stock index by expanding the stock index trend (upward trend, boxed, downward trend) to the multiple classification system in the existing binary index method. In order to solve this multi-classification problem, a technique such as Multinomial Logistic Regression Analysis (MLOGIT), Multiple Discriminant Analysis (MDA) or Artificial Neural Networks (ANN) we propose an optimization model using Genetic Algorithm as a wrapper for improving the performance of this model using Multi-classification Support Vector Machines (MSVM), which has proved to be superior in prediction performance. In particular, the proposed model named GA-MSVM is designed to maximize model performance by optimizing not only the kernel function parameters of MSVM, but also the optimal selection of input variables (feature selection) as well as instance selection. In order to verify the performance of the proposed model, we applied the proposed method to the real data. The results show that the proposed method is more effective than the conventional multivariate SVM, which has been known to show the best prediction performance up to now, as well as existing artificial intelligence / data mining techniques such as MDA, MLOGIT, CBR, and it is confirmed that the prediction performance is better than this. Especially, it has been confirmed that the 'instance selection' plays a very important role in predicting the stock index trend, and it is confirmed that the improvement effect of the model is more important than other factors. To verify the usefulness of GA-MSVM, we applied it to Korea's real KOSPI200 stock index trend forecast. Our research is primarily aimed at predicting trend segments to capture signal acquisition or short-term trend transition points. The experimental data set includes technical indicators such as the price and volatility index (2004 ~ 2017) and macroeconomic data (interest rate, exchange rate, S&P 500, etc.) of KOSPI200 stock index in Korea. Using a variety of statistical methods including one-way ANOVA and stepwise MDA, 15 indicators were selected as candidate independent variables. The dependent variable, trend classification, was classified into three states: 1 (upward trend), 0 (boxed), and -1 (downward trend). 70% of the total data for each class was used for training and the remaining 30% was used for verifying. To verify the performance of the proposed model, several comparative model experiments such as MDA, MLOGIT, CBR, ANN and MSVM were conducted. MSVM has adopted the One-Against-One (OAO) approach, which is known as the most accurate approach among the various MSVM approaches. Although there are some limitations, the final experimental results demonstrate that the proposed model, GA-MSVM, performs at a significantly higher level than all comparative models.

• Korean Journal of Ecology and Environment
• /
• v.45 no.2
• /
• pp.158-173
• /
• 2012

The objectives of this study were to analyze the longitudinal gradient and temporal variations of water quality in Daecheong Reservoir in relation to the major inflowing streams from the watershed, during 2001~2010. For the study, we selected 7 main-stream sites of the reservoir along the main axis of the reservoir, from the headwater to the dam and 8 tributary streams. In-reservoir nutrients of TN and TP showed longitudinal declines from the headwater to the dam, which results in a distinct zonation of the riverine ($R_z$, M1~M3), transition ($T_z$, M4~M6), and lacustrine zone ($L_z$, M7) in water quality, as shown in other foreign reservoirs. Chlorophyll-a (CHL) and BOD as an indicator of organic matter, were maximum in the $T_z$. Concentration of total phosphorus (TP) was the highest (8.52 $mg\;L^{-1}$) on March in the $R_z$, and was the highest (165 ${\mu}g\;L^{-1}$) in the $L_z$ on July. Values of TN was the maximum (377 ${\mu}g\;L^{-1}$) on August in the $R_z$, and was the highest (3.76 $mg\;L^{-1}$) in the $L_z$ on August. Ionic dilution was evident during September~October, after the monsoon rain. The mean ratios of TN : TP, as an indicator of limiting factor, were 88, which indicates that nitrogen is a surplus for phytoplankton growth in this system. Nutrient analysis of inflowing streams showed that major nutrient sources were headwater streams of T1~T2 and Ockcheon-Stream of T5, and the most influential inflowing stream to the reservoir was T5, which is located in the mid-reservoir, and is directly influenced by the waste-water treatment plants. The key parameters, influenced by the monsoon rain, were TP and suspended solids (SS). Empirical models of trophic variables indicated that variations of CHL in the $R_z$ ($R^2$=0.044, p=0.264) and $T_z$ ($R^2$=0.126, p=0.054) were not accounted by TN, but were significant (p=0.032) in the $L_z$. The variation of the log-transformed $I_r$-CHL was not accounted ($R^2$=0.258, p=0.110) by $I_w$-TN of inflowing streams, but was determined ($R^2$=0.567, p=0.005) by $I_w$-TP of inflowing streams. In other words, TP inputs from the inflowing streams were the major determinants on the in-reservoir phytoplankton growth. Regression analysis of TN : TP suggested that the ratio was determined by P, rather than N. Overall, our data suggest that TP and suspended solids, during the summer flood period, should be reduced from the eutrophication control and P-input from Ockcheon-Stream should be controlled for water quality improvement.

• Journal of Intelligence and Information Systems
• /
• v.18 no.3
• /
• pp.185-202
• /
• 2012

.Since the value of information has been realized in the information society, the usage and collection of information has become important. A facial expression that contains thousands of information as an artistic painting can be described in thousands of words. Followed by the idea, there has recently been a number of attempts to provide customers and companies with an intelligent service, which enables the perception of human emotions through one's facial expressions. For example, MIT Media Lab, the leading organization in this research area, has developed the human emotion prediction model, and has applied their studies to the commercial business. In the academic area, a number of the conventional methods such as Multiple Regression Analysis (MRA) or Artificial Neural Networks (ANN) have been applied to predict human emotion in prior studies. However, MRA is generally criticized because of its low prediction accuracy. This is inevitable since MRA can only explain the linear relationship between the dependent variables and the independent variable. To mitigate the limitations of MRA, some studies like Jung and Kim (2012) have used ANN as the alternative, and they reported that ANN generated more accurate prediction than the statistical methods like MRA. However, it has also been criticized due to over fitting and the difficulty of the network design (e.g. setting the number of the layers and the number of the nodes in the hidden layers). Under this background, we propose a novel model using Support Vector Regression (SVR) in order to increase the prediction accuracy. SVR is an extensive version of Support Vector Machine (SVM) designated to solve the regression problems. The model produced by SVR only depends on a subset of the training data, because the cost function for building the model ignores any training data that is close (within a threshold ${\varepsilon}$) to the model prediction. Using SVR, we tried to build a model that can measure the level of arousal and valence from the facial features. To validate the usefulness of the proposed model, we collected the data of facial reactions when providing appropriate visual stimulating contents, and extracted the features from the data. Next, the steps of the preprocessing were taken to choose statistically significant variables. In total, 297 cases were used for the experiment. As the comparative models, we also applied MRA and ANN to the same data set. For SVR, we adopted '${\varepsilon}$-insensitive loss function', and 'grid search' technique to find the optimal values of the parameters like C, d, ${\sigma}^2$, and ${\varepsilon}$. In the case of ANN, we adopted a standard three-layer backpropagation network, which has a single hidden layer. The learning rate and momentum rate of ANN were set to 10%, and we used sigmoid function as the transfer function of hidden and output nodes. We performed the experiments repeatedly by varying the number of nodes in the hidden layer to n/2, n, 3n/2, and 2n, where n is the number of the input variables. The stopping condition for ANN was set to 50,000 learning events. And, we used MAE (Mean Absolute Error) as the measure for performance comparison. From the experiment, we found that SVR achieved the highest prediction accuracy for the hold-out data set compared to MRA and ANN. Regardless of the target variables (the level of arousal, or the level of positive / negative valence), SVR showed the best performance for the hold-out data set. ANN also outperformed MRA, however, it showed the considerably lower prediction accuracy than SVR for both target variables. The findings of our research are expected to be useful to the researchers or practitioners who are willing to build the models for recognizing human emotions.

• Journal of Intelligence and Information Systems
• /
• v.18 no.3
• /
• pp.171-183
• /
• 2012

In the development of technologies, the exhibition industry has received much attention from governments and companies as an important way of marketing activities. Also, the exhibitors have considered the exhibition as new channels of marketing activities. However, the growing size of exhibitions for net square feet and the number of visitors naturally creates the competitive environment for them. Therefore, to make use of the effective marketing tools in these environments, they have planned and implemented many promotion technics. Especially, through smart environment which makes them provide real-time information for visitors, they can implement various kinds of promotion. However, promotions ignoring visitors' various needs and preferences can lose the original purposes and functions of them. That is, as indiscriminate promotions make visitors feel like spam, they can't achieve their purposes. Therefore, they need an approach using STP strategy which segments visitors through right evidences (Segmentation), selects the target visitors (Targeting), and give proper services to them (Positioning). For using STP Strategy in the smart exhibition environment, we consider these characteristics of it. First, an exhibition is defined as market events of a specific duration, which are held at intervals. According to this, exhibitors who plan some promotions should different events and promotions in each exhibition. Therefore, when they adopt traditional STP strategies, a system can provide services using insufficient information and of existing visitors, and should guarantee the performance of it. Second, to segment automatically, cluster analysis which is generally used as data mining technology can be adopted. In the smart exhibition environment, information of visitors can be acquired in real-time. At the same time, services using this information should be also provided in real-time. However, many clustering algorithms have scalability problem which they hardly work on a large database and require for domain knowledge to determine input parameters. Therefore, through selecting a suitable methodology and fitting, it should provide real-time services. Finally, it is needed to make use of data in the smart exhibition environment. As there are useful data such as booth visit records and participation records for events, the STP strategy for the smart exhibition is based on not only demographical segmentation but also behavioral segmentation. Therefore, in this study, we analyze a case of the promotion methodology which exhibitors can provide a differentiated service to segmented visitors in the smart exhibition environment. First, considering characteristics of the smart exhibition environment, we draw evidences of segmentation and fit the clustering methodology for providing real-time services. There are many studies for classify visitors, but we adopt a segmentation methodology based on visitors' behavioral traits. Through the direct observation, Veron and Levasseur classify visitors into four groups to liken visitors' traits to animals (Butterfly, fish, grasshopper, and ant). Especially, because variables of their classification like the number of visits and the average time of a visit can estimate in the smart exhibition environment, it can provide theoretical and practical background for our system. Next, we construct a pilot system which automatically selects suitable visitors along the objectives of promotions and instantly provide promotion messages to them. That is, based on the segmentation of our methodology, our system automatically selects suitable visitors along the characteristics of promotions. We adopt this system to real exhibition environment, and analyze data from results of adaptation. As a result, as we classify visitors into four types through their behavioral pattern in the exhibition, we provide some insights for researchers who build the smart exhibition environment and can gain promotion strategies fitting each cluster. First, visitors of ANT type show high response rate for promotion messages except experience promotion. So they are fascinated by actual profits in exhibition area, and dislike promotions requiring a long time. Contrastively, visitors of GRASSHOPPER type show high response rate only for experience promotion. Second, visitors of FISH type appear favors to coupon and contents promotions. That is, although they don't look in detail, they prefer to obtain further information such as brochure. Especially, exhibitors that want to give much information for limited time should give attention to visitors of this type. Consequently, these promotion strategies are expected to give exhibitors some insights when they plan and organize their activities, and grow the performance of them.