• Korean Journal of Agricultural and Forest Meteorology
• /
• v.8 no.1
• /
• pp.1-9
• /
• 2006

Regardless of the recent observed warmer winters in Korea, more freeze injuries and associated economic losses are reported in fruit industry than ever before. Existing freeze-frost forecasting systems employ only daily minimum temperature for judging the potential damage on dormant flowering buds but cannot accommodate potential biological responses such as short-term acclimation of plants to severe weather episodes as well as annual variation in climate. We introduce 'dormancy depth', in addition to daily minimum temperature, as a complementary criterion for judging the potential damage of freezing temperatures on dormant flowering buds of grape vines. Dormancy depth can be estimated by a phonology model driven by daily maximum and minimum temperature and is expected to make a reasonable proxy for physiological tolerance of buds to low temperature. Dormancy depth at a selected site was estimated for a climatological normal year by this model, and we found a close similarity in time course change pattern between the estimated dormancy depth and the known cold tolerance of fruit trees. Inter-annual and spatial variation in dormancy depth were identified by this method, showing the feasibility of using dormancy depth as a proxy indicator for tolerance to low temperature during the winter season. The model was applied to 10 vineyards which were recently damaged by a cold spell, and a temperature-dormancy depth-freeze injury relationship was formulated into an exponential-saturation model which can be used for judging freeze risk under a given set of temperature and dormancy depth. Based on this model and the expected lowest temperature with a 10-year recurrence interval, a freeze risk probability map was produced for Hwaseong County, Korea. The results seemed to explain why the vineyards in the warmer part of Hwaseong County have been hit by more freeBe damage than those in the cooler part of the county. A dormancy depth-minimum temperature dual engine freeze warning system was designed for vineyards in major production counties in Korea by combining the site-specific dormancy depth and minimum temperature forecasts with the freeze risk model. In this system, daily accumulation of thermal time since last fall leads to the dormancy state (depth) for today. The regional minimum temperature forecast for tomorrow by the Korea Meteorological Administration is converted to the site specific forecast at a 30m resolution. These data are input to the freeze risk model and the percent damage probability is calculated for each grid cell and mapped for the entire county. Similar approaches may be used to develop freeze warning systems for other deciduous fruit trees.

• Journal of Intelligence and Information Systems
• /
• v.25 no.4
• /
• pp.141-154
• /
• 2019

Rapid growth of internet technology and social media is progressing. Data mining technology has evolved to enable unstructured document representations in a variety of applications. Sentiment analysis is an important technology that can distinguish poor or high-quality content through text data of products, and it has proliferated during text mining. Sentiment analysis mainly analyzes people's opinions in text data by assigning predefined data categories as positive and negative. This has been studied in various directions in terms of accuracy from simple rule-based to dictionary-based approaches using predefined labels. In fact, sentiment analysis is one of the most active researches in natural language processing and is widely studied in text mining. When real online reviews aren't available for others, it's not only easy to openly collect information, but it also affects your business. In marketing, real-world information from customers is gathered on websites, not surveys. Depending on whether the website's posts are positive or negative, the customer response is reflected in the sales and tries to identify the information. However, many reviews on a website are not always good, and difficult to identify. The earlier studies in this research area used the reviews data of the Amazon.com shopping mal, but the research data used in the recent studies uses the data for stock market trends, blogs, news articles, weather forecasts, IMDB, and facebook etc. However, the lack of accuracy is recognized because sentiment calculations are changed according to the subject, paragraph, sentiment lexicon direction, and sentence strength. This study aims to classify the polarity analysis of sentiment analysis into positive and negative categories and increase the prediction accuracy of the polarity analysis using the pretrained IMDB review data set. First, the text classification algorithm related to sentiment analysis adopts the popular machine learning algorithms such as NB (naive bayes), SVM (support vector machines), XGboost, RF (random forests), and Gradient Boost as comparative models. Second, deep learning has demonstrated discriminative features that can extract complex features of data. Representative algorithms are CNN (convolution neural networks), RNN (recurrent neural networks), LSTM (long-short term memory). CNN can be used similarly to BoW when processing a sentence in vector format, but does not consider sequential data attributes. RNN can handle well in order because it takes into account the time information of the data, but there is a long-term dependency on memory. To solve the problem of long-term dependence, LSTM is used. For the comparison, CNN and LSTM were chosen as simple deep learning models. In addition to classical machine learning algorithms, CNN, LSTM, and the integrated models were analyzed. Although there are many parameters for the algorithms, we examined the relationship between numerical value and precision to find the optimal combination. And, we tried to figure out how the models work well for sentiment analysis and how these models work. This study proposes integrated CNN and LSTM algorithms to extract the positive and negative features of text analysis. The reasons for mixing these two algorithms are as follows. CNN can extract features for the classification automatically by applying convolution layer and massively parallel processing. LSTM is not capable of highly parallel processing. Like faucets, the LSTM has input, output, and forget gates that can be moved and controlled at a desired time. These gates have the advantage of placing memory blocks on hidden nodes. The memory block of the LSTM may not store all the data, but it can solve the CNN's long-term dependency problem. Furthermore, when LSTM is used in CNN's pooling layer, it has an end-to-end structure, so that spatial and temporal features can be designed simultaneously. In combination with CNN-LSTM, 90.33% accuracy was measured. This is slower than CNN, but faster than LSTM. The presented model was more accurate than other models. In addition, each word embedding layer can be improved when training the kernel step by step. CNN-LSTM can improve the weakness of each model, and there is an advantage of improving the learning by layer using the end-to-end structure of LSTM. Based on these reasons, this study tries to enhance the classification accuracy of movie reviews using the integrated CNN-LSTM model.

• Journal of the Korean Society of Groundwater Environment
• /
• v.7 no.3
• /
• pp.103-115
• /
• 2000

The formation of brown-colored precipitates is one of the serious problems frequently encountered in the development and supply of groundwater in Korea, because by it the water exceeds the drinking water standard in terms of color. taste. turbidity and dissolved iron concentration and of often results in scaling problem within the water supplying system. In groundwaters from the Pajoo area, brown precipitates are typically formed in a few hours after pumping-out. In this paper we examine the process of the brown precipitates' formation using the equilibrium thermodynamic and kinetic approaches, in order to understand the origin and geochemical pathway of the generation of turbidity in groundwater. The results of this study are used to suggest not only the proper pumping technique to minimize the formation of precipitates but also the optimal design of water treatment methods to improve the water quality. The bed-rock groundwater in the Pajoo area belongs to the Ca-$HCO_3$type that was evolved through water/rock (gneiss) interaction. Based on SEM-EDS and XRD analyses, the precipitates are identified as an amorphous, Fe-bearing oxides or hydroxides. By the use of multi-step filtration with pore sizes of 6, 4, 1, 0.45 and 0.2 $\mu\textrm{m}$, the precipitates mostly fall in the colloidal size (1 to 0.45 $\mu\textrm{m}$) but are concentrated (about 81%) in the range of 1 to 6 $\mu\textrm{m}$in teams of mass (weight) distribution. Large amounts of dissolved iron were possibly originated from dissolution of clinochlore in cataclasite which contains high amounts of Fe (up to 3 wt.%). The calculation of saturation index (using a computer code PHREEQC), as well as the examination of pH-Eh stability relations, also indicate that the final precipitates are Fe-oxy-hydroxide that is formed by the change of water chemistry (mainly, oxidation) due to the exposure to oxygen during the pumping-out of Fe(II)-bearing, reduced groundwater. After pumping-out, the groundwater shows the progressive decreases of pH, DO and alkalinity with elapsed time. However, turbidity increases and then decreases with time. The decrease of dissolved Fe concentration as a function of elapsed time after pumping-out is expressed as a regression equation Fe(II)=10.l exp(-0.0009t). The oxidation reaction due to the influx of free oxygen during the pumping and storage of groundwater results in the formation of brown precipitates, which is dependent on time, $Po_2$and pH. In order to obtain drinkable water quality, therefore, the precipitates should be removed by filtering after the stepwise storage and aeration in tanks with sufficient volume for sufficient time. Particle size distribution data also suggest that step-wise filtration would be cost-effective. To minimize the scaling within wells, the continued (if possible) pumping within the optimum pumping rate is recommended because this technique will be most effective for minimizing the mixing between deep Fe(II)-rich water and shallow $O_2$-rich water. The simultaneous pumping of shallow $O_2$-rich water in different wells is also recommended.

• Korean Journal of Poultry Science
• /
• v.7 no.2
• /
• pp.54-76
• /
• 1980

A survey was conducted to determine the status of health and productivity of poultry farms in Korea. Area included Was Kyunggido where exist nearly 50% of national poultry population. From this area, 41 layer and 34 broiler farms covering 21 Countries were selected randomly for the survey. When farms were divided in the operation size, 95.1% of layer and 82.3% of broiler farms were classified as business or industrial level while the rest were managed in a small scale as part time job. Generally layer farms had been established much earlier than broiler farms. Geographically 10.7% of layer farms were sited near the housing area such as field foreast and rice field. No farms were located near the seashore. The distance from one farm from the other was very close, being 80% of the farms within the distance of 1km and as many as 28% of the farms within loom. This concentrated poultry farming in a certain area created serious problems for the sanitation and preventive measures, especially in case of outbreak of infectious diseases. Average farm size was 5,016${\times}$3.3㎡ for layers and 1,037${\times}$3.3㎡ for broilers. 89.5% of layer ana 70.6% of broiler farms owned the land for farming while the rest were on lease. In 60% of layer farms welters were employed for farming while in the rest their own labour was used. Majority of farms were equipped poorly for taking necessary practice of hygiene and sanitation. The amount of disinfectant used by farms was considerably low. As many as 97.6% of lave. farms were practised with Newcastle(ND) and fowl pox(F$.$pox) vaccine, whereas only 43.6% and 5.1% of broiler farms were practised with ND and F$.$pox vaccine, respectively. In 17-32.7% of farms ND vaccine was used less than twice until 60 days of age and in only 14.6% of farms adult birds were vaccinated every 4months. Monthly expense for preventive measures was over 200,000W in 32% of farms. Only 4.9-2.7% of vaccine users were soaking advice from veterinarians before practising vaccination, 85% of the users trusted the efficacy of the vaccines. Selection of medicine was generally determined by the farm owner rather than by veterinarans on whom 33.3% of farms were dependant. When diseases outbroke, 49.3% of farms called for veterinary hospital and the rest were handled by their own veterinarians, salesmen or professionals. Approximately 70% of farms were satisfied with the diagnosis made by the veterinarians. Frequency of disease outbreaks varied according to the age and type of birds. The livabilities of layers during the period of brooding, rearing ana adultwere 90.5, 98.9 and 75.2%, respectively while the livalibility of broilers until marketing was 92.2%. In layers, average culling age, was 533.3 day and hen housed eggs were 232.7. Average feed conversion rates of layers and broilers were 3.30 and 2.48, respectively. Those figures were considerably higher than anticipated but still far lower than those in developed countries.

• Journal of Intelligence and Information Systems
• /
• v.19 no.1
• /
• pp.35-55
• /
• 2013

Information technology is a critical resource necessary for any company hoping to support and realize its strategic goals, which contribute to growth promotion and sustainable development. The selection of information technology and its strategic use are imperative for the enhanced performance of every aspect of company management, leading a wide range of companies to have invested continuously in information technology. Despite researchers, managers, and policy makers' keen interest in how information technology contributes to organizational performance, there is uncertainty and debate about the result of information technology investment. In other words, researchers and managers cannot easily identify the independent factors that can impact the investment performance of information technology. This is mainly owing to the fact that many factors, ranging from the internal components of a company, strategies, and external customers, are interconnected with the investment performance of information technology. Using an agent-based simulation technique, this research extracts factors expected to affect investment performance on information technology, simplifies the analyses of their relationship with economic modeling, and examines the performance dependent on changes in the factors. In terms of economic modeling, I expand the model that highlights the way in which product quality moderates the relationship between information technology investments and economic performance (Thatcher and Pingry, 2004) by considering the cost of information technology investment and the demand creation resulting from product quality enhancement. For quality enhancement and its consequences for demand creation, I apply the concept of information quality and decision-maker quality (Raghunathan, 1999). This concept implies that the investment on information technology improves the quality of information, which, in turn, improves decision quality and performance, thus enhancing the level of product or service quality. Additionally, I consider the effect of word of mouth among consumers, which creates new demand for a product or service through the information diffusion effect. This demand creation is analyzed with an agent-based simulation model that is widely used for network analyses. Results show that the investment on information technology enhances the quality of a company's product or service, which indirectly affects the economic performance of that company, particularly with regard to factors such as consumer surplus, company profit, and company productivity. Specifically, when a company makes its initial investment in information technology, the resultant increase in the quality of a company's product or service immediately has a positive effect on consumer surplus, but the investment cost has a negative effect on company productivity and profit. As time goes by, the enhancement of the quality of that company's product or service creates new consumer demand through the information diffusion effect. Finally, the new demand positively affects the company's profit and productivity. In terms of the investment strategy for information technology, this study's results also reveal that the selection of information technology needs to be based on analysis of service and the network effect of customers, and demonstrate that information technology implementation should fit into the company's business strategy. Specifically, if a company seeks the short-term enhancement of company performance, it needs to have a one-shot strategy (making a large investment at one time). On the other hand, if a company seeks a long-term sustainable profit structure, it needs to have a split strategy (making several small investments at different times). The findings from this study make several contributions to the literature. In terms of methodology, the study integrates both economic modeling and simulation technique in order to overcome the limitations of each methodology. It also indicates the mediating effect of product quality on the relationship between information technology and the performance of a company. Finally, it analyzes the effect of information technology investment strategies and information diffusion among consumers on the investment performance of information technology.

• Journal of the Society of Cosmetic Scientists of Korea
• /
• v.41 no.4
• /
• pp.361-373
• /
• 2015

This study was carried out to evaluate the antioxidant and antibacterial activities of Glycyrriza uralensis Fisher (Jecheon, Korea) extracts obtained by various extraction conditions (85% ethanol, heating temperatures and times), and to establish the optimal extraction condition of G. uralensis for the application as cosmetic ingredients. The extracts obtained under different conditions were concentrated and made in the powdered (sample-1) and were the crude extract solutions without concentration (sample-2). The antioxidant effects were determined by free radical scavenging activity ($FSC_{50}$), ROS scavenging activity ($OSC_{50}$), and cellular protective effects. Antibacterial activity was determined by minimum inhibitory concentration (MIC) on human skin flora. DPPH free radical scavenging activity of sample-1 ($100{\mu}g/mL$) was 10% higher in group extracted for 6 h than 12 h, but sample-2 didn't show any significant differences. The extraction yield extracted with same temperature for 12 h was 2.6 times higher than 6 h, but total flavonoid content was 1.1 times higher. These results indicated that total flavonoid content hardly increased with increasing extraction time. Free radical scavenging activity, ROS scavenging activity and cellular protective effects were not dependent on the yield of extraction, but total flavonoid content of extraction. Antibacterial activity on three skin flora (S. aureus, B. subtilis, P. acnes)of sample-1 in different extraction conditions were evaluated on same concentration, and the group extracted at 25 and $40^{\circ}C$ showed 16 times higher than methyl paraben ($2,500{\mu}g/mL$). In conclusion, 85% ethanol extracts of G. uralensis extracted at $40^{\circ}C$ for 6 h showed the highest antioxidant and antibacterial activity. These results indicate that the extraction condition is important to be optimized by comprehensive evaluation of extraction yield with various conditions, yield of active component, and activity test with concentrations, and activity of 100% extract, for manufacturing process of products.

• Journal of the Society of Cosmetic Scientists of Korea
• /
• v.34 no.3
• /
• pp.233-244
• /
• 2008

In this study, the antioxidative effects, inhibitory effects on elastase and tyrosinase, and component analysis of Psidium guajava leaf extracts were investigated. The free radical (1,1-diphenyl-2-picrylhydrazyl, DPPH) scavenging activities $(FSC_{50})$ of extract/fractions of Psidium guajava leaf were in the order: 50% ethanol extract $(7.05{\mu}g/mL)$ < ethyl acetate fraction $(3.36{\mu}g/mL)$ < deglycosylated flavonoid aglycone fraction $(3.24{\mu}g/mL)$. Reactive oxygen species (ROS) scavenging activities $(OSC_{50})$ of some Psidium guajava leaf extracts on ROS generated in $Fe^{3+}-EDTA/H_2O_2$ system were investigated using the luminol-dependent chemiluminescence assay. The order of ROS scavenging activities were 50% ethanol extract $(OSC_{50},\;2.17{\mu}g/mL)$ < ethyl acetate fraction $(0.64{\mu}g/mL)$ < deglycosylated flavonoid aglycone fraction $(3.39{\mu}g/mL)$. Aglycone fraction showed the most prominent ROS scavenging activity. The protective effects of extract/fractions of Psidium guajava leaf on the rose-bengal sensitized photohemolysis of human erythrocytes were investigated. The Psidium guajava leaf extracts suppressed photohemolysis in a concentration dependent manner $(1{\sim}10{\mu}g/mL)$, particularly deglycosylated flavonoid aglycone fraction exhibited the most prominent celluar protective effect ${\tau}_{50}\;107.5min\;at\;1{\mu}g/mL)$. Aglycone fraction obtained from the deglycosylation reaction of ethyl acetate fraction among the Psidium guajava leaf extracts, showed 1 band in TLC and 1 peak in HPLC experiments (360 nm). One component was identified as quercetin. TLC chromatogram of ethyl acetate fraction of Psidium guajava leaf extract revealed 5 bands and HPLC chromatogram showed 5 peaks, which were identified as quercetin 3-O-gentobioside (10.32%) , quercetin 3-O-${\beta}$-D-glucoside (isoquercitin, 13.30%), quercetin 3-O-${\beta}$-D-galactoside (hyperin, 11.34%), quercetin 3-O-${\alpha}$-L-arabinoside (guajavarin, 19.70%), quercetin 3-O-${\beta}$-L-rhamnoside (quercitrin, 45.33%) in the order of elution time. The inhibitory effect of Psidium guajava leaf extracts on tyrosinase were investigated to assess their whitening efficacy. Finally, their anti-elastase activities were measured to predict the anti-wrinkle efficacy in the human skin. Inhibitory effects $(IC_{50})$ on tyrosinase of some Psidium guajava leaf extracts was 50% ethanol extract $(149.67{\mu}g/mL)$ < ethylacetate fraction $(30.67{\mu}g/mL)$ < deglycosylated aglycone fraction $(17.10{\mu}g/mL)$. Inhibitory effects $(IC_{50})$ on elastase of some Psidium guajava leaf extracts was 50% ethanol extract $(6.60{\mu}g/mL)$ < deglycosylated aglycone fraction $(5.66{\mu}g/mL)$ < ethylacetate fraction $(3.44{\mu}g/mL)$. These results indicate that extract/fractions of Psidium guajava leaf can function as antioxidants in bioloigcal systems, particularly skin exposed to UV radiation by scavenging $^1O_2$ and other ROS, and protect cellular membranes against ROS. And component analysis of Psidium guajava leaf extract and inhibitory activity on elastase of the aglycone fraction could be applicable to new functional cosmetics for smoothing wrinkles.

• Korean Journal of Heritage: History & Science
• /
• v.52 no.3
• /
• pp.172-187
• /
• 2019

Knowledge of intangible cultural heritage is usually disseminated through word-of-mouth and actions rather than written records. Thus, people assemble to teach others about it and form communities. Accordingly, to understand and spread information about intangible cultural heritage properly, it is necessary to understand not only their attributes but also a community's relational characteristics. Community members include specialized transmitters who work under the auspices of institutions, and general transmitters who enjoy intangible cultural heritage in their daily lives. They converse about intangible cultural heritage in close relationships. However, to date, research has focused only on professionals. Thus, this study focused on the roles of general transmitters of intangible cultural heritage information by investigating intangible cultural heritage communities centering around Jeongseon Arirang; a social network analysis was performed. Regarding the research objectives presented in the introduction, the main findings of the study are summarized as follows. First, there were 197 links between 74 members of the Jeongseon Arirang Transmission Community. One individual had connections with 2.7 persons on average, and all were connected through two steps in the community. However, the density and the clustering coefficient were low, 0.036 and 0.32, respectively; therefore, the cohesiveness of this community was low, and the relationships between the members were not strong. Second, 'Young-ran Yu', 'Nam-gi Kim' and 'Gil-ja Kim' were found to be the prominent figures of the Jeongseon Arirang Transmission Community, and the central structure of the network was concentrated around these three individuals. Being located in the central structure of the network indicates that a person is popular and ranked high. Also, it means that a person has an advantage in terms of the speed and quantity of the acquisition of information and resources, and is in a relatively superior position in terms of bargaining power. Third, to understand the replaceability of the roles of Young-ran Yu, Nam-gi Kim, and Gil-ja Kim, who were found to be the major figures through an analysis of the central structure, structural equivalence was profiled. The results of the analysis showed that the positions and roles of Young-ran Yu, Nam-gi Kim, and Gil-ja Kim were unrivaled and irreplaceable in the Jeongseon Arirang Transmission Community. However, considering that these three members were in their 60s and 70s, it seemed that it would be necessary to prepare measures for the smooth maintenance and operation of the community. Fourth, to examine the subgroup hidden in the network of the Jeongseon Arirang Transmission Community, an analysis of communities was conducted. A community refers to a subgroup clearly differentiated based on modularity. The results of the analysis identified the existence of four communities. Furthermore, the results of an analysis of the central structure showed that the communities were formed and centered around Young-ran Yu, Hyung-jo Kim, Nam-gi Kim, and Gil-ja Kim. Most of the transmission TAs recommended by those members, students who completed a course, transmission scholarship holders, and the general members taught in the transmission classes of the Jeongseon Arirang Preservation Society were included as members of the communities. Through these findings, it was discovered that it is possible to maintain the transmission genealogy, making an exchange with the general members by employing the present method for the transmission of Jeongseon Arirang, the joint transmission method. It is worth paying attention to the joint transmission method as it overcomes the demerits of the existing closed one-on-one apprentice method and provides members with an opportunity to learn their masters' various singing styles. This study is significant for the following reasons: First, by collecting and examining data using a social network analysis method, this study analyzed phenomena that had been difficult to investigate using existing statistical analyses. Second, by adopting a different approach to the previous method in which the genealogy was understood, looking at oral data, this study analyzed the structures of the transmitters' relationships with objective and quantitative data. Third, this study visualized and presented the abstract structures of the relationships among the transmitters of intangible cultural heritage information on a 2D spring map. The results of this study can be utilized as a baseline for the development of community-centered policies for the protection of intangible cultural heritage specified in the UNESCO Convention for the Safeguarding of Intangible Cultural Heritage. To achieve this, it would be necessary to supplement this study through case studies and follow-up studies on more aspects in the future.

• Korean Journal of Heritage: History & Science
• /
• v.39
• /
• pp.351-378
• /
• 2006

With the development of media, the methods for the documentation of intangible cultural heritage have been also developed and diversified. As well as the previous analogue ways of documentation, the have been recently applying new multi-media technologies focusing on digital pictures, sound sources, movies, etc. Among the new technologies, the documentation of intangible cultural heritage using the method of 'Motion Capture' has proved itself prominent especially in the fields that require three-dimensional documentation such as dances and performances. Motion Capture refers to the documentation technology which records the signals of the time varing positions derived from the sensors equipped on the surface of an object. It converts the signals from the sensors into digital data which can be plotted as points on the virtual coordinates of the computer and records the movement of the points during a certain period of time, as the object moves. It produces scientific data for the preservation of intangible cultural heritage, by displaying digital data which represents the virtual motion of a holder of an intangible cultural heritage. National Research Institute of Cultural Properties (NRICP) has been working on for the development of new documentation method for the Important Intangible Cultural Heritage designated by Korean government. This is to be done using 'motion capture' equipments which are also widely used for the computer graphics in movie or game industries. This project is designed to apply the motion capture technology for 3 years- from 2005 to 2007 - for 11 performances from 7 traditional dances of which body gestures have considerable values among the Important Intangible Cultural Heritage performances. This is to be supported by lottery funds. In 2005, the first year of the project, accumulated were data of single dances, such as Seungmu (monk's dance), Salpuri(a solo dance for spiritual cleansing dance), Taepyeongmu (dance of peace), which are relatively easy in terms of performing skills. In 2006, group dances, such as Jinju Geommu (Jinju sword dance), Seungjeonmu (dance for victory), Cheoyongmu (dance of Lord Cheoyong), etc., will be documented. In the last year of the project, 2007, education programme for comparative studies, analysis and transmission of intangible cultural heritage and three-dimensional contents for public service will be devised, based on the accumulated data, as well as the documentation of Hakyeonhwadae Habseolmu (crane dance combined with the lotus blossom dance). By describing the processes and results of motion capture documentation of Salpuri dance (Lee Mae-bang), Taepyeongmu (Kang seon-young) and Seungmu (Lee Mae-bang, Lee Ae-ju and Jung Jae-man) conducted in 2005, this report introduces a new approach for the documentation of intangible cultural heritage. During the first year of the project, two questions have been raised. First, how can we capture motions of a holder (dancer) without cutoffs during quite a long performance? After many times of tests, the motion capture system proved itself stable with continuous results. Second, how can we reproduce the accurate motion without the re-targeting process? The project re-created the most accurate motion of the dancer's gestures, applying the new technology to drew out the shape of the dancers's body digital data before the motion capture process for the first time in Korea. The accurate three-dimensional body models for four holders obtained by the body scanning enhanced the accuracy of the motion capture of the dance.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.163-177
• /
• 2019

As smartphones are getting widely used, human activity recognition (HAR) tasks for recognizing personal activities of smartphone users with multimodal data have been actively studied recently. The research area is expanding from the recognition of the simple body movement of an individual user to the recognition of low-level behavior and high-level behavior. However, HAR tasks for recognizing interaction behavior with other people, such as whether the user is accompanying or communicating with someone else, have gotten less attention so far. And previous research for recognizing interaction behavior has usually depended on audio, Bluetooth, and Wi-Fi sensors, which are vulnerable to privacy issues and require much time to collect enough data. Whereas physical sensors including accelerometer, magnetic field and gyroscope sensors are less vulnerable to privacy issues and can collect a large amount of data within a short time. In this paper, a method for detecting accompanying status based on deep learning model by only using multimodal physical sensor data, such as an accelerometer, magnetic field and gyroscope, was proposed. The accompanying status was defined as a redefinition of a part of the user interaction behavior, including whether the user is accompanying with an acquaintance at a close distance and the user is actively communicating with the acquaintance. A framework based on convolutional neural networks (CNN) and long short-term memory (LSTM) recurrent networks for classifying accompanying and conversation was proposed. First, a data preprocessing method which consists of time synchronization of multimodal data from different physical sensors, data normalization and sequence data generation was introduced. We applied the nearest interpolation to synchronize the time of collected data from different sensors. Normalization was performed for each x, y, z axis value of the sensor data, and the sequence data was generated according to the sliding window method. Then, the sequence data became the input for CNN, where feature maps representing local dependencies of the original sequence are extracted. The CNN consisted of 3 convolutional layers and did not have a pooling layer to maintain the temporal information of the sequence data. Next, LSTM recurrent networks received the feature maps, learned long-term dependencies from them and extracted features. The LSTM recurrent networks consisted of two layers, each with 128 cells. Finally, the extracted features were used for classification by softmax classifier. The loss function of the model was cross entropy function and the weights of the model were randomly initialized on a normal distribution with an average of 0 and a standard deviation of 0.1. The model was trained using adaptive moment estimation (ADAM) optimization algorithm and the mini batch size was set to 128. We applied dropout to input values of the LSTM recurrent networks to prevent overfitting. The initial learning rate was set to 0.001, and it decreased exponentially by 0.99 at the end of each epoch training. An Android smartphone application was developed and released to collect data. We collected smartphone data for a total of 18 subjects. Using the data, the model classified accompanying and conversation by 98.74% and 98.83% accuracy each. Both the F1 score and accuracy of the model were higher than the F1 score and accuracy of the majority vote classifier, support vector machine, and deep recurrent neural network. In the future research, we will focus on more rigorous multimodal sensor data synchronization methods that minimize the time stamp differences. In addition, we will further study transfer learning method that enables transfer of trained models tailored to the training data to the evaluation data that follows a different distribution. It is expected that a model capable of exhibiting robust recognition performance against changes in data that is not considered in the model learning stage will be obtained.