• Journal of Korea Water Resources Association
• /
• v.53 no.12
• /
• pp.1159-1172
• /
• 2020

This study applied deep convolution neural network based on U-Net and SegNet using long period weather radar data to very short-term rainfall prediction. And the results were compared and evaluated with the translation model. For training and validation of deep neural network, Mt. Gwanak and Mt. Gwangdeoksan radar data were collected from 2010 to 2016 and converted to a gray-scale image file in an HDF5 format with a 1km spatial resolution. The deep neural network model was trained to predict precipitation after 10 minutes by using the four consecutive radar image data, and the recursive method of repeating forecasts was applied to carry out lead time 60 minutes with the pretrained deep neural network model. To evaluate the performance of deep neural network prediction model, 24 rain cases in 2017 were forecast for rainfall up to 60 minutes in advance. As a result of evaluating the predicted performance by calculating the mean absolute error (MAE) and critical success index (CSI) at the threshold of 0.1, 1, and 5 mm/hr, the deep neural network model showed better performance in the case of rainfall threshold of 0.1, 1 mm/hr in terms of MAE, and showed better performance than the translation model for lead time 50 minutes in terms of CSI. In particular, although the deep neural network prediction model performed generally better than the translation model for weak rainfall of 5 mm/hr or less, the deep neural network prediction model had limitations in predicting distinct precipitation characteristics of high intensity as a result of the evaluation of threshold of 5 mm/hr. The longer lead time, the spatial smoothness increase with lead time thereby reducing the accuracy of rainfall prediction The translation model turned out to be superior in predicting the exceedance of higher intensity thresholds (> 5 mm/hr) because it preserves distinct precipitation characteristics, but the rainfall position tends to shift incorrectly. This study are expected to be helpful for the improvement of radar rainfall prediction model using deep neural networks in the future. In addition, the massive weather radar data established in this study will be provided through open repositories for future use in subsequent studies.

• Korean Journal of Environment and Ecology
• /
• v.34 no.6
• /
• pp.558-572
• /
• 2020

The purpose of this study was to classify the types of degraded areas of Mt. Jirisan section in Baekdudaegan and survey the actual condition of each damage type to use it as basic data for the direction of the restoration of damaged areas according to damage type based on the vegetation information of reference ecosystem. The analysis of the Mt. Jirisan section's actual degraded conditions showed that the total number of patches of degraded areas was 57, and the number of patches and size of degraded areas was higher at the low average altitude and gentle slope. Grasslands (deserted lands) and cultivated areas accounted for a high portion of the damage types, indicating that agricultural land use was a major damage factor. The survey on the conditions of 14 degraded areas showed that the types of damage were classified into the grassland, cultivated area, restoration area, logged-off land, and bare ground. The analysis of the degree of disturbance (the ratio of annual and biennial herb, urbanized index, and disturbance index) by each type showed that the simple single-layer vegetation structure mostly composed of the herbaceous and the degree of disturbance were high in the grassland and cultivated land. The double-layer vegetation structure appeared in the restoration area where the pine seedlings were planted, and the inflow of naturalized plants was especially high compared to other degraded areas due to disturbances caused by the restoration project and the nearby hiking trails. Although the inflow of naturalized plants was low because of high altitude in bare ground, the proportion of annual and biennial herb was high, indicating that all surveyed degraded areas were in early succession stages. The stand ordination by type of damage showed the restoration area on the I-axis, cultivated area, grassland, logged-off land, and bare ground in that order, indicating the arrangement by the damage type. Moreover, the stand ordination of the degraded areas and reference ecosystem based on floristic variation showed a clear difference in species composition. This study diagnosed the status of each damage type based on the reference ecosystem information according to the ecological restoration procedure and confirmed the difference in species composition between the diagnosis result and the reference ecosystem. These findings can be useful basic data for establishing the restoration goal and direction in the future.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.67-86
• /
• 2020

Recommender Systems have been huge influence users and business more and more. Recently the importance of E-commerce has been reached rapid growth greatly in world-wide COVID-19 pandemic. Recommender system is the center of E-commerce lively. Top ranked E-commerce managers mentioned that recommender systems have a major influence on customer's purchase such as about 50% of Netflix, Amazon sales from their recommender systems. Most algorithms have been focused on improving accuracy of recommender system regardless of novelty, diversity, serendipity etc. Recommender systems with only high accuracy cannot satisfy business long-term profit because of generating sales polarization. In addition, customers do not experience enjoyment of shopping from only focusing accuracy recommender system because customer's preference is changed constantly. Therefore, recommender systems with various values need to be developed for user's high satisfaction. Reranking is the most useful methodology to realize diversity of recommender system. In this paper, diversity of recommender system is represented through constructing high similarity with users who have different preference using each user's purchased item's category algorithm. It is distinguished from past research approach which is changing the algorithm of recommender system without user's diversity preference level. We tried to discover user's diversity preference level and observed the results how the effect was different according to user's diversity preference level. In addition, graph-based recommender system was used to show diversity through user's network, not collaborative filtering. In this paper, Amazon Grocery and Gourmet Food data was used because the low-involvement product, such as habitual product, foods, low-priced goods etc., had high probability to show customer's diversity. First, a bipartite graph with users and items simultaneously is constructed to make graph-based recommender system. However, each users and items unipartite graph also need to be established to show diversity of recommender system. The weight of each unipartite graph has played crucial role changing Jaccard Distance of item's category. We can observe two important results from the user's unipartite network. First, the user's diversity preference level is observed from the network and second, dissimilar users can be discovered in the user's network. Through the research process, diversity of recommender system is presented highly with small accuracy loss and optimalization for higher accuracy is possible controlling diversity ratio. This paper has three important theoretical points. First, this research expands recommender system research for user's satisfaction with various values. Second, the graph-based recommender system is developed newly. Third, the evaluation indicator of diversity is made for diversity. In addition, recommender systems are useful for corporate profit practically and this paper has contribution on business closely. Above all, business long-term profit can be improved using recommender system with diversity and the recommender system can provide right service according to user's diversity level. Lastly, the corporate selling low-involvement products have great effect based on the results.

• Korean Journal of Environmental Biology
• /
• v.39 no.2
• /
• pp.195-206
• /
• 2021

In 2020, South Korea initiated research and development of a longitudinal connectivity evaluation between upstream and downstream based on stream ecosystem health. This study analyzed the migration of upstream and downstream migratory fish species, fish distribution characteristics, trophic guilds, tolerance guilds, and species composition changes from 2015 to 2020 at Songrim weir in Yeongok stream, where the cross-structure of an ice harbor-type fishway for fish movement was recently improved. A total of 5,136 fish, including 36 species, were collected and three major migratory fishes were identified, namely, Tribolodon hakonensis, Plecoglossus altivelis altivelis, and Oncorhynchus keta. According to the comparative analysis before (Pre-I) and after (Post-I) improvement of the fishway, the relative abundance of primary freshwater fish increased in the upstream section, while the number of migratory fishes decreased. The fish species that used the fishway in the Songrim weir were Tribolodon hakonensis (58.4%) and Plecoglossus altivelis altivelis(11.8%). According to the Wilcoxon Signed-Rank Test migratory fish showed a statistically significant difference (p<0.05) in the upstream and downstream, showing a biological improvement effect of the crossstructure. On the other hand, the annual change of migratory fish based on the MannKendall trend test did not significantly increase or decrease (p>0.05). Therefore, in the fish passage improvement project, it is necessary not only for physical, hydrological, and structural tests, but also for pre- and post-biological tests on the use and improvement effect of fishway.

• Korean Journal of Environmental Biology
• /
• v.39 no.1
• /
• pp.108-118
• /
• 2021

The purpose of this study was to provide baseline ecological data for the conservation of the Manchurian trout habitat through the investigation of the growth status of Brachymystax lenok tsinlingensis, and Pearson's correlation analysis (PCA) between the B. lenok tsinlingensis population and the use of the land around Gyebangcheon stream. Sampling was conducted twice in July, September, and October 2018. During the July and September surveys, 882 individuals belonging to 13 species from six families were collected. The dominant species was Rhynchocypris kumgangensis and the subdominant species was Zacco koreanus. The total number of B. lenok tsinlingensis collected was 99. The results of the length-weight relationship in the B. lenok tsinlingensis population were analyzed with a regression coefficient b value of 3.1272 and a condition factor (k) value of 0.0006. Therefore, the growth condition of B. lenok tsinlingensis was regarded as fairly good. The QHEI(Qualitative habitat evaluation index) value in the B. lenok tsinlingensis habitat was 119.5(±0.5)-153.5(±0.5), indicating optimal-suboptimal conditions. As a result of the HIS (Habitat suitability index) analysis, it was confirmed that the optimal habitat for B. lenok tsinlingensis was 0.45-0.55m and >1 m in water depth, 0.55-0.65 m s^-1 in water velocity, and boulder in the substrate. The ratio of the land use in this study site was analyzed as 66.26-96.31% for forest and grassland areas, 0.00-23.79% for agricultural areas, 0.00-4.19% for urbanized areas, and 3.69-8.87% for others. Correlation analysis of the number of B. lenok tsinlingensis and various factors revealed statistically significant correlations between QHEI and forest and grassland areas, agricultural areas, and urbanized areas.

• Journal of Intelligence and Information Systems
• /
• v.27 no.2
• /
• pp.17-32
• /
• 2021

Speaker recognition is generally divided into speaker identification and speaker verification. Speaker recognition plays an important function in the automatic voice system, and the importance of speaker recognition technology is becoming more prominent as the recent development of portable devices, voice technology, and audio content fields continue to expand. Previous speaker recognition studies have been conducted with the goal of automatically determining who the speaker is based on voice files and improving accuracy. Speech is an important sociolinguistic subject, and it contains very useful information that reveals the speaker's attitude, conversation intention, and personality, and this can be an important clue to speaker recognition. The final ending used in the speaker's speech determines the type of sentence or has functions and information such as the speaker's intention, psychological attitude, or relationship to the listener. The use of the terminating ending has various probabilities depending on the characteristics of the speaker, so the type and distribution of the terminating ending of a specific unidentified speaker will be helpful in recognizing the speaker. However, there have been few studies that considered speech in the existing text-based speaker recognition, and if speech information is added to the speech signal-based speaker recognition technique, the accuracy of speaker recognition can be further improved. Hence, the purpose of this paper is to propose a novel method using speech style expressed as a sentence-final ending to improve the accuracy of Korean speaker recognition. To this end, a method called sentence sequencing that generates vector values by using the type and frequency of the sentence-final ending appearing in the utterance of a specific person is proposed. To evaluate the performance of the proposed method, learning and performance evaluation were conducted with a actual drama script. The method proposed in this study can be used as a means to improve the performance of Korean speech recognition service.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.22 no.4
• /
• pp.268-278
• /
• 2020

Flowering time has been put forward due to the recent abnormally warm winter, which often caused damages of flower buds by late frosts persistently. In the present study, cumulative chill unit and cumulative heat unit of Niitaka pear, which are required for releasing the endogenous dormancy and for flowering after breaking dormancy, respectively, were compared between flowering time prediction models used in South K orea. Observation weather data were collected at eight locations for the recent three years from 2018-2020. The dates of full bloom were also collected to determine the confidence level of models including DVR, mDVR and CD models. It was found that mDVR model tended to have smaller values (8.4%) of the coefficient of variation (cv) of chill units than any other models. The CD model tended to have a low value of cv (17.5%) for calculation of heat unit required to reach flowering after breaking dormancy. The mDVR model had the most accurate prediction of full bloom during the study period compared with the other models. The DVR model usually had poor skills in prediction of full bloom dates. In particular, the error of the DVR model was large especially in southern coastal areas (e.g., Ulju and Sacheon) where the temperature was warm. Our results indicated that the mDVR model had relatively consistent accuracy in prediction of full bloom dates over region and years of interest. When observation data for full bloom date are compiled for an extended period, the full bloom date can be predicted with greater accuracy improving the mDVR model further.

• Journal of The Korean Society of Grassland and Forage Science
• /
• v.40 no.4
• /
• pp.227-235
• /
• 2020

This study was conducted to investigate the growth characteristics of the shoot and roots and to analyse the morphological characteristics of roots of waterlogging resistant and susceptible maize inbred lines. Six maize inbred lines were treated with waterlogging for 10 days at V3, and the degree of leaf senescence was evaluated for waterlogging resistance. As a result of waterlogging resistance evaluation, KS85 was the most damaged inbred line with 3.33 senescence leaves and 5.54 degree, and KS141 was the least damaged inbred line with 1.33 senescence leaves and 3 degree. At 20 days after treatment, the effect of waterlogging stress on the shoot dry matter accumulation of KS85 and KS141 were decreased by 86.1% and 77.0%, respectively, compared to the control. Similarly, root dry matter accumulation of KS85 and KS141 were decreased by 77.6% and 65.0%. As a result of SEM photographs of the nodal roots of the two maize inbred lines, the thickness of cortex of KS141 was thicker than that of KS85, and the distortion of the cortex was observed in KS85 at 20 days after waterlogging. It was concluded that the thickness of cortex was related to maize waterlogging resistance.

• Journal of the Society of Cosmetic Scientists of Korea
• /
• v.46 no.4
• /
• pp.319-327
• /
• 2020

Particulate matters (PM) are small particulate pollution that decrease the function of skin barrier, which causes inflammatory skin diseases and extrinsic aging. In this study, we evaluated the effect of preventing the adherence of PMs from several cosmetic products applied to human skin using iron oxide black. The PM floating chamber consists of skin exposure area, PM inlet, floating power device, and an outlet so that PM can be naturally attached to the skin while floating in the chamber. The change in skin brightness according to the floating concentration of alternative fine dust was checked to confirm the optimal floating concentration conditions. The intensity difference (before-after intensity, Δ) before and after adhesion of iron oxide black was proportional to the amount of PM adhered. The anti-adherence effect of iron oxide black on five cosmetic products were evaluated through 20 each subjects by comparing the amount of iron oxide black adhered on the control and treatment. The difference in brightness before and after the iron oxide black attached to the skin was calculated and compared with the control group(p < 0.05). When over 150 mg of iron oxide black was adhered on the skin, the interference of intensity was low and clearly showed the skin adhered pattern. According to the application of the five cosmetics, the intensity difference was significantly lower than the control group. This means that depending on the product, it prevented the attachment of iron oxide black. This study is a safe and useful method to confirm the prevention of PM skin adherence. In conclusion, cosmetics can prevent the adherence of PM on the skin according to the formulation or ingredients characteristics.

• Journal of The Korean Society of Grassland and Forage Science
• /
• v.40 no.4
• /
• pp.236-243
• /
• 2020

This study was to evaluate the feed value of whole crop rice silage (WCRS) and to investigate a suitable ratio of the WCRS and concentrate by an analysis of rumen fermentation. A total of 6 treatments were used according to WCRS: concentrate ratio on in vitro rumen fermentation: T1 (100:0), T2 (60:40), T3 (40:60), T4 (20:80), T5 (10:90), and T6 (0:100). The ruminal pH, total gas emission, ammonia nitrogen, and volatile fatty acid (VFA) were determined as fermentation parameters. Total nutrients digestibility trial was conducted by 4 treatments according to WCRS: concentrate ratio at 40:60 (W40), 20:80 (W20), and 10:90 (W10), respectively. Feed value was analyzed according to AOAC (2019) and nutrient digestibility was calculated based on NRC (2001). The levels of crude protein (CP), crude fat, and neutral detergent fiber of the WCRS were 12.29%, 1.67%, and 59.79%, respectively. It was found to be 51.49% as a result of predicting the total digestible nutrient of WCRS using the NRC (2001) model. In vitro rumen fermentation, T4, T5, and T6 treatments showed a greater gas emission and total VFA concentration compared with other treatments (p<0.05). Acetate and acetate to propionate ratio of T4, T5, and T6 were significantly higher than other treatments (p<0.05). There was a significant difference in the level of propionate and butyrate according to the WCRS: concentrate ratio (p<0.05). The digestibility of dry matter and CP was significantly lower in W40 than in other treatments (p<0.05); however, there was no difference in W20 and W10. In conclusion, the 20:80 (WCRS: concentrate) is beneficial for stabilizing the rumen that does not inhibit rumen fermentation and nutrient digestion. This ratio might have a positive effect on the economics of farms as a valuable feed.