• Korean Journal of Soil Science and Fertilizer
• /
• v.5 no.2
• /
• pp.1-38
• /
• 1972

This study, designed to establish a classification system of paddy soils and suitability groups on productivity and management of paddy land based on soil characteristics, has been made for the paddy soils on the Gimje-Mangyeong plains. The morphological, physical and chemical properties of the 15 paddy soil series found on these plains are briefly as follows: Ten soil series (Baeggu, Bongnam, Buyong, Gimje, Gongdeog, Honam, Jeonbug, Jisan, Mangyeong and Suam) have a B horizon (cambic B), two soil series (Geugrag and Hwadong) have a Bt horizon (argillic B), and three soil series (Gwanghwal, Hwagye and Sindab) have no B or Bt horizons. Uniquely, both the Bongnam and Gongdeog series contain a muck layer in the lower part of subsoil. Four soil series (Baeggu, Gongdeog, Gwanghwal and Sindab) generally are bluish gray and dark gray, and eight soil series (Bongnam, Buyong, Gimje, Honam, Jeonbug, Jisan, Mangyeong and Suam) are either gray or grayish brown. Three soil series (Geugrag, Hwadong and Hwagye), however, are partially gleyed in the surface and subsurface, but have a yellowish brown to brown subsoil or substrata. Seven soil series (Bongnam, Buyong, Geugrag, Gimje, Gongdeog, Honam and Hwadong) are of fine clayey texture, three soil series (Baeggu, Jeonbug and Jisan) belong to fine loamy and fine silty, three soil series (Gwanghwal, Mangyeong and Suam) to coarse loamy and coarse silty, and two soil series (Hwagye and Sindab) to sandy and sandy skeletal texture classes. The carbon content of the surface soil ranges from 0.29 to 2.18 percent, mostly 1.0 to 2.0 percent. The total nitrogen content of the surface soil ranges from 0.03 to 0.25 percent, showing a tendency to decrease irregularly with depth. The C/N ratio in the surface soil ranges from 4.6 to 15.5, dominantly from 8 to 10. The C/N ratio in the subsoil and substrata, however, has a wide range from 3.0 to 20.25. The soil reaction ranges from 4.5 to 8.0. All soil series except the Gwanghwal and Mangyeong series belong to the acid reaction class. The cation exchange cpacity in the surface soil ranges from 5 to 13 milliequivalents per 100 grams of soil, and in all the subsoil and substrata except those of a sandy texture, from 10 to 20 milliequivalents per 100 grams of soil. The base saturation of the soil series except Baeggu and Gongdeog is more than 60 percent. The active iron content of the surface soil ranges from 0.45 to 1.81 ppm, easily-reduceable manganese from 15 to 148 ppm, and available silica from 36 to 366 ppm. The iron and manganese are generally accumulated in a similar position (10 to 70cm. depth), and silica occurs in the same horizon with that of iron and manganese, or in the deeper horizons in the soil profile. The properties of each soil series extending from the sea shore towards the continental plains change with distance and they are related with distance (x) as follows: y(surface soil, clay content) = $$-0.2491x^2+6.0388x-1.1251$$ y(subsoil or subsurface soil, clay content) = $$-0.31646x^2+7.84818x-2.50008$$ y(surface soil, organic carbon content) = $$-0.0089x^2+0.2192x+0.1366$$ y(subsoil or subsurface soil, pH) = $$-0.0178x^2-0.04534x+8.3531$$ Soil profile development, soil color, depositional and organic layers, soil texture and soil reaction etc. are thought to be the major items that should be considered in a paddy soil classification. It was found that most of the soils belonging to the moderately well, somewhat poorly and poorly drained fine and medium textured soils and moderately deep fine textured soils over coarse materials, produce higher paddy yields in excess of 3,750 kg/ha. and most of the soils belonging to the coarse textured soils, well drained fine textured soils, moderately deep medium textured soils over coarse materials and saline soils, produce yields less than 3,750kg/ha. Soil texture of the profile, available soil depth, salinity and gleying of the surface and subsurface soils etc. seem to be the major factors determining rice yields, and these factors are considered when establishing suitability groups for paddy land. The great group, group, subgroup, family and series are proposed for the classification categories of paddy soils. The soil series is the basic category of the classification. The argillic horizon (Bt horizon) and cambic horizon (B horizon) are proposed as two diagnostic horizons of great group level for the determination of the morphological properties of soils in the classification. The specific soil characteristics considered in the group and subgroup levels are soil color of the profile (bluish gray, gray or yellowish brown), salinity (salic), depositonal (fluvic) and muck layers (mucky), and gleying of surface and subsurface soils (gleyic). The family levels are classified on the basis of soil reaction, soil texture and gravel content of the profile. The definitions are given on each classification category, diagnostic horizons and specific soil characteristics respectively. The soils on these plains are classified in eight subgroups and examined under the existing classification system. Further, the suitability group, can be divided into two major categories, suitability class and subclass. The soils within a suitability class are similar in potential productivity and limitation on use and management. Class 1 through 4 are distinguished from each other by combination of soil characteristics. Subclasses are divided from classes that have the same kind of dominant limitations such as slope(e), wettness(w), sandy(s), gravels(g), salinity(t) and non-gleying of the surface and subsurface soils(n). The above suitability classes and subclasses are examined, and the definitions are given. Seven subclasses are found on these plains for paddy soils. The classification and suitability group of 15 paddy soil series on the Gimje-Mangyeong plains may now be tabulated as follows.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.14 no.1
• /
• pp.1-14
• /
• 1978

To study the fluctuation of cold water in the East China Sea in summer heat budget of the Yellow Sea in winter was analysed based on the oceanographic and meteorological data compiled from 1951 to 1974. The maintain value of insolation was observed in December($160{\sim}190ly/day$), while the maximum in February ($250{\sim}260ly/day$). The range of the annual variation was found to be less than 50 ly/day. The value of the radiation term ($Q_s-Q_r-Q_h$) was remarkably small (mean 20 ly/day) in winter. It was negative value in December and January, and a positive value in February. The minimum total heat exchange from the sea ($Q_({h+c}$) was found value (471 ly/day) in February 1962, and the maximum (882 ly/day) in January 1963. The annual total heat exchange was minimum (588 ly/day) in 1962, and maximum (716 ly/day) in 1968. If the average deviation of mean water temperature at 50m depth layer were assumed to be the horizontal index ($C_h$) of colder water, $C_h$ is $C_h=\frac{{\Sigma}\limit_i\;A_i\;T_i}{{\Sigma}\limit_i\;A_i}$ where $A_i$ denotes the area of isothermal region and $T_i$ the value of deviation from mean sea water temperature. The vertical index ($C_v$) of cold water can be expressed similarly. Consequently the total index (C) of cold water equals to the sum of the two components, i.e. $C=C_h$＋$C_v$. Taking the deviation of mean sea surface temperature(T'w) in the third ten-day of Novembers in the Yellow Sea as the value of the initial condition, the following expressions are deduced : $C－T'w＝32.06 - 0.049$ $\;Q_T$ $C_h-T'w/2=12.20-0.019\;Q_T$ $C_v-T'w/2＝18.07-0.027\;Q_T$ where $Q_T$ denotes the total heat exchange of the sea. The correlation coefficients of these regression equations were found to be greater than 0.9. Heat budget was 588 ly/day in winter, and minimum water temperature of cold water was $18^{\circ}C$ in summer of 1962. The isotherm of $23^{\circ}C$ extended narrowly to southward up to $29^{\circ}N$ in summer. However, heat budget was 716 ly/day, and minimum water temperature of cold water was $12^{\circ}C$ in summer of 1968. The isotherm of $23^{\circ}C$ extended widely to southward up to $28^{\circ}30'N$ in summer. As a result of the present study, it may be concluded that the fluctuation of cold water of the East China Sea in summer can be predicted by the calculation of heat budget of the Yellow Sea in winter.

• Korean Journal of Soil Science and Fertilizer
• /
• v.6 no.1
• /
• pp.35-52
• /
• 1973

Red Yellow Soils occur very commonly in Korea and constitute the important upland soils of the country which are either presently being cultivated or are suitable for reclaiming and cultivating. These soils are distributed on rolling, moutain foot slopes, and terraces in the southern and western parts of the central districts of Korea, and are derived from granite, granite gneiss, old alluvium and locally from limestone and shale. This report is a summary of the morphology, physical and chemical characteristics of Red Yellow Soils. The data obtained from detailed soil surveys since 1964 are summarized as follows. 1. Red-Yellows Soils have an A, Bt, C profile. The A horizon is dark colored coarse loamy or fine loamy with the thin layer of organic matter. The B horizon is dominantly strong brown, reddish brown or yellowish red, clayey or fine loamy with clay cutans on the soil peds. The C horizon varies with parent materials, and is coarser texture and has a less developed structure than the Bt horizon. Soil depth, varied with relief and parent materials, is predominantly around 100cm. 2. In the physical characteristics, the clay content of surface soil is 18 to 35 percent, and of subsoil is 30 to 90 percent nearly two times higher than the surface soil. Bulk density is 1.2 to 1.3 in the surface soil and 1.3 to 1.5 in the subsoil. The range of 3-phase is mostly narrow with 45 to 50 percent in solid phase, 30 to 45 percent in liquid one, and 5 to 25 percent in gaseous state in the surface soil; and 50 to 60 solid, 35 to 45 percent liquid and less than 15 percent gaseous in the subsoil. Available soil moisture capacity ranges from 10 to 23 percent in the surface soil, and 5 to 16 percent in the subsoil. 3. Chemically, soil reaction is neutral to alkaline in soils derived from limestone or old fluviomarine deposits, and acid to strong acid in other ones. The organic matter content of surface soil varying considerably with vegetation, erosion and cultivation, ranges from 1.0 to 5.0 percent. The cation exchange capacity is 5 to 40 me/100gr soil and closely related to the content of organic matter, clay and silt. Base saturation is low, on the whole, due to the leaching of extractable cations, but is high in soils derived from limestone with high content of lime and magnesium. 4. Most of these soils mainly contain halloysite (a part of kaolin minerals), vermiculite (weathered mica), and illite, including small amount of chlorite, gibbsite, hematite, quartz and feldspar. 5. Characteristically they are similar to Red Yellow Podzolic Soils and a part of Reddish Brown Lateritic Soils of the United States, and Red Yellow Soils of Japan. According to USDA 7th Approximation, they can be classified as Udu Its or Udalfs, and in FAO classification system to Acrisols, Luvisols, and Nitosols.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.18 no.4
• /
• pp.297-308
• /
• 1985

Differences in lipid composition including fatty acid, lipid class, sterol and especially carotenoid between fleshes of wild and cultured prawn, Penaeus japonicus, were studied. Total lipids were extracted from the flesh during the spawning period and fractionated into two lipid classes of polar and nonpolar lipids by silicic acid column chromatography. The fatty acid composition of each lipid classes, total lipid (TL), nonpolar lipid (NL) and polar lipid (PL) were analyzed by gas liquid chromatography. The sterol and carotenoid composition of total lipids were determined by using thin layer chromatography, gas liquid chromatography and column chromatography using MgO-celite 545 and silicic acid-celite 545 as an absorbent, and by UV spectrophotometry. Total lipid contents of both fleshes from the wild and cultured prawn were about $2.0\%$ on average, but the content of the unsaponifiable matters in the cultured prawn (about $16.2\%$ in total lipid) showed a little higher than that of the wild prawn (about $13.9\%$ in total lipid) and the ratio of NL to PL in total lipid was 1:1.7. In the fatty acid composition of TL, the contents of $Cl_{16:0}\;and\;C_{20:3}$ fatty acids were higher in wild prawn than in cultured prawn, while the contents of $Cl_{18:1}\;and\;C_{20:5}$ fatty acids in cultured prawn were higher than those in wild prawn. The cultured prawn contained higher amounts of monoenoic acids and lower amounts of polyenoic acids than the wild prawn. In the fatty acid composition of NL, the wild prawn showed higher levels in $Cl_{18:0}\;and\;C_{20:1}$ fatty acid contents than the cultured prawn, while the cultured prawn contained much amout of $Cl_{16:0}\;and\;C_{18:1}$ fatty acids. On the other hand, the fatty acid composition of PL showed that $Cl_{16:1}\;and\;C_{17:1}$ fatty acid were higher in the wild prawn than in the cultured prawn, but in $Cl_{16:0}\;and\;C_{18:1}$ fatty acids, the levels were reversed. Consequently, the cultured prawn contained higher amount of monoenoic acids, and similar amounts of saturated acids and polyenoic acids to the wild prawn in NL. And the cultured prawn contained lower amount of monoenoic acids, and similar amounts of saturated acids and polyenoic acids to the wild prawn in PL. In sterol composition of both the wild and cultured prawn, the predominant sterol was cholesterol with the proportion of $78.7{\sim}88.9\%$ to the total sterol. In addition to the cholesterol, the other minor sterols such as 24-methylene cholesterol and sitosterol were detected. Total carotenoid content in flesh of the wild prawn was relatively higher than that of the cultured prawn marking 70 mg/100g of lipid in wild prawn and 40 mg/100 g of lipid in cultured prawn, respectively. The main carotenoids of the both prawns were astaxanthin($54.1{\sim}60.8%$), phoenicoxanthin ($16.5{sim}22.9%$),${\bata}-carotene\;(20.0{\sim}22.0%)$.

• Korean Journal of Poultry Science
• /
• v.38 no.1
• /
• pp.59-69
• /
• 2011

This study was conducted to investigate the effects of dietary supplementation of CS682, a fermentation product of Actinomycetae(Nocardia sp. CS682), and its commercial product DSC682$^{(R)}$ on the performance, blood parameters, intestinal microflora, and immune response in laying hens. Hy-Line Brown$^{(R)}$ laying hens were housed in two bird cages. Feeding trial lasted 5 wk under 16.5 h:7.5 h(L:D) lighting regimen. In Exp.1, a total of 480 birds of 86 wk old were assigned to four dietary treatments: Control, Antibiotics (6 ppm avilamycin), CS682-0.1 (CS682 0.1%) and CS682-1.0 (CS682 1.0% supplementation). Each treatment was replicated five times with 24 birds (or 12 cages) per replication. In Exp. 2, a total of 1,000 birds of 26 wk old were assigned to five dietary treatments: Control, Antibiotics (6 ppm avilamycin), DCS682-0.05 (DCS682 0.05%), DCS682-0.1 (DCS682 0.1%), DCS682-0.2 (DCS682 0.2% supplementation). Each treatment was replicated five times with 40 birds (or 20 cages) per replication. In Exp. 1, there were no significant differences among treatments in egg production, egg weight, broken & soft egg production, feed intake, and feed conversion ratio. Also, there were no significant differences among treatments in eggshell thickness, eggshell color and Haugh unit. However, eggshell strength was significantly (p<0.05) greater in CS682 and Antibiotics treatments than Control, and egg yolk color was significantly (p<0.05) higher in CS682-1.0 than Control. In Exp. 2, feed intake was significantly (p<0.05) lower in DSC682-0.05 than Control. Lightness(L) of Hunter Lab color of eggshell of DCS and Antibiotics treatments was significantly (p<0.05) lower than Control. Egg yolk color of DCS 0.1 and 0.2 treatments was significantly (p<0.05) higher than Control. Haugh unit increased significantly (p<0.05) in Antibiotics and DCS682-0.1 treatments. The immunoglobulin levels of plasma (IgG and IgA) and eggyolk (IgY) were not significantly affected by treatments. Antibiotics and CS682 or DCS682 treatments significantly (p<0.05 or 0.01) influenced some of the erythrocytes and leukocytes parameters in blood. In Exp.1, mean corpuscular volume (MCV) decreased by CS682 treatments and mean corpuscular hemoglobin (MCH) was highest in Antibiotics treatments. In Exp.2, the level of monocyte (MO) decreased in DCS682-0.10 and 0.20 treatments. The cfu of C. perfringens and S. typhimurium in small intestinal content were highest in Control and lowest in Antibiotics in both experiments. In Exp. 2, DSC682-0.05 and -0.1 treatments were highest and Antibiotic treatment was lowest in Lactobacilli spp. The results of the present layer experiments indicated that supplementation of 0.1~0.2% CS682 or DCS682 may increase eggshell strength, color of eggshell and eggyolk, Haugh unit, and control harmful intestinal microbes.

• Journal of the Korean Wood Science and Technology
• /
• v.13 no.1
• /
• pp.3-18
• /
• 1985

In Korea, a study on the anatomical features of pitch pine (pinus rigida Miller) branch wood through photo-microscopical method was reported in 1972 by Lee. Therefore, as a further study of Lee's on the anatomical features in branch wood of pinus rigida miller that grows in Korea, compression wood, opposite wood, and side wood were selected and treated for the purpose of comparing their structures revealed on cross and radial surface through scanning electron microscope in this study. The obtained results in this study were summarized as follows; 1. The trachied transition from earlywood to late wood is very gradual and the tracheids are nearly regular in both arrangement and size in compression wood but this transition in opposite wood and side wood is abrupt and the tracheids in opposite wood and side wood are less regular than those in compression wood. Also, the annual ring width of opposite wood is narrower than that of compression wood or side wood and the rays revealed on cross surface of side wood are more distinct than compression wood and opposite wood rays. 2. The tracheids of compression wood show roundish trends especially in earlywood but those of opposite wood and side wood show some angular trends. And intercellular space, helical cavity, and spiral check are present in both earlywood and latewood of compression wood but not present in opposite wood and side wood irrespective of earlywood and latewood. 3. The wall thickness of latewood tracheid is similar to that of earlywood tracheid in compression wood whereas the wall thickness of latewood tracheid is by far thicker than that of earlywood tracheid in opposite wood and side wood and the S3 layer of secondary wall is lack in compression wood tracheid unlike opposite wood and side wood tracheid. 4. The tracheids in compression wood are often distorted at their tips unlike those in opposite wood and side wood and the bordered pit in compression wood tracheid is located at the bottom of helical groove unlike that in opposite wood and side wood tracheid. 5. The bordered pits in radial wall of opposite wood and side wood tracheids are oval in shape but those of compression wood tracheids show some modified oval shape. 6. In earlywood of side wood, the small apertures of cross-field pits are roundish triangle to rectangle and the large one are fenestriform through the coalition of two small ones. However, the small apertures of cross-field pits are upright oval and the large ones are procumbent oval shape in earlywood of opposite wood and the apertures of cross-field pits in compression wood are tilted bifacial convex lens shape in earlywood and slit in late wood because of the border on tracheid side.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.5 no.2
• /
• pp.105-118
• /
• 2000

Anchored measurements (12.5 hr) of suspended sediment concentration and other hydrodynamic parameters were carried out at two stations located at the entrance to Hampyung Bay in summer (August 1999). Tidal variations in water temperature and salinity were in the range of 26.0-27.9$^{\circ}C$ and 30.9-31.5, respectively, indicating exchange offshore and offshore water mass. Active tidal mixing processes at the entrance appear to destroy the otherwise vertical stratification in temperature and salinity in spite of strong solar heating in summer. On the contrary, suspended sediment concentrations show a marked stratification with increasing concentrations toward bottom layer. Clastic particles in suspended sediments consist mostly of very fine to fine silt (4-16 ${\mu}$m) with a poorly-sorted value of 14.7-25.9 ${\mu}$m. However, at slack time with less turbulent energy, flocs larger than 40 ${\mu}$m are formed by cohesion and inter-collision of particles, resulting in a higher settling velocity. Strong ebb-dominated and weak flood dominated tidal currents, in the southwestern and the northeastern part, respectively, result in a seaward residual flow of -10${\sim}$-20 cm $s^{-1}$ at station H1 and a bayward residual flow less than 5.0 cm $s^{-1}$ at station H2. However, mean concentration of suspended sediments at station H1 is higher at flood (95.0-144.1 mg $1^{-1}$) than in ebb (75.8-120.9 mg $1^{-1}$). On the contrary, at the station H2, the trend is reversed with higher concentration at the ebb (84.7-158.4 mg $1^{-1}$) than that at the flood (53.0-107.9 mg $1^{-1}$). As a result, seaward net suspended sediment fluxes ($f_{s}$) are calculated to be -1.7 ${\sim}$-$15.610^{3}$ kg $m^{-2}$ $s^{-1}$ through the whole water column. However, the stations H1 and H2 show definitely different values of the flux with higher ones in the former than in the latter. Alternatively, depth-integrated net suspended sediment loads ($\c{Q}_{s}$) for one tidal cycle are also toward the offshore with ranges of 0.37${\times}$$10^{3}$ kg $m^{-1}$ and 0.21${\times}$$10^{3}$ kg $m^{-1}$, at station H1 and H2, respectively. This seaward transport of suspended sediment in summer suggests that summer-time erosion in the Hampyung muddy tidal flats is a rather exceptional phenomenon compared to the general deposition reported for many other tidal flats on the west coast of Korea.

• Journal of the Korean Institute of Traditional Landscape Architecture
• /
• v.36 no.3
• /
• pp.115-127
• /
• 2018

This study was attempted to find out the value of cultural assets through the clear diagnosis and prescription of the dead and weakness factors of the Population of Retusa Fringe Trees in Pyeongji-ri, Jinan(Natural Monument No. 214), The results are as follows. First, Since the designation of 13 natural monuments in 1968, since 1973, many years have passed since then. In particular, despite the removal of some of the buried soil during the maintenance process, such as retreating from the fence of the primary school after 2010, Second, The first and third surviving tree of the designated trees also have many branches that are dead, the leaves are dull, and the amount of leaves is small. vitality of tree is 'extremely bad', and the first branch has already been faded by a large number of branches, and the amount of leaves is considerably low this year, so that only two flowers are bloomed. The second is also in a 'bad'state, with small leaves, low leaf density, and deformed water. The largest number 1 in the world is added to the concern that the s coverd oil is assumed to be paddy soils. Third, It is found that the composition ratio of silt is high because it is known as '[silty loam(SiL)]'. In addition, the pH of the northern soil at pH 1 was 6.6, which was significantly different from that of the other soil. In addition, the organic matter content was higher than the appropriate range, which is considered to reflect the result of continuous application for protection management. Fourth, It is considered that the root cause of failure and growth of Jinan pyeongji-ri Population of Retusa Fringe Trees group is chronic syndrome of serious menstrual deterioration due to covered soil. This can also be attributed to the newly planted succession and to some of the deaths. Fifthly, It is urgent to gradually remove the subsoil part, which is estimated to be the cause of the initial damage. Above all, it is almost impossible to remove the coverd soil after grasping the details of the soil, such as clayey soil, which is buried in the rootstock. After removal of the coverd soil, a pestle is installed to improve the respiration of the roots and the ground with Masato. And the dead 4th dead wood and the 5th and 6th dead wood are the best, and the lower layer vegetation is mown. The viable neck should be removed from the upper surface, and the bark defect should undergo surgery and induce the development of blindness by vestibule below the growth point. Sixth, The underground roots should be identified to prepare a method to improve the decompression of the root and the respiration of the soil. It is induced by the shortening of rotten roots by tracing the first half of the rootstock to induce the generation of new roots. Seventh, We try mulching to suppress weed occurrence, trampling pressure, and soil moisturizing effect. In addition, consideration should be given to the fertilization of the foliar fertilizer, the injection of the nutrients, and the soil management of the inorganic fertilizer for the continuous nutrition supply. Future monitoring and forecasting plans should be developed to check for changes continuously.

• 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.

• Journal of Intelligence and Information Systems
• /
• v.27 no.4
• /
• pp.1-22
• /
• 2021

Aspect Based Sentiment Analysis (ABSA), which analyzes sentiment based on aspects that appear in the text, is drawing attention because it can be used in various business industries. ABSA is a study that analyzes sentiment by aspects for multiple aspects that a text has. It is being studied in various forms depending on the purpose, such as analyzing all targets or just aspects and sentiments. Here, the aspect refers to the property of a target, and the target refers to the text that causes the sentiment. For example, for restaurant reviews, you could set the aspect into food taste, food price, quality of service, mood of the restaurant, etc. Also, if there is a review that says, "The pasta was delicious, but the salad was not," the words "steak" and "salad," which are directly mentioned in the sentence, become the "target." So far, in ABSA, most studies have analyzed sentiment only based on aspects or targets. However, even with the same aspects or targets, sentiment analysis may be inaccurate. Instances would be when aspects or sentiment are divided or when sentiment exists without a target. For example, sentences like, "Pizza and the salad were good, but the steak was disappointing." Although the aspect of this sentence is limited to "food," conflicting sentiments coexist. In addition, in the case of sentences such as "Shrimp was delicious, but the price was extravagant," although the target here is "shrimp," there are opposite sentiments coexisting that are dependent on the aspect. Finally, in sentences like "The food arrived too late and is cold now." there is no target (NULL), but it transmits a negative sentiment toward the aspect "service." Like this, failure to consider both aspects and targets - when sentiment or aspect is divided or when sentiment exists without a target - creates a dual dependency problem. To address this problem, this research analyzes sentiment by considering both aspects and targets (Target-Aspect-Sentiment Detection, hereby TASD). This study detected the limitations of existing research in the field of TASD: local contexts are not fully captured, and the number of epochs and batch size dramatically lowers the F1-score. The current model excels in spotting overall context and relations between each word. However, it struggles with phrases in the local context and is relatively slow when learning. Therefore, this study tries to improve the model's performance. To achieve the objective of this research, we additionally used auxiliary loss in aspect-sentiment classification by constructing CNN(Convolutional Neural Network) layers parallel to existing models. If existing models have analyzed aspect-sentiment through BERT encoding, Pooler, and Linear layers, this research added CNN layer-adaptive average pooling to existing models, and learning was progressed by adding additional loss values for aspect-sentiment to existing loss. In other words, when learning, the auxiliary loss, computed through CNN layers, allowed the local context to be captured more fitted. After learning, the model is designed to do aspect-sentiment analysis through the existing method. To evaluate the performance of this model, two datasets, SemEval-2015 task 12 and SemEval-2016 task 5, were used and the f1-score increased compared to the existing models. When the batch was 8 and epoch was 5, the difference was largest between the F1-score of existing models and this study with 29 and 45, respectively. Even when batch and epoch were adjusted, the F1-scores were higher than the existing models. It can be said that even when the batch and epoch numbers were small, they can be learned effectively compared to the existing models. Therefore, it can be useful in situations where resources are limited. Through this study, aspect-based sentiments can be more accurately analyzed. Through various uses in business, such as development or establishing marketing strategies, both consumers and sellers will be able to make efficient decisions. In addition, it is believed that the model can be fully learned and utilized by small businesses, those that do not have much data, given that they use a pre-training model and recorded a relatively high F1-score even with limited resources.