• Journal of Intelligence and Information Systems
• /
• v.25 no.4
• /
• pp.105-122
• /
• 2019

Dimensionality reduction is one of the methods to handle big data in text mining. For dimensionality reduction, we should consider the density of data, which has a significant influence on the performance of sentence classification. It requires lots of computations for data of higher dimensions. Eventually, it can cause lots of computational cost and overfitting in the model. Thus, the dimension reduction process is necessary to improve the performance of the model. Diverse methods have been proposed from only lessening the noise of data like misspelling or informal text to including semantic and syntactic information. On top of it, the expression and selection of the text features have impacts on the performance of the classifier for sentence classification, which is one of the fields of Natural Language Processing. The common goal of dimension reduction is to find latent space that is representative of raw data from observation space. Existing methods utilize various algorithms for dimensionality reduction, such as feature extraction and feature selection. In addition to these algorithms, word embeddings, learning low-dimensional vector space representations of words, that can capture semantic and syntactic information from data are also utilized. For improving performance, recent studies have suggested methods that the word dictionary is modified according to the positive and negative score of pre-defined words. The basic idea of this study is that similar words have similar vector representations. Once the feature selection algorithm selects the words that are not important, we thought the words that are similar to the selected words also have no impacts on sentence classification. This study proposes two ways to achieve more accurate classification that conduct selective word elimination under specific regulations and construct word embedding based on Word2Vec embedding. To select words having low importance from the text, we use information gain algorithm to measure the importance and cosine similarity to search for similar words. First, we eliminate words that have comparatively low information gain values from the raw text and form word embedding. Second, we select words additionally that are similar to the words that have a low level of information gain values and make word embedding. In the end, these filtered text and word embedding apply to the deep learning models; Convolutional Neural Network and Attention-Based Bidirectional LSTM. This study uses customer reviews on Kindle in Amazon.com, IMDB, and Yelp as datasets, and classify each data using the deep learning models. The reviews got more than five helpful votes, and the ratio of helpful votes was over 70% classified as helpful reviews. Also, Yelp only shows the number of helpful votes. We extracted 100,000 reviews which got more than five helpful votes using a random sampling method among 750,000 reviews. The minimal preprocessing was executed to each dataset, such as removing numbers and special characters from text data. To evaluate the proposed methods, we compared the performances of Word2Vec and GloVe word embeddings, which used all the words. We showed that one of the proposed methods is better than the embeddings with all the words. By removing unimportant words, we can get better performance. However, if we removed too many words, it showed that the performance was lowered. For future research, it is required to consider diverse ways of preprocessing and the in-depth analysis for the co-occurrence of words to measure similarity values among words. Also, we only applied the proposed method with Word2Vec. Other embedding methods such as GloVe, fastText, ELMo can be applied with the proposed methods, and it is possible to identify the possible combinations between word embedding methods and elimination methods.

• 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.

• Proceedings of the Korean Society of Near Infrared Spectroscopy Conference
• /
• 2001.06a
• /
• pp.1031-1031
• /
• 2001

The mammary gland is made up of remarkably sensitive tissue, which has the capability of producing a large volume of secretion, milk, under normal or healthy conditions. When bacteria enter the gland and establish an infection (mastitis), inflammation is initiated accompanied by an influx of white cells from the blood stream, by altered secretory function, and changes in the volume and composition of secretion. Cell numbers in milk are closely associated with inflammation and udder health. These somatic cell counts (SCC) are accepted as the international standard measurement of milk quality in dairy and for mastitis diagnosis. NIR Spectra of unhomogenized composite milk samples from 14 cows (healthy and mastitic), 7days after parturition and during the next 30 days of lactation were measured. Different multivariate analysis techniques were used to diagnose the disease at very early stage and determine how the spectral properties of milk vary with its composition and animal health. PLS model for prediction of somatic cell count (SCC) based on NIR milk spectra was made. The best accuracy of determination for the 1100-2500nm range was found using smoothed absorbance data and 10 PLS factors. The standard error of prediction for independent validation set of samples was 0.382, correlation coefficient 0.854 and the variation coefficient 7.63%. It has been found that SCC determination by NIR milk spectra was indirect and based on the related changes in milk composition. From the spectral changes, we learned that when mastitis occurred, the most significant factors that simultaneously influenced milk spectra were alteration of milk proteins and changes in ionic concentration of milk. It was consistent with the results we obtained further when applied 2DCOS. Two-dimensional correlation analysis of NIR milk spectra was done to assess the changes in milk composition, which occur when somatic cell count (SCC) levels vary. The synchronous correlation map revealed that when SCC increases, protein levels increase while water and lactose levels decrease. Results from the analysis of the asynchronous plot indicated that changes in water and fat absorptions occur before other milk components. In addition, the technique was used to assess the changes in milk during a period when SCC levels do not vary appreciably. Results indicated that milk components are in equilibrium and no appreciable change in a given component was seen with respect to another. This was found in both healthy and mastitic animals. However, milk components were found to vary with SCC content regardless of the range considered. This important finding demonstrates that 2-D correlation analysis may be used to track even subtle changes in milk composition in individual cows. To find out the right threshold for SCC when used for mastitis diagnosis at cow level, classification of milk samples was performed using soft independent modeling of class analogy (SIMCA) and different spectral data pretreatment. Two levels of SCC - 200 000 cells/$m\ell$ and 300 000 cells/$m\ell$, respectively, were set up and compared as thresholds to discriminate between healthy and mastitic cows. The best detection accuracy was found with 200 000 cells/$m\ell$ as threshold for mastitis and smoothed absorbance data: - 98% of the milk samples in the calibration set and 87% of the samples in the independent test set were correctly classified. When the spectral information was studied it was found that the successful mastitis diagnosis was based on reviling the spectral changes related to the corresponding changes in milk composition. NIRS combined with different ways of spectral data ruining can provide faster and nondestructive alternative to current methods for mastitis diagnosis and a new inside into disease understanding at molecular level.

• Korean Journal of Cognitive Science
• /
• v.15 no.2
• /
• pp.1-15
• /
• 2004

The structure of sensibility from motion was developed for the purpose of understanding relationship between sensibilities and physical factors to apply it to dynamic visual display. Seventy adjectives were collected by assessing adequacy to express sensibilities from motion and reporting sensibilities recalled from dynamic displays with achromatic color. Various motion displays with a moving single dot were rated according to the degree of sensibility corresponding to each adjective, on the basis of the Semantic Differential (SD) method. The results of assessment were analyzed by means of the factor analysis to reduce 70 words into 19 fundamental sensibilities from motion. The Multidimensional Scaling (MDS) technique constructed the sensibility space in motion, in which 19 sensibilities were scattered with two dimensions, active-passive and bright-dark Motion types systemically varied in kinematic factors were placed on the two-dimensional space of motion sensibility, in order to analyze important variables affecting sensibility from motion. Patterns of placement indicate that speed and both of cycle and amplitude in trajectories tend to partially determine sensibility. Although color and motion affected sensibility according to the in dimensions, it seemed that combination of motion and color made each have dominant effect individually in a certain sensibility dimension, motion to active-passive and color to bright-dark.

• The Korean Journal of Pesticide Science
• /
• v.8 no.3
• /
• pp.151-161
• /
• 2004

Three dimensional quantitative structure-activity relationships (3D-QSAR) studies for the protox inhibition activities against root and shoot of rice plant (Orysa sativa L.) and barnyardgrass (Echinochloa crus-galli) by a series of new A=3,4,5,6-tetrahydrophthalimino, B=3-chloro-4,5,6,7-tetrahydro-2H-indazolyl and C=3,4-dimethylmaleimino group, and R-group substituted on the phenyl ring in 1-(5-methyl-3-phenylisoxazolin-5-yl)methoxy-2chloro-4-fluorobenzene derivatives were performed using comparative molecular field analyses (CoMFA) methodology with Gasteiger-Huckel charge. Four CoMFA models for the protox inhibition activities against root and shoot of the two plants were generated using 46 molecules as training set and the predictive ability of the each models was evaluated against a test set of 8 molecules. And the statistical results of these models with combination (SIH) of standard field, indicator field and H-bond field showed the best predictability of the protox inhibition activities based on the cross-validated value $r^2_{cv.}$ $(q^2=0.635\sim0.924)$, conventional coefficient $(r^2_{ncv.}=0.928\sim0.977)$ and PRESS value $(0.091\sim0.156)$, respectively. The activities exhibited a strong correlation with steric $(74.3\sim87.4%)$, electrostatic $(10.10\sim18.5%)$ and hydrophobic $(1.10\sim8.30%)$ factors of the molecules. The steric feature of molecule may be an important factor for the activities. We founded that an novel selective and higher protox inhibitors between the two plants may be designed by modification of X-subsitutents for barnyardgrass based upon the results obtained from CoMFA analyses.

• Journal of Global Scholars of Marketing Science
• /
• v.18 no.2
• /
• pp.147-187
• /
• 2008

Online game researchers focus on the flow and factors influencing flow. Flow is conceptualized as an optimal experience state and useful explaining game experience in online. Many game studies focused on the customer loyalty and flow in playing online game, In showing specific game experience, however, it doesn't examine multidimensional experience process. Flow is not construct which show absorbing process, but construct which show absorbing result. Hence, Flow is not adequate to examine multidimensional experience of games. Online game is included in hedonic consumption. Hedonic consumption is a relatively new field of study in consumer research and it explores the consumption experience as a experiential view(Hirschman and Holbrook 1982). Hedonic consumption explores the consumption experience not as an information processing event but from a phenomenological of experiential view, which is a primarily subjective state. It includes various playful leisure activities, sensory pleasures, daydreams, esthetic enjoyment, and emotional responses. In online game experience, therefore, it is right to access through a experiential view of hedonic consumption. The objective of this paper was to make up for lacks in our understanding of online game experience by developing a framework for better insight into the hedonic experience of online game. We developed this framework by integrating and extending existing research in marketing, online game and hedonic responses. We then discussed several expectations for this framework. We concluded by discussing the results of this study, providing general recommendation and directions for future research. In hedonic response research, Lacher's research(1994)and Jongho lee and Yunhee Jung' research (2005;2006) has served as a fundamental starting point of our research. A common element in this extended research is the repeated identification of the four hedonic responses: sensory response, imaginal response, emotional response, analytic response. The validity of these four constructs finds in research of music(Lacher 1994) and movie(Jongho lee and Yunhee Jung' research 2005;2006). But, previous research on hedonic response didn't show that constructs of hedonic response have cause-effect relation. Also, although hedonic response enable to different by stimulus properties. effects of stimulus properties is not showed. To fill this gap, while largely based on Lacher(1994)' research and Jongho Lee and Yunhee Jung(2005, 2006)' research, we made several important adaptation with the primary goal of bringing the model into online game and compensating lacks of previous research. We maintained the same construct proposed by Lacher et al.(1994), with four constructs of hedonic response:sensory response, imaginal response, emotional response, analytical response. In this study, the sensory response is typified by some physical movement(Yingling 1962), the imaginal response is typified by images, memories, or situations that game evokes(Myers 1914), and the emotional response represents the feelings one experiences when playing game, such as pleasure, arousal, dominance, finally, the analytical response is that game player engaged in cognition seeking while playing game(Myers 1912). However, this paper has several important differences. We attempted to suggest multi-dimensional experience process in online game and cause-effect relation among hedonic responses. Also, We investigated moderate effects of perceived complexity. Previous studies about hedonic responses didn't show influences of stimulus properties. According to Berlyne's theory(1960, 1974) of aesthetic response, perceived complexity is a important construct because it effects pleasure. Pleasure in response to an object will increase with increased complexity, to an optimal level. After that, with increased complexity, pleasure begins with a linearly increasing line for complexity. Therefore, We expected this perceived complexity will influence hedonic response in game experience. We discussed the rationale for these suggested changes, the assumptions of the resulting framework, and developed some expectations based on its application in Online game context. In the first stage of methodology, questions were developed to measure the constructs. We constructed a survey measuring our theoretical constructs based on a combination of sources, including Yingling(1962), Hargreaves(1962), Lacher (1994), Jongho Lee and Yunhee Jung(2005, 2006), Mehrabian and Russell(1974), Pucely et al(1987). Based on comments received in the pretest, we made several revisions to arrive at our final survey. We investigated the proposed framework through a convenience sample, where participation in a self-report survey was solicited from various respondents having different knowledges. All respondents participated to different degrees, in these habitually practiced activities and received no compensation for their participation. Questionnaires were distributed to graduates and we used 381 completed questionnaires to analysis. The sample consisted of more men(n=225) than women(n=156). In measure, the study used multi-item scales based previous study. We analyze the data using structural equation modeling(LISREL-VIII; Joreskog and Sorbom 1993). First, we used the entire sample(n=381) to refine the measures and test their convergent and discriminant validity. The evidence from both the factor analysis and the analysis of reliability provides support that the scales exhibit internal consistency and construct validity. Second, we test the hypothesized structural model. And, we divided the sample into two different complexity group and analyze the hypothesized structural model of each group. The analysis suggest that hedonic response plays different roles from hypothesized in our study. The results indicate that hedonic response-sensory response, imaginal response, emotional response, analytical response- are related positively to respondents' level of game satisfaction. And game satisfaction is related to higher levels of game loyalty. Additionally, we found that perceived complexity is important to online game experience. Our results suggest that importance of each hedonic response different by perceived game complexity. Understanding the role of perceived complexity in hedonic response enables to have a better understanding of underlying mechanisms at game experience. If game has high complexity, analytical response become important response. So game producers or marketers have to consider more cognitive stimulus. Controversy, if game has low complexity, sensorial response respectively become important. Finally, we discussed several limitations of our study and suggested directions for future research. we concluded with a discussion of managerial implications. Our study provides managers with a basis for game strategies.

• The Journal of Korean Academy of Prosthodontics
• /
• v.52 no.4
• /
• pp.305-311
• /
• 2014

Purpose: When the full veneer crown was treated in the tooth with abfraction lesion due to various causes, the prognosis of it may be compromised according to the location of the finish line, but there is few study about the location of its buccal finish line. The purpose of this study was to investigate the effect of location of the finish line of the full veneer crown on stress distribution of the tooth with abfraction lesion. Materials and methods: The two dimensional finite element model was developed to express tooth, surrounding tissue and full veneer crown. The stress distribution under eccentric 144 N occlusal load was analyzed using finite element analysis. The location of finish line was set just at the lower border of the lesion (Group 0), 1 mm (Group 1) and 2 mm (Group 2) below the lower border of the lesion. Results: In the Group 0, von Mises stress was concentrated at the finish line and the apex of the lesion. Also, the stress at the bucal finish line propagated to the lingual side. In the Group 1 and Group 2, stress distribution was similar each other. Stress was concentrated at the apex of lesion, but the stress at the buccal finish line did not propagate to the lingual side. That implied decrease of the possibility of horizontal crown fracture. Conclusion: Full veneer crown alleviated the stress concentrated at the apex of the abfraction lesion, when the finish line of full veneer crown was set below the lower border of abfraction lesion.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.25 no.2
• /
• pp.63-75
• /
• 2013

The Inner Port Phase 2 area of the Pyeongtaek-Dangjin Port is enclosed by a total of three permeable sea-walls, and the disposal site to the east of the Inner Port Phase 2 is also enclosed by two permeable sea-walls. The maximum tidal range measured in the Inner Port Phase 2 and in the disposal site in May 2010 is 4.70 and 2.32 m, respectively. It reaches up to 54 and 27%, respectively of 8.74 m measured simultaneously in the exterior. Regression formulas between the difference of hydraulic head and the rate of interior water volume change, are induced. A three-dimensional numerical hydrodynamic model for the Asan Bay is constructed incorporating a module to compute water discharge through the permeable sea-walls at each computation time step by employing the formulas. Hydrodynamics for the period from 13th to 27th May, 2010 is simulated by driving forces of real-time reconstructed tide with major five constituents($M_2$, $S_2$, $K_1$, $O_1$ and $N_2$) and freshwater discharges from Asan, Sapkyo, Namyang and Seokmoon Sea dikes. The skill scores of modeled mean high waters, mean sea levels and mean low waters are excellent to be 96 to 100% in the interior of permeable sea-walls. Compared with the results of simulation to obstruct the flow through the permeable sea-walls, the maximum current speed increases by 0.05 to 0.10 m/s along the main channel and by 0.1 to 0.2 m/s locally in the exterior of the Outer Sea-wall of Inner Port. The maximum bottom shear stress is also intensified by 0.1 to 0.4 $N/m^2$ in the main channel and by more than 0.4 $N/m^2$ locally around the arched Outer Sea-wall. The module developed to compute the flow through impermeable seawalls can be practically applied to simulate and predict the advection and dispersion of materials, the erosion or deposion of sediments, and the local scouring around coastal structures where large-scale permeable sea-walls are maintained.

• Journal of Yeungnam Medical Science
• /
• v.6 no.2
• /
• pp.223-239
• /
• 1989

To observe the change in the status of stresses according to three different postural angulation of an intervertebral disc with or without nucleus pulposus, 6 specimens of a 3-dimensional photoelastic model of the s pine were made of epoxy. The nucleus pulposus portion was replaced with silicon in three models, and the three were made without silicon. Through axial application of a vertical compressive load of 8kg, the peculiar patterns of the isochromatic fringes were observed. Stresses on the intervertebral disc were analyzed according to three different postural angulations of the intervertebral disc with the nucleus pulposus and without the nucleus pulposus. The results of these study are as follow : 1. In an erect neutral posture with the nucleus pulposus, the stress concentration was much increased at the posterior portion rather than at the anterior portion. Also, the high stress was concentrated at the medial and central portion. In an erect neutral posture without the nucleus pulposus, the stress concentration was much increased at the anterior portion rather than at the posterior portion and the stress distribution seemed to be locally concentrated. 2. In a maximal flexed posture, the stress concentration was much increased at the posterior portion rather than at the anterior portion. Comparing the presence of the nucleus pulposus with the absence of the nucleus pulposus, the stress concentration was lower at the anterior portion in the presence of the nucleus pulposus than in the absence of the nucleus pulposus. However, the stress distribution at the posterior portion was nearly same in the two groups. According to the analysis of the stress distribution diagram, as a whole, the stress pattern around the disc was evenly distributed. 3. In a maximal extended posture, the higher concentration of the stress distribution at the anterior and medial portion rather than in the posterior and lateral portion was observed. The stress concentration was higher in the presence of the nucleus pulposus than in the absence of the nucleus pulposus. 4. Comparing the maximal flexed posture with the erect neutral posture, the stress concentration in the flexed posture was much decreased in the posterior portion rather than in the erect neutral posture, and an even distribution of the stress pattern in the flexed posture was observed. 5. In the presence of the nucleus pulposus, at the anterior and posterior portion, the stress concentration in the flexed posture was much decreased compared with the extended posture. In the absence of the nucleus pulposus, at the anterior and posterior portion, the stress concentration in the extended posture was much decreased compared with the flexed posture.

• Journal of the Korean Geotechnical Society
• /
• v.21 no.2
• /
• pp.113-120
• /
• 2005

An understanding of soil-structure interaction is the key to rational and economical design for laterally loaded drilled shafts. It is very difficult to formulate the ultimate lateral capacity into a general equation because of the inherent soil nonlincarity, nonhomogeneity, and complexity enhanced by the three dimensional and asymmetric nature of the problem though extensive research works on the behavior of deep foundations subjected to lateral loads have been conducted for several decades. This study reviews the four most well known methods (i.e., Reese, Broms, Hansen, and Davidson) among many design methods according to the specific site conditions, the drilled shaft geometric characteristics (D/B ratios), and the loading conditions. And the hyperbolic lateral capacities (H$_h$) interpreted by the hyperbolic transformation of the load-displacement curves obtained from model tests carried out as a part of this research have been compared with the ultimate lateral capacities (Hu) predicted by the four methods. The H$_u$ / H$_h$ ratios from Reese's and Hansen's methods are 0.966 and 1.015, respectively, which shows both the two methods yield results very close to the test results. Whereas the H$_u$ predicted by Davidson's method is larger than H$_h$ by about $30\%$, the C.0.V. of the predicted lateral capacities by Davidson is the smallest among the four. Broms' method, the simplest among the few methods, gives H$_u$ / H$_h$ : 0.896, which estimates the ultimate lateral capacity smaller than the others because some other resisting sources against lateral loading are neglected in this method. But it results in one of the most reliable methods with the smallest S.D. in predicting the ultimate lateral capacity. Conclusively, none of the four can be superior to the others in a sense of the accuracy of predicting the ultimate lateral capacity. Also, regardless of how sophisticated or complicated the calculating procedures are, the reliability in the lateral capacity predictions seems to be a different issue.