• Journal of Intelligence and Information Systems
• /
• v.24 no.2
• /
• pp.59-83
• /
• 2018

With the increasing importance of sentiment analysis to grasp the needs of customers and the public, various types of deep learning models have been actively applied to English texts. In the sentiment analysis of English texts by deep learning, natural language sentences included in training and test datasets are usually converted into sequences of word vectors before being entered into the deep learning models. In this case, word vectors generally refer to vector representations of words obtained through splitting a sentence by space characters. There are several ways to derive word vectors, one of which is Word2Vec used for producing the 300 dimensional Google word vectors from about 100 billion words of Google News data. They have been widely used in the studies of sentiment analysis of reviews from various fields such as restaurants, movies, laptops, cameras, etc. Unlike English, morpheme plays an essential role in sentiment analysis and sentence structure analysis in Korean, which is a typical agglutinative language with developed postpositions and endings. A morpheme can be defined as the smallest meaningful unit of a language, and a word consists of one or more morphemes. For example, for a word '예쁘고', the morphemes are '예쁘(= adjective)' and '고(=connective ending)'. Reflecting the significance of Korean morphemes, it seems reasonable to adopt the morphemes as a basic unit in Korean sentiment analysis. Therefore, in this study, we use 'morpheme vector' as an input to a deep learning model rather than 'word vector' which is mainly used in English text. The morpheme vector refers to a vector representation for the morpheme and can be derived by applying an existent word vector derivation mechanism to the sentences divided into constituent morphemes. By the way, here come some questions as follows. What is the desirable range of POS(Part-Of-Speech) tags when deriving morpheme vectors for improving the classification accuracy of a deep learning model? Is it proper to apply a typical word vector model which primarily relies on the form of words to Korean with a high homonym ratio? Will the text preprocessing such as correcting spelling or spacing errors affect the classification accuracy, especially when drawing morpheme vectors from Korean product reviews with a lot of grammatical mistakes and variations? We seek to find empirical answers to these fundamental issues, which may be encountered first when applying various deep learning models to Korean texts. As a starting point, we summarized these issues as three central research questions as follows. First, which is better effective, to use morpheme vectors from grammatically correct texts of other domain than the analysis target, or to use morpheme vectors from considerably ungrammatical texts of the same domain, as the initial input of a deep learning model? Second, what is an appropriate morpheme vector derivation method for Korean regarding the range of POS tags, homonym, text preprocessing, minimum frequency? Third, can we get a satisfactory level of classification accuracy when applying deep learning to Korean sentiment analysis? As an approach to these research questions, we generate various types of morpheme vectors reflecting the research questions and then compare the classification accuracy through a non-static CNN(Convolutional Neural Network) model taking in the morpheme vectors. As for training and test datasets, Naver Shopping's 17,260 cosmetics product reviews are used. To derive morpheme vectors, we use data from the same domain as the target one and data from other domain; Naver shopping's about 2 million cosmetics product reviews and 520,000 Naver News data arguably corresponding to Google's News data. The six primary sets of morpheme vectors constructed in this study differ in terms of the following three criteria. First, they come from two types of data source; Naver news of high grammatical correctness and Naver shopping's cosmetics product reviews of low grammatical correctness. Second, they are distinguished in the degree of data preprocessing, namely, only splitting sentences or up to additional spelling and spacing corrections after sentence separation. Third, they vary concerning the form of input fed into a word vector model; whether the morphemes themselves are entered into a word vector model or with their POS tags attached. The morpheme vectors further vary depending on the consideration range of POS tags, the minimum frequency of morphemes included, and the random initialization range. All morpheme vectors are derived through CBOW(Continuous Bag-Of-Words) model with the context window 5 and the vector dimension 300. It seems that utilizing the same domain text even with a lower degree of grammatical correctness, performing spelling and spacing corrections as well as sentence splitting, and incorporating morphemes of any POS tags including incomprehensible category lead to the better classification accuracy. The POS tag attachment, which is devised for the high proportion of homonyms in Korean, and the minimum frequency standard for the morpheme to be included seem not to have any definite influence on the classification accuracy.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.12 no.3
• /
• pp.208-213
• /
• 2008

Purpose: Generally a whole body bone scan has been known as one of the most frequently executed exams in the nuclear medicine fields. Asan medical center, usually use various gamma camera systems - manufactured by PHILIPS (PRECEDENCE, BRIGHTVIEW), SIEMENS (ECAM, ECAM signature, ECAM plus, SYMBIA T2), GE (INFINIA) - to execute whole body scan. But, as we know, each camera's sensitivity is not same so it is hard to consistent diagnosis of patients. So our purpose is when we execute whole body bone scans, we exclude uncontrollable factors and try to correct controllable factors such as inherent sensitivity of gamma camera. In this study, we're going to measure each gamma camera's sensitivity and study about reasonable correction factors of whole body bone scan to follow up patient's condition using different gamma cameras. Materials and Methods: We used the $^{99m}Tc$ flood phantom, it recommend by IAEA recommendation based on general counts rate of a whole body scan and measured counts rates by the use of various gamma cameras - PRECEDENCE, BRIGHTVIEW, ECAM, ECAM signature, ECAM plus, IFINIA - in Asan medical center nuclear medicine department. For measuring sensitivity, all gamma camera equipped LEHR collimator (Low Energy High Resolution multi parallel Collimator) and the $^{99m}Tc$ gamma spectrum was adjusted around 15% window level, the photo peak was set to 140-kev and acquirded for 60 sec and 120 sec in all gamma cameras. In order to verify whether can apply calculated correction factors to whole body bone scan or not, we actually conducted the whole body bone scan to 27 patients and we compared it analyzed that results. Results: After experimenting using $^{99m}Tc$ flood phantom, sensitivity of ECAM plus was highest and other sensitivity order of all gamma camera is ECAM signature, SYMBIA T2, ECAM, BRIGHTVIEW, IFINIA, PRECEDENCE. And yield sensitivity correction factor show each gamma camera's relative sensitivity ratio by yielded based on ECAM's sensitivity. (ECAM plus 1.07, ECAM signature 1.05, SYMBIA T2 1.03, ECAM 1.00, BRIGHTVIEW 0.90, INFINIA 0.83, PRECEDENCE 0.72) When analyzing the correction factor yielded by $^{99m}Tc$ experiment and another correction factor yielded by whole body bone scan, it shows statistically insignificant value (p<0.05) in whole body bone scan diagnosis. Conclusion: In diagnosing the bone metastasis of patients undergoing cancer, whole body bone scan has been conducted as follow up tests due to its good points (high sensitivity, non invasive, easily conducted). But as a follow up study, it's hard to perform whole body bone scan continuously using same gamma camera. If we use same gamma camera to patients, we have to consider effectiveness of equipment's change by time elapsed. So we expect that applying sensitivity correction factor to patients who tested whole body bone scan regularly will add consistence in diagnosis of patients.

• Journal of Biosystems Engineering
• /
• v.1 no.1
• /
• pp.15-15
• /
• 1976

In analyzing the operational characteristics of a rice whitening machine, the internal radial pressure of the machine was measured using strain gage equipment. Changes in cylinder and feed screw configurations, screen type, cylinder speed and counter-pressure levels were examined to determine their impact on the quality and quantity of milled rice and the performance of the machine. The results are summarized as follows: 1. The internal radial pressure in the whitening chamber varied with the surface condition of the grain being processed. During the first or second pass through the machine, pressure was relatively low, reached a maximum after two to three passes with combinations I and II, three to six with combination III and then began to fall. 2. The pitch of the feed screw and the size of the feed gate opening which determine the rate of entry of grain into the whitening chamber, appeared to be the most important factor aff-::cting the degree of radial pressure, quality and quantity of milled rice and the efficiency of the machine. Using a feed screw with a wide pitch (4.8cm), radial pressure was relatively high and head rice recovery ratio \vere quite low. In this case capacity and machine effic?iency were much higher than obtained when using a feed screw with a narrow pitch (2.3cm). Very significant responses in radial pressure, head rice recovery rates and machine capacity were observed with changes in cylinder speed and counter-pressure levels when using the wide pitch feed screw. 3. The characteristics of the screen which surrounds the whitening chamber had an important effect on whitening efficiency. The existence of small protuberances on the original screen resulted in significant increases in both machine capacity and efficiency but without a significant decrease in head rice recovery or development of excessive radial pressure. Further work is required to determine the effects of screen surface conditions and the shape of the cylinderical steel roller on the rate of bran removal, machine efficiency and recovery rates. The size of the slotted perforations 0:1 the screen affects total milled rice recovery. The opening size on the original screen was fabricated to accommodate the round shape of Japonica rice varieties but was not suitable for the more slender Indica type. Milling Indica varieties with this screen resulted in a reduction in total milled rice recovery. 4. An increase in cylinder speed from 380 to 820 rpm produced a positive effect on head rice recovery for all machine combinations at every level of counter-pressure used in the tests. Head rice recovery was considerably lower at 380rpm using a wide screw pitch when compared to the results obtained at speeds from 600 to 820 r.p.m. The effects of cylinder speed On radial pressure, capacity and machine efficiency showed contrasting results, depending on the width of the feed screw pitch. With a narrow feed screw pitch (2.3cm), a direct proportional relationship was observed bet?ween cylinder speed and both radial pressure and machine efficiency. In contrast, using a 4.8 centimeter pitch feed roller produced a series of inverse relationships between the above variables. Based on the results of this study it is recommended when milling Indica type long grain rice varieties that the cylinder speed of the original machine be increased from 500-600 rmp up to a minimum of 800 rpm to obtain a greater abrasive effect between the grain and the screen. The pitch of the feed screw should be also reduced to decr?ease the level of internal radial pressure and to obtain higher machine efficiency and increased quality of milled rice with increased cylinder speeds. Further study on the interaction between cylinder speed and feed screw pitch is recommended. 5. An increase in the counter pressure level produced a negative effect On the head rice recovery with an increase in radial pressure, capacity, and machine efficiency over all combinations and at every level of cylinder speed. 6. Head rice recovery rates were conditioned primarily by the pressure inside the whitening chamber. According to the empirical cha racteristics curve developed in this study, the relationships of head rice recovery ($Y_h$) and machine capacity ($Y_c$/TEX>) to internal radial pressure ($X_p$) followed an inverse quadratic function and a linear function respectively: $$Y_h^\Delta=\frac{1}{{1.4383-0.2951X_p^\ast+0.1425X_p^{\ast\ast}}^2} , (R^2=0.98)$$$$Y_c^\Delta=-305.83+374.37X_p^{\ast\ast}, (R^2=0.88)$$The correlation between capacity and power consumption per unit of brown rice expressed in the following exponential function: $$Y_c^\Delta=1.63Y_c^{-0.7786^\{\ast\ast}, (R^2=0.94)$$These relationships indicate that when radial pressure increases above a certain range (1. 6 to 2.0 kg/$cm^2$ based On the results of the experiment) head ricerecovery decrea?ses in a quadratic relation with a inear increase in capacity but without any decrease in power consump tion per unit of brown rice. On the other hand, if radial pressure is below the range shown above, power consumption increases dramatically with a lin?ear decrease in capacity but without significant increases in head rice recovery. During the operation of a given whitening machine, the optimum radial pressure range or the correct capacity range should be selected by controlling the feed rate and/or counter-pressure keeping in mind the condition of the grain, particulary the hardness. It was observed that the total number of passes is related to radial pessure level, feed rate and counter-pressure level. The higher theradial pressure the fewer num?ber of pass required but with decreased head rice recovery. In particular, when using high feed rates, the total number of passes should be increased to more than three by reducing the counter-pressure level to avoid decreaseases in head rice recovery (less than 65 percent head rice recovery on the basis of brown rice) at every cylinder speed. 7. A rapid rise in grain temperature seemed to have a close relationship with the pressure generated inside the whitening chamber and, subsequently with head rice reco?very rates. The higher the rate of increase, the lower were the resulting head rice recoveries.

• Journal of Biosystems Engineering
• /
• v.1 no.1
• /
• pp.17-48
• /
• 1976

In analyzing the operational characteristics of a rice whitening machine, the internal radial pressure of the machine was measured using strain gage equipment. Changes in cylinder and feed screw configurations, screen type, cylinder speed and counter-pressure levels were examined to determine their impact on the quality and quantity of milled rice and the performance of the machine. The results are summarized as follows: 1. The internal radial pressure in the whitening chamber varied with the surface condition of the grain being processed. During the first or second pass through the machine, pressure was relatively low, reached a maximum after two to three passes with combinations I and II, three to six with combination III and then began to fall. 2. The pitch of the feed screw and the size of the feed gate opening which determine the rate of entry of grain into the whitening chamber, appeared to be the most important factor aff-::cting the degree of radial pressure, quality and quantity of milled rice and the efficiency of the machine. Using a feed screw with a wide pitch (4.8cm), radial pressure was relatively high and head rice recovery ratio \vere quite low. In this case capacity and machine effic\ulcorneriency were much higher than obtained when using a feed screw with a narrow pitch (2.3cm). Very significant responses in radial pressure, head rice recovery rates and machine capacity were observed with changes in cylinder speed and counter-pressure levels when using the wide pitch feed screw. 3. The characteristics of the screen which surrounds the whitening chamber had an important effect on whitening efficiency. The existence of small protuberances on the original screen resulted in significant increases in both machine capacity and efficiency but without a significant decrease in head rice recovery or development of excessive radial pressure. Further work is required to determine the effects of screen surface conditions and the shape of the cylinderical steel roller on the rate of bran removal, machine efficiency and recovery rates. The size of the slotted perforations 0:1 the screen affects total milled rice recovery. The opening size on the original screen was fabricated to accommodate the round shape of Japonica rice varieties but was not suitable for the more slender Indica type. Milling Indica varieties with this screen resulted in a reduction in total milled rice recovery. 4. An increase in cylinder speed from 380 to 820 rpm produced a positive effect on head rice recovery for all machine combinations at every level of counter-pressure used in the tests. Head rice recovery was considerably lower at 380rpm using a wide screw pitch when compared to the results obtained at speeds from 600 to 820 r.p.m. The effects of cylinder speed On radial pressure, capacity and machine efficiency showed contrasting results, depending on the width of the feed screw pitch. With a narrow feed screw pitch (2.3cm), a direct proportional relationship was observed bet\ulcornerween cylinder speed and both radial pressure and machine efficiency. In contrast, using a 4.8 centimeter pitch feed roller produced a series of inverse relationships between the above variables. Based on the results of this study it is recommended when milling Indica type long grain rice varieties that the cylinder speed of the original machine be increased from 500-600 rmp up to a minimum of 800 rpm to obtain a greater abrasive effect between the grain and the screen. The pitch of the feed screw should be also reduced to decr\ulcornerease the level of internal radial pressure and to obtain higher machine efficiency and increased quality of milled rice with increased cylinder speeds. Further study on the interaction between cylinder speed and feed screw pitch is recommended. 5. An increase in the counter pressure level produced a negative effect On the head rice recovery with an increase in radial pressure, capacity, and machine efficiency over all combinations and at every level of cylinder speed. 6. Head rice recovery rates were conditioned primarily by the pressure inside the whitening chamber. According to the empirical cha racteristics curve developed in this study, the relationships of head rice recovery ($Y_h$) and machine capacity ($Y_c$/TEX>) to internal radial pressure ($X_p$) followed an inverse quadratic function and a linear function respectively: $$Y_h^\Delta=\frac{1}{{1.4383-0.2951X_p^\ast+0.1425X_p^{\ast\ast}}^2} , (R^2=0.98)$$ $$Y_c^\Delta=-305.83+374.37X_p^{\ast\ast}, (R^2=0.88)$$ The correlation between capacity and power consumption per unit of brown rice expressed in the following exponential function: $$Y_c^\Delta=1.63Y_c^{-0.7786^\{\ast\ast}, (R^2=0.94)$$ These relationships indicate that when radial pressure increases above a certain range (1. 6 to 2.0 kg/$cm^2$ based On the results of the experiment) head ricerecovery decrea\ulcornerses in a quadratic relation with a inear increase in capacity but without any decrease in power consump tion per unit of brown rice. On the other hand, if radial pressure is below the range shown above, power consumption increases dramatically with a lin\ulcornerear decrease in capacity but without significant increases in head rice recovery. During the operation of a given whitening machine, the optimum radial pressure range or the correct capacity range should be selected by controlling the feed rate and/or counter-pressure keeping in mind the condition of the grain, particulary the hardness. It was observed that the total number of passes is related to radial pessure level, feed rate and counter-pressure level. The higher theradial pressure the fewer num\ulcornerber of pass required but with decreased head rice recovery. In particular, when using high feed rates, the total number of passes should be increased to more than three by reducing the counter-pressure level to avoid decreaseases in head rice recovery (less than 65 percent head rice recovery on the basis of brown rice) at every cylinder speed. 7. A rapid rise in grain temperature seemed to have a close relationship with the pressure generated inside the whitening chamber and, subsequently with head rice reco\ulcornervery rates. The higher the rate of increase, the lower were the resulting head rice recoveries.

• Journal of Nutrition and Health
• /
• v.6 no.1
• /
• pp.71-76
• /
• 1973

1. This survey is somewhat different from that conducted by Yonsei University, although being in many respects, very similar. We found the average per capita caloric intake to be 7 or 8% of what it should be. Of that caloric intake, 84.5% in the model village here and 82.2% in the compared village Bupyong, by and large depends on cereal grains. Since such grains tend to distend the stomach, the farmers, it seems,should substitute fat for a part of their diet so as to dimish digestive pain and still receive the necessary calories. 2. Protein is the most important nutrient for the development of physical strength and improvement of health, but the average daily intake is only 68.3% of the necessary amount. It is desirable that the ratio of vegetable protein to that of animal be one to three for maintaining one's health. Most of the villager's protein, however, comes from plants: the intake of animal protein, at a level of only 13.6 g, is far below such a one to three ratio. 3. In the model village, 497.6 mg, of inorganic calcium is the daily intake level. In the compared village it is 505.5 mg, making a difference of only 8mg. This, however, is 35% less than the recommended intake. More than 50% of this calcium comes from cereals and other plants. Moreover, plant calcium which has much oxakuc acud us not as nutritional as animal calcium, so their calcium diet is less than it would appear. We must, therefore, make efforts to receive as good nutritional calcium as possible. 4. Among the vitamin group, the daily average intake of vitamin A and vitamin $B_{2}$ are respectively 40% and 32% less than the desired intake, while vitamin $B_{1}$ happens to be taken in sufficient quantities and more niacin is taken than which is even necessary. The intake of vitamin C is much more than the necessary quantity. However, this figure was calculated from uncooked food; if the loss from cooking were to be considered, the real intake might well be a little less. Also, as this survey was carried out in May, some of these results were influenced by the fact that lettuce and spinach are seasonally popular. In conclusion, except for a few nutrients which are in abundance, the normal food intake in a day is, on the whole, less than the average recommended. Furthermore despite the fact that both of these places are model villages in the development of nutrition, it seems that they have not gotten out of such conventional eating habiys as the almost dependancy on cereal grains. Cow's milk, sheep's milk, eggs and so on produced by each farmhouse are not used for their own families but are taken to the market for the purpose of making money. Accordingly, I think from now we must seek to improve, guide and enlighten the farmers as to how to correct their eating habits and implement changes in their lives so that our firm purpose may be achieved.

• Journal of Oral Medicine and Pain
• /
• v.32 no.2
• /
• pp.157-175
• /
• 2007

This study was conducted through an interview process in which questionnaires were administered to 293 people. The questionnaires related to the behaviors of oral hygiene care, and disease history related to halitosis, and status of halitosis, halitosis measurement, oral examination, and caries activity tests such as the snyder test, Salivary flow rate test, and Salivary buffering capacity test. Our sample was taken from 293 rural residents within the period from 4th to 21st of January 2006. This was done in order to provide basic data to prepare both policies of halitosis prevention and a device to efficiently measure halitosis status and investigate the factors related therein. The major findings of this study results are as follows: 1. As for frequency of tooth brushing, twice a day occupied the greatest portion at 46.1 % Women exceeded men in frequency of tooth brushing. Tongue brushing everyday produced a 25.6 % result among subjects and The use of auxiliary oral hygiene devices occupied 9.2 %. 2. As for degree of usual self-awareness of halitosis: 62.5 %. This result also demonstrate that the severest time of self-awareness in regards to halitosis is wake up time in the morning. The time period produced the highest portion of 72.7 % in times of self-awareness. In terms of the area in which halitosis was observed, gum resulted in 23.0 %. As for types of halitosis, fetid smell was the most frequent at 37.2 %. 3. As for the result of halitosis measurement, values of OG less than 50 ppm occupied 54.3 % and $50{\sim}100ppm$ occupied 41.6 %. As for $NH_3$ values, $20{\sim}60ppm$ showed the highest value range of 52.6 %. 4. As for OG per disease history related to halitosis, values of OG were significantly high in the ranges of $50{\sim}100ppm$ within family history groups of food impaction by dental caries, diabetes mellitus and halitosis. As for values of $NH_3$, there showed a significant difference in respiratory system disease groups. 5 Value range of OG per ordinary halitosis self-awareness degree: values ranging less than 50 ppm were recorded at 55.9 % from the group realizing not aware of smell. 57.5 % from groups only realizing sometimes, while values range of $50{\sim}100ppm$ were recorded at 52.0 % from groups always aware of smell. 63.6 % from groups always strongly aware of smell. Meanwhile as for the values ranges of $NH_3$, $20{\sim}60ppm$. they occupied high portions for all groups of exams. 6. Values of OG per oral examination: the more pulp-exposed teeth and food impaction and the higher the tongue plaque index, values of OG increased within the range of $50{\sim}100ppm$. As for values of $NH_3$, the more prosthetic teeth and the higher the tongue plaque index, this value increased significantly, and the values increased up to no less than 60 ppm for groups of mandibular partial denture. 7. Within the realm of caries activity test: as for the Snyder test, high activity was highest by 43.0 % wherewith the higher the activity of acidogenic bacteria the higher the OG values. As for the salivary flow rate test, the number of cases below 8.0 ml showed the highest tendency by 62.5 %. The larger the salivary flow rate the more decreased OG values distribution. As for the salivary buffering capacity test, $6{\sim}10$ drops of 0.1N lactic acid showed the overwhelming trend by 58.7 % whereby the higher the salivary buffering capacity the greater distribution occupancy ratio of OG values below 50 ppm which is scentless to on ordinary person. 8. As for the correlation between oral environment and halitosis, OG showed the positive correlation with pulp exposed teeth, filled teeth, present teeth, tongue plaque index, and food impaction, while the negative correlation with salivary flow rate and prosthetic teeth. $NH_3$ showed a positive correlation with prosthetic teeth and frequency of tooth brushing, while decayed teeth was negative correlation. 9. As for the multiple regression analysis result, there have been selected female, pulp exposed teeth, prosthetic teeth, food impaction, salivary flow rate, tongue plaque index and severe activities in the Snyder test as factors affecting OG wherein explanatory power on it was 45.1 %. There have been selected females, pulp exposed teeth, tongue plaque index, and prosthetic teeth as factors affecting on $NH_3$ wherein explanatory power on it was 6.6 %. With the aforementioned results in mind, the status of halitosis among rural residents is considered to bare a close relation with oral environments and other factors related to halitosis such as the Snyder test from caries activity test, and salivary flow rate test. For the prevention of halitosis of residents in rural areas, we have to focus on correct tooth brushing methods and tongue brushing, with using auxiliary oral hygiene devices to remove fur of tongue plaque and food impaction. Also, when the cause and ingredients of halitosis are diverse and complex, in order to analyze exactly the factors of individual halitosis development, we need continuous and systematic study in order to provide rural residents with programs of oral hygiene education and encourage the use of dental hygienists in public health centers.