• Asia pacific journal of information systems
• /
• v.19 no.2
• /
• pp.157-178
• /
• 2009

Corporate credit rating is a very important factor in the market for corporate debt. Information concerning corporate operations is often disseminated to market participants through the changes in credit ratings that are published by professional rating agencies, such as Standard and Poor's (S&P) and Moody's Investor Service. Since these agencies generally require a large fee for the service, and the periodically provided ratings sometimes do not reflect the default risk of the company at the time, it may be advantageous for bond-market participants to be able to classify credit ratings before the agencies actually publish them. As a result, it is very important for companies (especially, financial companies) to develop a proper model of credit rating. From a technical perspective, the credit rating constitutes a typical, multiclass, classification problem because rating agencies generally have ten or more categories of ratings. For example, S&P's ratings range from AAA for the highest-quality bonds to D for the lowest-quality bonds. The professional rating agencies emphasize the importance of analysts' subjective judgments in the determination of credit ratings. However, in practice, a mathematical model that uses the financial variables of companies plays an important role in determining credit ratings, since it is convenient to apply and cost efficient. These financial variables include the ratios that represent a company's leverage status, liquidity status, and profitability status. Several statistical and artificial intelligence (AI) techniques have been applied as tools for predicting credit ratings. Among them, artificial neural networks are most prevalent in the area of finance because of their broad applicability to many business problems and their preeminent ability to adapt. However, artificial neural networks also have many defects, including the difficulty in determining the values of the control parameters and the number of processing elements in the layer as well as the risk of over-fitting. Of late, because of their robustness and high accuracy, support vector machines (SVMs) have become popular as a solution for problems with generating accurate prediction. An SVM's solution may be globally optimal because SVMs seek to minimize structural risk. On the other hand, artificial neural network models may tend to find locally optimal solutions because they seek to minimize empirical risk. In addition, no parameters need to be tuned in SVMs, barring the upper bound for non-separable cases in linear SVMs. Since SVMs were originally devised for binary classification, however they are not intrinsically geared for multiclass classifications as in credit ratings. Thus, researchers have tried to extend the original SVM to multiclass classification. Hitherto, a variety of techniques to extend standard SVMs to multiclass SVMs (MSVMs) has been proposed in the literature Only a few types of MSVM are, however, tested using prior studies that apply MSVMs to credit ratings studies. In this study, we examined six different techniques of MSVMs: (1) One-Against-One, (2) One-Against-AIL (3) DAGSVM, (4) ECOC, (5) Method of Weston and Watkins, and (6) Method of Crammer and Singer. In addition, we examined the prediction accuracy of some modified version of conventional MSVM techniques. To find the most appropriate technique of MSVMs for corporate bond rating, we applied all the techniques of MSVMs to a real-world case of credit rating in Korea. The best application is in corporate bond rating, which is the most frequently studied area of credit rating for specific debt issues or other financial obligations. For our study the research data were collected from National Information and Credit Evaluation, Inc., a major bond-rating company in Korea. The data set is comprised of the bond-ratings for the year 2002 and various financial variables for 1,295 companies from the manufacturing industry in Korea. We compared the results of these techniques with one another, and with those of traditional methods for credit ratings, such as multiple discriminant analysis (MDA), multinomial logistic regression (MLOGIT), and artificial neural networks (ANNs). As a result, we found that DAGSVM with an ordered list was the best approach for the prediction of bond rating. In addition, we found that the modified version of ECOC approach can yield higher prediction accuracy for the cases showing clear patterns.

• Journal of Korean Society of Forest Science
• /
• v.26 no.1
• /
• pp.57-65
• /
• 1975

This study was carried out to examine the effect of manufacturing factors on physical properties of hardboard (S.I.S) made from the juvenile wood of sycamore (Platanus orientalis L.) The results obtained may be summarized as follows: 1. The difference among the yields of pulp treated with preheat time (defibrate condition) was significant in those of treatments. There was no difference in the yield of pulp treated with defibrate time. The yields of pulp on the tree age classes were shown the difference by 2<4<6<8 years. 2. The specific gravities of hardboard that were treated with hot pressing conditions showed us significantly in those of treatments. There was no difference on the specific gravities among hardboards, treated with resin and wax contents. But in all cases of the specific gravities satisfied the standard which specified the KS F 3203 (Hardboard) 3. The moisture contents of hardboard satisfied the standard which calls for 13-percent below. There were difference in moisture contents between hardboard, treated with preheating time, resin and wax contents and hot pressing conditions. And the moisture contents of hardboard on the tree age classes showed the difference by 2<4<6<8 years. 4. The water absorption and thickness swelling of hardboard treated with defibrations, resin and wax contents, and hot pressing conditions were significant in those of treatments. And the water absorption and thickness swelling of hardboard on the tree age classes showed us the significant difference by 8<6<4<2 years. 5. The difference among the flexural strength in using tested three conditions showed us the difference by defibration$200kg/cm^2$) of hardboard, it is likely to be recommened that the juvenile wood of sycamore is valuable for the raw materials of hardboards.

• Microbiology and Biotechnology Letters
• /
• v.41 no.3
• /
• pp.327-334
• /
• 2013

For the development of hardy kiwi wine, we arranged for the post-maturity of hardy kiwi fruit, treated them with calcium carbonate and a pectinase enzyme complex, investigated the resulting physicochemical properties and conducted a sensory evaluation. The period determined for creating post-maturity in the hardy kiwi fruit was determined as 5 days storage at room temperature following maturity. During this time the yield of fruit juice was increased from 22.1% to 53.5% using 0.1% (v/v) cytolase PCL5 for 2 h at room temperature. 0.1% (w/v) calcium carbonate was also added during the process of aging, for the reduction of the sour taste. The fermentation trial of the hardy kiwi wine was prepared using water (25% or 50%), sugar ($24^{\circ}brix$), 0.1% (w/v) $CaCO_3$, 0.1% (v/v) cytolase PCL5, $K_2S_2O_5$ (200 ppm), and yeast ($1.5{\times}10^7$ cell/ml). Fermentation then occurred for 2 weeks at $20^{\circ}C$. The pH value, total acidity, alcohol, and reducing sugar content of the resulting hardy kiwi wines of 25% (v/w) and 50% (v/w) water, were in a range of pH 3.4-3.7, 1.12-1.21%, 14.3-14.4%, and 15-16 g/l, respectively. Citric acid and fructose constituted the major organic acids and the free sugar of the 25% and 50% hardy kiwi wine, respectively. Volatile flavor components, including 10 kinds of esters, 8 kinds of alcohols, 5 kinds of acids, 3 kinds of others and aldehydes, were determined by GC analysis. The results of sensory evaluation demonstrated that 50% hardy kiwi wine is more palatable than 25% hardy kiwi wine.

• Food Science and Preservation
• /
• v.22 no.5
• /
• pp.674-682
• /
• 2015

In this study, vinegar was produced using urushiol-free fermented Rhus verniciflua extract to create a lacquer with added value. The effect of manufacturing conditions on the quality of vinegar using detoxified R. verniciflua extract for fermentation was investigated. The acidity of the vinegar for inoculations with various liquid starter contents was 4.8~4.9%, and it was similar among all treatment groups. The acidity of vinegar was higher when the initial alcohol content was high. The acetic acid yields were 82.8%, 84.4%, 77.7%, and 69.5%, and the maximum yield was observed when the initial alcohol content was 6%. For acetic acid fermentation using different amounts of detoxified R. verniciflua extracts, the acidity of the vinegar with the extract after fermentation was 5.3~5.9%. However, the acidity of vinegar without the extract was 5.5%. The intensity of the brown color was high for vinegar without the extract. Hunter's L values were high for vinegar with an extract content of 2%. Acetic acid (53.3~65.8 mg/mL) was the predominant acid. Arginine ($190.3{\sim}333.3{\mu}g/mL$), proline ($125.6{\sim}290.8{\mu}g/mL$), alanine ($126.1{\sim}270.9{\mu}g/mL$), and glutamic acid ($159.0{\sim}262.4{\mu}g/mL$) were the predominant amino acids in detoxified R. verniciflua vinegar.

• Resources Recycling
• /
• v.31 no.2
• /
• pp.20-32
• /
• 2022

This study implemented a chelating agent (Ethylenediaminetetraacetic acid, EDTA) in purification to obtain high-purity vanadium pentoxide (V₂O₅) for use in VRFB (Vanadium Redox Flow Battery). V₂O₅ (powder) was produced through the precipitation recovery of ammonium metavanadate (NH₄VO₃) from a vanadium solution, which was prepared using a low-purity vanadium raw material. The initial purity of the powder was estimated to be 99.7%. However, the use of a chelating agent improved its purity up to 99.9% or higher. It was conjectured that the added chelating agent reacted with the impurity ions to form a complex, stabilizing them. This improved the selectivity for vanadium in the recovery process. However, the prepared V₂O₅ powder exhibited higher contents of K, Mn, Fe, Na, and Al than those in the standard counterparts, thus necessitating additional research on its impurity separation. Furthermore, the vanadium electrolyte was prepared using the high-purity V₂O₅ powder in a newly developed direct electrolytic process. Its analytical properties were compared with those of commercial electrolytes. Owing to the high concentration of the K, Ca, Na, Al, Mg, and Si impurities in the produced vanadium electrolyte, the purity was analyzed to be 99.97%, lower than those (99.98%) of its commercial counterparts. Thus, further research on optimizing the high-purity V₂O₅ powder and electrolyte manufacturing processes may yield a process capable of commercialization.

• Applied Biological Chemistry
• /
• v.10
• /
• pp.69-100
• /
• 1968

1. In order to investigate on the microflora and enzyme activity of mold wheat 'Nuruk' , the major source of microorganisms for the brewing of Takju (a Korean Sake), two samples of Nuruk, one prepared at the College of Agriculture, Chung Nam University (S) and the other perchased at a market (T), were taken for the study. The molds, aerobic bacteria, lactic acid bacteria, and yeasts were examined and counted. The yeasts were classified by the treatment with TTC (2, 3, 5 triphenyltetrazolium chloride) agar that yields a varied shade of color. The amylase and protease activities of Nuruk were measured. The results were as the followings. a) In the Nuruk S found were: Aspergillus oryzae group, $204{\times}10^5$; Black Aspergilli, $163{\times}10^5$; Rhizogus, $20{\times}10^5$; Penicillia, $134{\times}10^5$; Areobic bacteria, $9{\times}10^6-2{\times}10^7$; Lactic acid bacteria, $3{\times}10^4$ In the Nuruk T found were: Aspergillus oryzae group, $836{\times}10^5$; Black Aspergilli, $286{\times}10^5$; Rhizopus, $623{\times}10^5$; Penicillia, $264{\times}10^5$; Aerobic bacteria, $5{\times}10^6-9{\times}10^6$; Lactic acid bacteria, $3{\times}10^4$ b) Eighty to ninety percent of the aerobic bacteria in Nuruk S appeared to belong to Bacillus subtilis while about 70% of those in Nuruk T seemed to be spherical bacteria. In both Nuruks about 80% of lactic acid bacteria were observed as spherical ones. c) The population of yeasts in 1g. of Nuruk S was about $6{\times}10^5$, 56.5% of which were TTC pink yeasts, 16% of which were TTC red pink yeasts, 8% of which were TTC red yeasts, 19.5% of which were TTC white yeasts. In Nuruk T(1g) the number of yeasts accounted for $14{\times}10^4$ and constituted of 42% TTC pink. 21% TTC red pink 28% TTC red and 9% TTC white. d) The enzyme activity of 1g Nuruk S was: Liquefying type Amylase, $D^{40}/_{30},=256$ W.V. Saccharifying type Amylase, 43.32 A.U. Acid protease, 181 C.F.U. Alkaline protease, 240C.F.U. The enzyme activity of 1g Nuruk T was: Liquefying type Amylase $D^{40}/_{30},=32$ W.V. Saccharifying type amylase $^{30}34.92$ A.U. Acid protease, 138 C.F.U. Alkaline protease 31 C.F.U. 2. During the fermentation of 'Takju' employing the Nuruks S and T the microflora and enzyme activity throughout the brewing were observed in 12 hour intervals. TTC pink and red yeasts considered to be the major yeasts were isolated and cultured. The strains ($1{\times}10^6/ml$) were added to the mashes S and T in which pH was adjusted to 4.2 and the change of microflora was examined during the fermentation. The results were: a) The molds disappeared from each sample plot since 2 to 3 days after mashing while the population of aerobic bacteria was found to be $10{\times}10^7-35{\times}10^7/ml$ inS plots and $8.2{\times}10^7-12{\times}10^7$ in plots. Among them the coccus propagated substantially until some 30 hours elasped in the S and T plots treated with lactic acid but decreased abruptly thereafter. In the plots of SP. SR. TP. and TR the coccus had not appeared from the beginning while the bacillus showed up and down changes in number and diminished by 1/5-1/10 the original at the end stage. b) The lactic acid bacteria observed in the S plot were about $7.4{\times}10^7$ in number per ml of the mash in 24 hours and increased up to around $2{\times}10^8$ until 3-4 days since. After this period the population decreased rapidly and reached about $4{\times}10^5$ at the end, In the plot T the lactic acid becteria found were about $3{\times}10^8$ at the period of 24 fours, about $3{\times}10$ in 3 days and about $2{\times}10^5$ at the end in number. In the plots SP. SR. TP, and TR the lactic acid bacteria observed were as less as $4{\times}10^5$ at the stage of 24 hours and after this period the organisms either remained unchanged in population or ceased to exist. c) The maiority of lactic acid bacteria found in each mash were spherical and the change in number displayed a tendency in accordance with the amount of lactic acid and alcohol produced in the mash. d) The yeasts had showed a marked propagation since the period of 24 hours when the number was about $2{\times}10^8$ ㎖ mash in the plot S. $4{\times}10^8$ in 48 hours and $5-7{\times}10^8$ in the end period were observed. In the plot T the number was $4{\times}10^8$ in 24 hours and thereafter changed up and down maintaining $2-5{\times}10^8$ in the range. e) Over 90% of the yeasts found in the mashes of S and T plots were TTC pink type while both TTC red pink and TTC red types held range of $2{\times}10-3{\times}10^7$ throughout the entire fermentation. f) The population of TTC pink yeasts in the plot SP was as $5{\times}10^8$ much as that is, twice of that of S plot at the period of 24 hours. The predominance in number continued until the middle and later stages but the order of number became about the same at the end. g) Total number of the yeasts observed in the plot SR showed little difference from that of the plot SP. The TTC red yeasts added appeared considerably in the early stage but days after the change in number was about the same as that of the plot S. In the plot TR the population of TTC red yeasts was predominant over the T plot in the early stage which there was no difference between two plots there after. For this reason even in the plot w hers TTC red yeasts were added TTC pink yeasts were predominant. TTC red yeasts observed in the present experiment showed continuing growth until the later stage but the rate was low. h) In the plot TP TTC pink yeasts were found to be about $5{\times}10^8$ in number at the period of 2 days and inclined to decrease thereafter. Compared with the plot T the number of TTC pink yeasts in the plot TP was predominant until the middle stage but became at the later stage. i) The productivity of alcohol in the mash was measured. The plot where TTC pink yeasts were added showed somewhat better yield in the earely stage but at and after the middle stage the difference between the yeast-added and the intact mashes was not recognizable. And the production of alcohol was not proportional to the total number of yeasts present. j) Activity of the liquefying amylase was the highest until 12 hours after mashing, somewhat lowered once after that, and again increased around 36-48 hours after mashing. Then the activity had decreased continuously. Activity of saccharifying amylase also decreased at the period of 24 hours and then increased until 48 hours when it reached the maximum. Since, the activity had gradually decreased until 72 hours and rapidly so did thereafter. k) Activity of alkaline protease during the fermentation of mash showed a tendency to decrease continusously although somewhat irregular. Activity of acid protease increased until hours at the maximum, then decreased rapidly, and again increased, the vigor of acid protease showed better shape than that of alkaline protease throughout. 3. TTC pink yeasts that were predominant in number, two strains of TTC red pink yeasts that appeared throughout the brewing, and TTC red yeasts were identified and the physiological characters examined. The results were as described below. a) TTC pinkyeasts (B-50P) and two strains of TTC red pink yeasts (B-54 RP & B-60 RP) w ere identified as the type of Saccharomyces cerevisiae and TTC pink red yeasts CB-53 R) were as the type of Hansenula subpelliculosa. b) The fermentability of four strains above mentioned were measured as follows. Two strains of TTC red pink yeasts were the highest, TTC pink yeasts were the lowest in the fermantability. The former three strains were active in the early stage of fermentation and found to be suitable for manufacturing 'Takju' TTC red yeasts were found to play an important role in Takju brewing due to its strong ability to produce esters although its fermentability was low. c) The tolerance against nitrous acid of strains of yeast was marked. That against lactic acid was only 3% in Koji extract, and TTC red yeasts showed somewhat stronger resistance. The tolerance against alcohol of TTC pink and red pink yeasts in the Hayduck solution was 7% while that in the malt extract was 13%. However, that of TTC red yeasts was much weaker than others. Liguefying activity of gelatin by those four strains of yeast was not recognized even in 40 days. 4. Fermentability during Takju brewing was shown in the first two days as much as 70-80% of total fermentation and around 90% of fermentation proceeded in 3-4 days. The main fermentation appeared to be completed during :his period. Productivity of alcohol during Takju brewing was found to be apporximately 65% of the total amount of starch put in mashing. 5. The reason that Saccharomyces coreanuss found be Saito in the mash of Takju was not detected in the present experiment is considered due to the facts that Aspergillus oryzae has been inoculated in the mold wheat (Nuruk) since around 1930 and also that Koji has been used in Takju brewing, consequently causing they complete change in microflora in the Takju brewing. This consideration will be supported by the fact that the original flavor and taste have now been remarkably changed.