• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.1
• /
• pp.17-24
• /
• 2020

In this paper, we present a movie rating prediction architecture based on a convolutional neural network (CNN). Our prediction architecture extends TextCNN, a popular CNN-based architecture for sentence classification, in three aspects. First, character embeddings are utilized to cover many variants of words since reviews are short and not well-written linguistically. Second, the attention mechanism (i.e., squeeze-and-excitation) is adopted to focus on important features. Third, a scoring function is proposed to convert the output of an activation function to a review score in a certain range (1-10). We evaluated our prediction architecture on a movie review dataset and achieved a low MSE (e.g., 3.3841) compared with an existing method. It showed the superiority of our movie rating prediction architecture.

• The Journal of Korean Association of Computer Education
• /
• v.14 no.4
• /
• pp.63-70
• /
• 2011

Recommender systems, which predict and recommend items that may possibly draw users' interests, have been applied in various fields as e-commerce systems are widespread. Collaborative filtering, one of the major methodologies of recommender systems, recommends either items similar to those preferred by the user, or items preferred by the other similar user. Therefore, two problems determine its performance; one is correct estimation of similarity and the other is predicting the real rating of the recommended item. This study addresses the latter problem. Previous studies predict the real rating based on the mean of the ratings, but this study proposes a prediction based on the range of the ratings and investigates its performance through experiments. As a result, it is demonstrated that the proposed method improves the mean absolute error significantly, compared to the previous method.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.6
• /
• pp.81-89
• /
• 2020

Online reviews we encounter commonly on SNS, although a complex range of assessment information affecting the consumer's preferences are included, it is general that such information is just provided by simple numbers or star ratings. Based on those review types, it is not easy to get specific information that consumers want and use it to make a decision for purchase. Therefore, in this study, we propose a prediction methodology that can provide ratings broken down by evaluation items by performing sentiment analysis on restaurant reviews written in Korean. To this end, we select 'food', 'price', 'service', and 'atmosphere' as the main evaluation items of restaurants, and build a new sentiment dictionary for each evaluation item. It also classifies review sentences by rating item, predicts granular ratings through sentiment analysis, and provides additional information that consumers can use to make decisions. Finally, using MAE and RMSE as evaluation indicators it shows that the rating prediction accuracy of the proposed methodology has been improved than previous studies and presents the use case of proposed methodology.

• 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.

• Smart Structures and Systems
• /
• v.26 no.6
• /
• pp.781-796
• /
• 2020

Prediction of total sediment load is essential in an extensive range of problems such as the design of the dead volume of dams, design of stable channels, sediment transport in the rivers, calculation of bridge piers degradation, prediction of sand and gravel mining effects on river-bed equilibrium, determination of the environmental impacts and dredging necessities. This paper is aimed to investigate and predict the total sediment load of the Wadi Arbaat in Eastern Sudan. The study was estimated the sediment load by separate total sediment load into bedload and Suspended Load (SL), independently. Although the sediment records are not sufficient to construct the discharge-sediment yield relationship and Sediment Rating Curve (SRC), the total sediment loads were predicted based on the discharge and Suspended Sediment Concentration (SSC). The turbidity data NTU in water quality has been used for prediction of the SSC in the estimation of suspended Sediment Yield (SY) transport of Wadi Arbaat. The sediment curves can be used for the estimation of the suspended SYs from the watershed area. The amount of information available for Khor Arbaat case study on sediment is poor data. However, the total sediment load is essential for the optimal control of the sediment transport on Khor Arbaat sediment and the protection of the dams on the upper gate area. The results show that the proposed model is found to be considered adequate to predict the total sediment load.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.1
• /
• pp.125-134
• /
• 2021

With the rapid development of the film industry, the number of films is significantly increasing and movie recommendation system can help user to predict the preferences of users based on their past behavior or feedback. This paper proposes a movie recommendation system based on the latent factor model with the adjustment of mean and bias in rating. Singular value decomposition is used to decompose the rating matrix and stochastic gradient descent is used to optimize the parameters for least-square loss function. And root mean square error is used to evaluate the performance of the proposed system. We implement the proposed system with Surprise package. The simulation results shows that root mean square error is 0.671 and the proposed system has good performance compared to other papers.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.30 no.6B
• /
• pp.579-587
• /
• 2010

The mid-range streamflow forecast was performed using NWP(Numerical Weather Prediction) provided by KMA. The NWP consists of RDAPS for 48-hour forecast and GDAPS for 240-hour forecast. To enhance the accuracy of the NWP, QPM to downscale the original NWP and Quantile Mapping to adjust the systematic biases were applied to the original NWP output. The applicability of the suggested streamflow prediction system which was verified in Geum River basin. In the system, the streamflow simulation was computed through the long-term continuous SSARR model with the rainfall prediction input transform to the format required by SSARR. The RQPM of the 2-day rainfall prediction results for the period of Jan. 1~Jun. 20, 2006, showed reasonable predictability that the total RQPM precipitation amounts to 89.7% of the observed precipitation. The streamflow forecast associated with 2-day RQPM followed the observed hydrograph pattern with high accuracy even though there occurred missing forecast and false alarm in some rainfall events. However, predictability decrease in downstream station, e.g. Gyuam was found because of the difficulties in parameter calibration of rainfall-runoff model for controlled streamflow and reliability deduction of rating curve at gauge station with large cross section area. The 10-day precipitation prediction using GQPM shows significantly underestimation for the peak and total amounts, which affects streamflow prediction clearly. The improvement of GDAPS forecast using post-processing seems to have limitation and there needs efforts of stabilization or reform for the original NWP.

• Journal of The Geomorphological Association of Korea
• /
• v.20 no.1
• /
• pp.21-33
• /
• 2013

Weight of evidence model was applied for potential mapping of mountainous wetland to reduce the range of the field survey and to increase the efficiency of operations because the surveys of mountainous wetland need a lot of time and money owing to inaccessibility and extensiveness. The relationship between mountainous wetland location and related factors is expressed as a probability by Weight of evidence model. For this, the spatial database consist of slope map, curvature map, vegetation index map, wetness index map, soil drainage rating map was constructed in Yeongnam area, Korea, and weights of evidence based on the relationship between mountainous wetland location and each factor rating were calculated. As a result of correlation analysis between mountainous wetland location and each factors rating using likelihood ratio values, the probability of mountainous wetlands were increased at condition of lower slope, lower curvature, lower vegetation index value, lower wetness value, moderate soil drainage rating. Mountainous Wetland Potential Index(MWPI) was calculated by summation of the likelihood ratio and mountainous wetland potential map was constucted from GIS integration. The mountain wetland potential map was verified by comparison with the known mountainous wetland locations. The result showed the 75.48% in prediction accuracy.

• Asian-Australasian Journal of Animal Sciences
• /
• v.28 no.8
• /
• pp.1090-1094
• /
• 2015

The aim of the study was to select traits that may constitute a prospective criterion for breeding value prediction of young horses. The results of 1,232 starts of 894 four-, five-, six-, and seven-year-old horses, obtained during jumping championships for young horses which had not been evaluated in, alternative to championships, training centres were analyed. Nine traits were chosen of those recorded: ranking in the championship, elimination (y/n), conformation, rating of style on day one, two, and three, and penalty points on day one, two, and three of a championship. (Co)variance components were estimated via the Gibbs sampling procedure and adequate (co)variance component ratios were calculated. Statistical classifications were trait dependent but all fitted random additive genetic and permanent environment effects. It was found that such characteristics as penalty points and jumping style are potential indicators of jumping ability, and the genetic variability of the traits was within the range of 14% to 27%. Given the low genetic correlations between the conformation and other results achieved on the parkour, the relevance of assessment of conformation in four-years-old horses has been questioned.

• Journal of Korean Society of Forest Science
• /
• v.89 no.1
• /
• pp.49-54
• /
• 2000

Stand level survival prediction model was developed that incorporated the incidence of fusiform rust(Cronartium quercuum [Berk.] Miyabe ex Shirai f. sp. fusiforme) and allowed the transition of trees from an uninfected stage to an infected stage. The influence of hardwood interspecific competition on the survival of unthinned planted stands of loblolly pine (Pinus taeda L.) was analyzed by using of information from twelve years of tracking a set of permanent plots representing a broad range of plantation parameters. Significant interaction effects between site index and hardwood basal area per acre were revealed in the survival model. Survival of the planted pines decreased with increasing density of hardwood trees per acre and site index as the productivity rating of the forest land. The effects of hardwood trees interspecific competition on loblolly pine tended to show a negative effect on predicted future number of planted pine trees.