• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.3
• /
• pp.125-132
• /
• 2022

Sentence compression is a natural language processing task that generates concise sentences that preserves the important meaning of the original sentence. For grammatically appropriate sentence compression, early studies utilized human-defined linguistic rules. Furthermore, while the sequence-to-sequence models perform well on various natural language processing tasks, such as machine translation, there have been studies that utilize it for sentence compression. However, for the linguistic rule-based studies, all rules have to be defined by human, and for the sequence-to-sequence model based studies require a large amount of parallel data for model training. In order to address these challenges, Deleter, a sentence compression model that leverages a pre-trained language model BERT, is proposed. Because the Deleter utilizes perplexity based score computed over BERT to compress sentences, any linguistic rules and parallel dataset is not required for sentence compression. However, because Deleter compresses sentences only considering perplexity, it does not compress sentences by reflecting the linguistic information of the words in the sentences. Furthermore, since the dataset used for pre-learning BERT are far from compressed sentences, there is a problem that this can lad to incorrect sentence compression. In order to address these problems, this paper proposes a method to quantify the importance of linguistic information and reflect it in perplexity-based sentence scoring. Furthermore, by fine-tuning BERT with a corpus of news articles that often contain proper nouns and often omit the unnecessary modifiers, we allow BERT to measure the perplexity appropriate for sentence compression. The evaluations on the English and Korean dataset confirm that the sentence compression performance of sentence-scoring based models can be improved by utilizing the proposed method.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.2
• /
• pp.235-245
• /
• 2012

Logistic discrimination is an useful statistical technique for quantitative analysis of financial service industry. Especially it is not only easy to be implemented, but also has good classification rate. Generalized additive model is useful for credit scoring since it has the same advantages of logistic discrimination as well as accounting ability for the nonlinear effects of the explanatory variables. It may, however, need too many additive terms in the model when the number of explanatory variables is very large and there may exist dependencies among the variables. Mixtures of factor analyzers can be used for dimension reduction of high-dimensional feature. This study proposes to use the low-dimensional factor scores of mixtures of factor analyzers as the new features in the generalized additive model. Its application is demonstrated in the classification of some real credit scoring data. The comparison of correct classification rates of competing techniques shows the superiority of the generalized additive model using factor scores.

• Tuberculosis and Respiratory Diseases
• /
• v.72 no.1
• /
• pp.15-23
• /
• 2012

Background: The survival of non-small cell lung cancer (NSCLC) patients with brain metastases is reported to be 3~6 months even with aggressive treatment. Some patients have very short survival after aggressive treatment and reliable prognostic scoring systems for patients with cancer have a strong correlation with outcome, often supporting decision making and treatment recommendations. Methods: A total of one hundred twenty two NSCLC patients with brain metastases who received gamma knife radiosurgery (GKRS) were analyzed. Survival analysis was calculated in all patients for thirteen available prognostic factors and four prognostic scoring systems: score index for radiosurgery (SIR), recursive partitioning analysis (RPA), graded prognostic assessment (GPA), and basic score for brain metastases (BSBM). Results: Age, Karnofsky performance status, largest brain lesion volume, systemic chemotherapy, primary tumor control, and medication of epidermal growth factor receptor tyrosine kinase inhibitor were statistically independent prognostic factors for survival. A multivariate model of SIR and RPA identified significant differences between each group of scores. We found that three-tiered indices such as SIR and RPA are more useful than four-tiered scoring systems (GPA and BSBM). Conclusion: There is little value of RPA class III (most unfavorable group) for the same results of 6-month and 1-year survival rate. Thus, SIR is the most useful index to sort out patients with poorer prognosis. Further prospective trials should be performed to develop a new molecular- and gene-based prognostic index model.

• Journal of Gastric Cancer
• /
• v.21 no.4
• /
• pp.368-378
• /
• 2021

Purpose: When patients with early gastric cancer (EGC) undergo non-curative endoscopic submucosal dissection requiring gastrectomy (NC-ESD-RG), additional medical resources and expenses are required for surgery. To reduce this burden, predictive model for NC-ESD-RG is required. Materials and Methods: Data from 2,997 patients undergoing ESD for 3,127 forceps biopsy-proven differentiated-type EGCs (2,345 and 782 in training and validation sets, respectively) were reviewed. Using the training set, the logistic stepwise regression analysis determined the independent predictors of NC-ESD-RG (NC-ESD other than cases with lateral resection margin involvement or piecemeal resection as the only non-curative factor). Using these predictors, a risk-scoring system for predicting NC-ESD-RG was developed. Performance of the predictive model was examined internally with the validation set. Results: Rate of NC-ESD-RG was 17.3%. Independent pre-ESD predictors for NC-ESD-RG included moderately differentiated or papillary EGC, large tumor size, proximal tumor location, lesion at greater curvature, elevated or depressed morphology, and presence of ulcers. A risk-score was assigned to each predictor of NC-ESD-RG. The area under the receiver operating characteristic curve for predicting NC-ESD-RG was 0.672 in both training and validation sets. A risk-score of 5 points was the optimal cut-off value for predicting NC-ESD-RG, and the overall accuracy was 72.7%. As the total risk score increased, the predicted risk for NC-ESD-RG increased from 3.8% to 72.6%. Conclusions: We developed and validated a risk-scoring system for predicting NC-ESD-RG based on pre-ESD variables. Our risk-scoring system can facilitate informed consent and decision-making for preoperative treatment selection between ESD and surgery in patients with EGC.

• International Journal of Quality Innovation
• /
• v.10 no.1
• /
• pp.25-36
• /
• 2009

The purpose of this study is to analyze the efficiency of the service quality activity itself by using the DEA Model, in contrast to previous quality evaluation methods, as an attempt to evaluate the service quality activities of the distribution industry. Furthermore, by complementing the shortfalls of the weighted value of the DEA Model, it recommends a DEA/PS Model that is appropriate in the evaluation of service quality activities. Based on this model, the study proposes the SQAE Model, an evaluation tool to complement the traditional measuring method. According to the results of the analysis of 18 sample distribution businesses, there was a discrepancy by business in the results of the Traditional Scoring System and the Evaluation Measuring System. Therefore, it is most desirable to not only be active in service quality activities but also increase efficiency at the same time.

• Communications of Mathematical Education
• /
• v.27 no.4
• /
• pp.581-609
• /
• 2013

Students' learning processes and mathematical levels should be correctly diagnosed in many different methods of assessment to help students learn mathematics. The study developed the model for the process-based assessment while using manipulatives in the middle school in order to improve problem solving, reasoning and communication which are emphasized in 2009 reformed curriculum as the areas of mathematical process. Identifying the principles of assessment, we created the assessment model for each area and carried out a preliminary study. Based on this, we revised the representative items and the observation checklist and then conducted a main study. Through the results of assessment, we found that students' thinking processes were well presented in scoring rubric for their responses on each item. It meant that the purpose of the assessment as a criterion-referenced test was achieved.

• The Korean journal of internal medicine
• /
• v.39 no.4
• /
• pp.625-639
• /
• 2024

Background/Aims: Intensive care unit (ICU) quality is largely determined by the mortality rate. Therefore, we aimed to develop and validate a novel prognostic model for predicting mortality in Korean ICUs, using national insurance claims data. Methods: Data were obtained from the health insurance claims database maintained by the Health Insurance Review and Assessment Service of South Korea. From patients who underwent the third ICU adequacy evaluation, 42,489 cases were enrolled and randomly divided into the derivation and validation cohorts. Using the models derived from the derivation cohort, we analyzed whether they accurately predicted death in the validation cohort. The models were verified using data from one general and two tertiary hospitals. Results: Two severity correction models were created from the derivation cohort data, by applying variables selected through statistical analysis, through clinical consensus, and from performing multiple logistic regression analysis. Model 1 included six categorical variables (age, sex, Charlson comorbidity index, ventilator use, hemodialysis or continuous renal replacement therapy, and vasopressor use). Model 2 additionally included presence/absence of ICU specialists and nursing grades. In external validation, the performance of models 1 and 2 for predicting in-hospital and ICU mortality was not inferior to that of pre-existing scoring systems. Conclusions: The novel and simple models could predict in-hospital and ICU mortality and were not inferior compared to the pre-existing scoring systems.

• Education of Primary School Mathematics
• /
• v.16 no.3
• /
• pp.229-250
• /
• 2013

The purpose of this study is analyzing 'observation and recommendation letter by teacher', which is being submitted to screen and enhance the utilization of gifted students in accordance with recently introduced gifted students observation, recommendation and screening system. For the purpose, this study will provide with objective securing plan of 'observation and recommendation letter by teacher' by developing an optimum evaluation model. The research findings were as follows: First, the result of analysis on the mathematically gifted students behavior characteristic as appeared in 'observation and recommendation letter by teacher' suggested that the recommending teachers have the tendency of giving superficial statement instead of giving concrete case description. When it was analyzed for frequency by the 'observation and recommendation letter by teacher' analysis framework devised by the author, the teachers showed the tendency of concentrating on specific questions. Meanwhile, there was a tendency that teachers concentrate on specific gifted behavior characteristic or area for which concrete case had been suggested. The reason is believed that such part is easy to observe and state while others are not, or, teachers did not judge the other part as the characteristic of gifted students. Second, the gifted students behavior characteristics as appeared in 'observation and recommendation letter by teacher' were made into scores by Rubric model. When the interrater reliability was analyzed based on these scores, the correlation coefficient of 1st scoring was .641. After a discussion session was taken and 2nd scoring was done 3 weeks later, the correlation coefficient of 2nd scoring increased to .732. The reason is believed that; i) the severity among scorers was adjusted by the discussion session after the 1st scoring, ii) the scorers established detail judgment standard on various situations which can appear because of the descriptive nature, and, (iii) they found a consensus on scoring for a new situation appeared. It implies that thorough understanding and application of scorers on evaluation model is as important as the development of optimum model for the differentiation of mathematically gifted elementary students.

• The Journal of Korean Association of Computer Education
• /
• v.20 no.2
• /
• pp.1-11
• /
• 2017

The demands of computer science education for non-majors in higher education is increasing but relevant evaluation tools for the students' computational artifacts are lack. This research aims to develop a scoring rubric to assess student's computational artifacts in non-major programming course at Computational Thinking point of view. The rubric was developed based on 'CT Practice Design Pattern' as a framework. The rubric consists of 'domain, skills, evaluation, evaluating resources, and scales'. Domains are 'Design of abstract model', 'Design and application of creative artifacts', and 'Analysis of the artifacts'. Experts reviewed the rubric to ensure contents validity. The rubric is resulted in reliable for consistency. This rubric can be revised and applied to application environment accordingly.

• Gut and Liver
• /
• v.12 no.6
• /
• pp.674-681
• /
• 2018

Background/Aims: The cumulative surgery rate and postoperative relapse of intestinal Behçet's disease (BD) have been reported to be high. This study aimed to establish a scoring system based on follow-up endoscopic findings that can predict intestinal BD recurrence after surgery. Methods: Fifty-four patients with intestinal BD who underwent surgery due to bowel complications and underwent follow-up colonoscopy were retrospectively investigated. Their clinical data, including colonoscopic findings, were retrieved. Classification and regression tree analysis was used to develop an appropriate endoscopic classification model that can explain the postsurgical recurrence of intestinal BD most accurately based on the following classification: e0, no lesions; e1, solitary ulcer <20 mm in size; e2, solitary ulcer ${\geq}20mm$ in size; and e3, multiple ulcers regardless of size. Results: Clinical relapse occurred in 37 patients (68.5%). Among 38 patients with colonoscopic recurrence, only 29 patients had clinically relapsed. Multivariate analysis identified higher disease activity index for intestinal BD at colonoscopy (hazard ratio [HR], 1.013; 95% confidence interval [CI], 1.005 to 1.021; p=0.002) and colonoscopic recurrence (HR, 2.829; 95% CI, 1.223 to 6.545; p=0.015) as independent risk factors for clinical relapse of intestinal BD. Endoscopic findings were classified into four groups, and multivariate analysis showed that the endoscopic score was an independent risk factor of clinical relapse (p=0.012). The risk of clinical relapse was higher in the e3 group compared to the e0 group (HR, 6.284; 95% CI, 2.036 to 19.391; p=0.001). Conclusions: This new endoscopic scoring system could predict clinical relapse in patients after surgical resection of intestinal BD.