• ETRI Journal
• /
• v.46 no.1
• /
• pp.137-153
• /
• 2024

This study introduces a new approach in Korean morphological analysis combining dictionary-based techniques with Transformer-based deep learning models. The key innovation is the use of a BERT-based reranking system, significantly enhancing the accuracy of traditional morphological analysis. The method generates multiple suboptimal paths, then employs BERT models for reranking, leveraging their advanced language comprehension. Results show remarkable performance improvements, with the first-stage reranking achieving over 20% improvement in error reduction rate compared with existing models. The second stage, using another BERT variant, further increases this improvement to over 30%. This indicates a significant leap in accuracy, validating the effectiveness of merging dictionary-based analysis with contemporary deep learning. The study suggests future exploration in refined integrations of dictionary and deep learning methods as well as using probabilistic models for enhanced morphological analysis. This hybrid approach sets a new benchmark in the field and offers insights for similar challenges in language processing applications.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.411-416
• /
• 1991

Operation simulation is a key factor to evaluate investment and operation in a power utility. Probabilistic production simulation is of major concern. With pumped-storage plant, production simulation is not an easy task, because its economy should fully be exploited. In addition, usual operation interval is a week rather than a day. Most existing models are based on approximate production simulation such as adopting simple priority orders of generations. This study is based on the more elaborate model developed by authors. Further, a policy of weekly coordination is established based on the Maximum Principle. Chronological load curve instead of usual load duration curve is used and the accuracy in simulation is enhahced. Resulting economics are compared. Deviation between these two toad curve is shown.

• Transactions of the Korean Society of Pressure Vessels and Piping
• /
• v.10 no.1
• /
• pp.51-55
• /
• 2014

Based on the results of Probabilistic Safety Assessment(PSA) for a Nuclear Power Plant (NPP), Common Cause Failure(CCF) events have been recognized as one of the main contributors to the risk. Also, the CCF data and estimation method used in domestic PSA models have been pointed out as an issue with respect to the quality. The existing method of MGL and non-staggered testing even widely used were considered conservative in estimating the safety and had a limited capability in uncertainty analyses. Therefore, this paper presents the CCF estimation using a new generic data source and Alpha factor method. The analyses showed that Alpha factor and staggered method are effective in estimating the CCF contribution and risk insights of reference plant. This method will be a common bases for the optimization of new design for the construction plants as well as for the updating of safety assessment on the operating nuclear power plants.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.23 no.1
• /
• pp.109-141
• /
• 1998

Many processes in the industrial realm exhibit sstochastic and nonlinear behavior. Consequently, an intelligent system must be able to nonlinear production processes as well as probabilistic phenomena. In order for a knowledge based system to control a manufacturing processes as well as probabilistic phenomena. In order for a knowledge based system to control manufacturing process, an important capability is that of prediction : forecasting the future trajectory of a process as well as the consequences of the control action. This paper examines the robustness of data mining tools under varying levels of noise while predicting nonlinear processes, includinb chaotic behavior. The evaluated models include the perceptron neural network using backpropagation (BPN), the recurrent neural network (RNN) and case based reasoning (CBR). The concepts are crystallized through a case study in predicting a chaotic process in the presence of various patterns of noise.

• Computers and Concrete
• /
• v.18 no.1
• /
• pp.69-97
• /
• 2016

Performance-based remaining life assessment of reinforced concrete bridge girders, subject to chloride-induced corrosion of reinforcement, is addressed in this paper. Towards this, a methodology that takes into consideration the human judgmental aspects in expert decision making regarding condition state assessment is proposed. The condition of the bridge girder is specified by the assignment of a condition state from a set of predefined condition states, considering both serviceability- and ultimate- limit states, and, the performance of the bridge girder is described using performability measure. A non-homogeneous Markov chain is used for modelling the stochastic evolution of condition state of the bridge girder with time. The thinking process of the expert in condition state assessment is modelled within a probabilistic framework using Brunswikian theory and probabilistic mental models. The remaining life is determined as the time over which the performance of the girder is above the required performance level. The usefulness of the methodology is illustrated through the remaining life assessment of a reinforced concrete T-beam bridge girder.

• Proceedings of the Korea Concrete Institute Conference
• /
• 2009.05a
• /
• pp.429-430
• /
• 2009

The flexural strength of reinforced concrete members is strongly governed by mechanical properties of reinforcing bars, especially by yield strength, which have many uncertainties. The correct choice of probabilistic models for yield strength of reinforcement is an essential step to assure the safety and reliability of members. In this paper, a probabilistic model of yield strength of reinforcing bars is proposed based on literature and own experimental data.

• Proceedings of the Korea Concrete Institute Conference
• /
• 2010.05a
• /
• pp.151-152
• /
• 2010

The objective of this study is to propose the new punching shear model for two-way concrete slabs of building structures and bridge decks structures reinforced with FRP or steel rebars. To do this, two evaluating methods are applied here. One is the ratio of test to model and the other is probability analysis with probabilistic uncertainties. In conclusion, it shows that the proposed punching shear model evaluates the tested punching shear strength as conservative with stability and it exhibits better probabilistic characteristics than existing punching shear models.

• Journal of Korean Institute of Industrial Engineers
• /
• v.40 no.3
• /
• pp.313-320
• /
• 2014

We consider the problem to find economic inventory quantity of a single commodity under stochastic demands and order cancellation. In contrast to the traditional economic production quantity (EPQ) model, we assume that once the amount of inventory reaches to a predetermined level of quantity then the production is not halted but its production speed decreases until the inventory level drops to zero. We establish two probabilistic models representing the behaviors of both the high-production period and low-production period, respectively, and derive the relationship between the level of inventory and costs of production, cancellation, and holding, from which the quantity of economic inventory is obtained.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1991.10a
• /
• pp.147-159
• /
• 1991

Stochastic convexity(concvity) of a stochastic process is a very useful concept for various stochastic optimization problems. In this study we first establish stochastic convexity of a certain class of Markov additive processes through the probabilistic construction based on the sample path approach. A Markov additive process is obtained by integrating a functional of the underlying Markov process with respect to time, and its stochastic convexity can be utilized to provide efficient methods for optimal design or for optimal operation schedule of a wide range of stochastic systems. We also clarify the conditions for stochatic monotonicity of the Markov process, which is required for stochatic convexity of the Markov additive process. This result shows that stochastic convexity can be used for the analysis of probabilistic models based on birth and death processes, which have very wide application area. Finally we demonstrate the validity and usefulness of the theoretical results by developing efficient methods for the optimal replacement scheduling based on the stochastic convexity property.

• Journal of the Korean Society for Precision Engineering
• /
• v.13 no.10
• /
• pp.130-138
• /
• 1996

In general, the scattler is observed in fatigue test data due to the nonhomogeneity of a material. Consequently. It is necessary to use the statistical method to describe the fatigue crack growth process precisely. Bogdanoff and Kozin suggested and developed the B-model which is the probabilistic models of cumulative damage using the Markov process in order to describe the damage process. But the B-model uses only constant probability ratior(r), so it is not consistent with the actual damage process. In this study, the r-decreasing model using a monotonic decreasing function is introduced to improve the B-model. To verify the model, thest data of fatigue crack growth of A12024-T351 and A17075-T651 are used. Compared with the empirical distribution of test data, the distribution from the r-decreasing model is satisfactory and damage process is well described from the probabilistic and physical viewpoint.