• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.111-115
• /
• 2017

목적 지향적 대화 시스템(Goal-oriented dialogue system) 은 텍스트나 음성을 통해 특정한 목적을 수행 할 수 있는 시스템이다. 최근 RNN(recurrent neural networks)을 기반으로 대화 데이터를 end-to-end learning 방식으로 학습하여 대화 시스템을 구축하는데에 활용한 연구가 있다. End-to-end 방식의 학습은 도메인에 대한 지식 없이 학습 데이터 자체만으로 대화 시스템 구축을 위한 학습이 가능하다는 장점이 있지만 도메인 지식을 학습하기 위해서는 많은 양의 데이터가 필요하다는 단점이 존재한다. 이에 본 논문에서는 도메인 특정 지식을 결합하여 end-to-end learning 방식의 학습이 가능한 Hybrid Code Network 구조를 기반으로 한국어로 구성된 식당 예약에 관련한 대화 데이터셋을 이용하여 식당 예약을 목적으로하는 대화 시스템을 구축하는 방법을 제안한다. 실험 결과 본 시스템은 응답 별 정확도 95%와 대화 별 정확도 63%의 성능을 나타냈다.

• Journal of Korean Institute of Industrial Engineers
• /
• v.40 no.6
• /
• pp.580-591
• /
• 2014

This is a report on the history and current status of production planning and control activities in South Korea with regard to Industrial Engineers' (IEs') role. One of the main characteristics of Korean global companies is that they make good use of IEs not only in their production but also in higher strategic fields like supply chain management. In these companies it is well demonstrated that instead of adding small part of IE skills to traditional engineers, it is much more competitive to have IEs work with them as a team since IEs are fully proficient in every kind of productivity and quality improvement tools. It is also observed that IEs become more productive when they have good domain knowledge in their own industries like semi-conductor, automobile, telecommunications, steel, shipbuilding, etc. So teaching the basics for manufacturing processes to IE students seems very important for their future career development.

• BMB Reports
• /
• v.43 no.5
• /
• pp.305-310
• /
• 2010

Syndecans, cell surface heparansulfate proteoglycans, have been proposed to act as cell surface receptors and/or coreceptors to play critical roles in multiple cellular functions. However, recent reports suggest that the function of syndecans can be further extended through shedding, a cleavage of extracellular domain. Shedding constitutes an additional level for controlling the function of syndecans, providing a means to attenuate and/or regulate amplitude and duration of syndecan signals by modulating the activity of syndecans as cell surface receptors. Whether these remaining cleavage products are still capable of functioning as cell surface receptors to efficiently transduce signals inside of cells is not clear. However, shedding transforms cell surface receptor syndecans into soluble forms, which, like growth factors, may act as novel ligands to induce cellular responses by association with other cell surface receptors. It is becoming interestingly evident that shed syndecans also contribute significantly to syndecan functions in cancer biology. This review presents current knowledge about syndecan shedding and its functional significance, particularly in the context of cancer.

• Journal for History of Mathematics
• /
• v.18 no.2
• /
• pp.75-94
• /
• 2005

This study is intended to develop the criterion for evaluating the creative products that mathematically gifted students produce in their education program to enhance the development of creative productive ability. 1 distinguish the mathematical creativity with the creativity in the general domain, and make the production model of the creative mathematical product grounded on the mathematicians' work through the mathematical history. The model has the following components; the mathematical knowledge, the mathematical thinking and the mathematical inquiry skill, surrounding the resultive creative product. The students products are focused on one component of the model. Thus the criterion for the creative products is grounded on the each component of the model. According to it, teachers could evaluate the students'work, which got the validity and the reliability.

• Genomics & Informatics
• /
• v.2 no.3
• /
• pp.142-146
• /
• 2004

Biological data and analysis tools are accumulated in distributed databases and web servers. For this reason, biologists who want to find information from the web should be aware of the various kinds of resources where it is located and how it is retrieved. Integrating the data from heterogeneous biological resources will enable biologists to discover new knowledge across the specific domain boundaries from sequences to expression, structure, and pathway. And inevitably biological databases contain noisy data. Therefore, consensus among databases will confirm the reliability of its contents. We have developed WeSAT that integrates distributed and heterogeneous biological databases and analysis tools, providing through Web Services protocols. In WeSAT, biologists are retrieved specific entries in SWISS-PROT/EMBL, PDB, and KEGG, which have annotated information about sequence, structure, and pathway. And further analysis is carried by integrated services for example homology search and multiple alignments. WeSAT makes it possible to retrieve real time updated data and analysis from the scattered databases in a single platform through Web Services.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.2
• /
• pp.123-128
• /
• 2002

A fuzzy traffic controller with the control rules and the membership functions generated by using genetic algorithm is presented for crossroad management. Conventional fuzzy traffic controllers use control rules and membership functions generated by human operators. However, this approach does not guarantee the optimal solution to design fuzzy control system. Genetic algorithm is a good solution for an optimal problem requiring domain-specific knowledge that is often heuristic. In this paper, we use genetic algorithms to automatically determine the near optimal rules and their membership functions of fuzzy traffic controllers. The effectiveness of our method was shown through simulation of crossroad network.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.8 no.6
• /
• pp.78-84
• /
• 1998

In this paper, a fuzzy traffic controller using genetic algorithms is presented. Conventional fuzzy traffic controllers use membership functions generated by humans. However, this approach does not guarantee the optimal solution to design the fuzzy controller. Genetic algorithm is a good problem solving method requiring domain-specific knowledge that is often heuristic. To find fuzzy membership functions showing good performance, a fitness function must be defined. However it's not easy in traffic control to define such a function as a numeric expression. Thus, we use simulation approach, namely, the fitness value of a solution is determined by use of a performance measure that is obtained by traffic simulator. The proposed method outperforms the conventional fuzzy controllers.

• Journal of the Korean Society for Library and Information Science
• /
• v.44 no.2
• /
• pp.285-308
• /
• 2010

This study analyzes the intellectual structure in domestic research of the Housing field, by utilizing text mining technique. Unlike the existing research that mainly uses text clustering in statistical analyses to identify subject specialties, core authors, and relationships between research areas, this study applied author profiling and factor analysis. To supplement the analysis of intellectual structure generated by text mining, and to perform evaluation on intellectual structure itself, two professionals in the housing field were interviewed. The intellectual structure, generated through text mining, was evaluated and showed its division of valid research areas that is slightly different from the traditional intellectual structure in the housing field.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.454-457
• /
• 2005

Mining frequent patterns in transaction databases, time-series databases, and many other kinds of databases has been studied popularly in data mining research. Most of the previous studies adopt an Apriori-like candidate set generation-and-test approach. However, candidate set generation is still costly, especially when there exist prolific patterns and/or long patterns. And calendar based on temporal association rules proposes the discovery of association rules along with their temporal patterns in terms of calendar schemas, but this approach is also adopt an Apriori-like candidate set generation. In this paper, we propose an efficient temporal frequent pattern mining using TFP-tree (Temporal Frequent Pattern tree). This approach has three advantages: (1) this method separates many partitions by according to maximum size domain and only scans the transaction once for reducing the I/O cost. (2) This method maintains all of transactions using FP-trees. (3) We only have the FP-trees of I-star pattern and other star pattern nodes only link them step by step for efficient mining and the saving memory. Our performance study shows that the TFP-tree is efficient and scalable for mining, and is about an order of magnitude faster than the Apriori algorithm and also faster than calendar based on temporal frequent pattern mining methods.

• Journal of Navigation and Port Research
• /
• v.35 no.7
• /
• pp.575-580
• /
• 2011

It is important to understand the trajectory of structure in launching process because of the short time of launching process may result in unexpected accidents or damage to structures. The high risk of structural failure is not avoidable without the fully comprehension of changing forces in launching procedure. The commercial software can evaluate the motion of launching event in calm water condition but there is the limitation of research application because of the programmed commercial software. The launching process of the spar hull is suggested with stage concept that is divided into 10 stages in time domain. A force equilibrium diagram is derived for each stage where the changes of force vector and motion characteristics take place. In particular, the effects of changes in buoyancy and drag force due to the progressive submergence of the spar hull are taken into account by means of a touch length concept. The results contained in this paper provide the valuable information of the trajectory motion evaluation with suggested methods in spar launching process with sliding barge. Furthermore, the presented stage concept and touch length concept will provide basic knowledge for understanding launching process and help to develop further research area for launching analysis.