• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.25-29
• /
• 2019

딥러닝을 이용한 한국어 의존 파싱은 전이 기반 방식과 그래프 기반 방식으로 나뉘어 연구되어 왔다. 전이 기반 방식은 입력 버퍼와 스택으로부터 자질을 추출하여 모델을 통해 액션을 결정하고 액션에 따라 파스트리를 생성해 나가는 상향식(Botton-Up)의 지역적 모델이고 그래프 기반 방식은 문장 내의 모든 단어에 대해 지배소, 의존소가 될 수 있는 점수를 딥러닝 모델을 통해 점수화하여 트리를 생성하는 전역적 모델이다. 본 논문에서는 Dual Decomposition을 이용하여 하이브리드 방식으로 전이 기반 파서와 그래프 기반 파서를 결합하는 방법을 제안하고 BERT 언어 모델을 반영하여 세종 데이터 셋에서 UAS 94.47%, LAS 92.58% 그리고 SPMRL '14 데이터 셋에서 UAS 94.74%, UAS 94.20%의 성능을 보여 기존 그래프 기반 파서의 성능을 더욱 개선하였다.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2005.05a
• /
• pp.130-134
• /
• 2005

본 연구는 애너그램(anagram) 문제해결 과제에서 발생하는 제약들(영어 철자연쇄 집합의 출현 확률과 단어 내 위치 확률)을 동시에 병렬적으로 만족시키는 확률 모델 학습과정을 보인다. 애너그램에 관한 많은 선행연구들은 이 문제해결 과정이 단순히 기호처리적인 층위뿐만 아니라 하위기호적(subsymbolic) 층위에서의 상향식 처리로 인해 일어남을 밝혀왔고, 주로 영어 철자의 연쇄체의 확률값을 이용해왔다. 본 연구는 확률 라이브러리 모델(Probabilistic Library Model)을 통해 애너그램 문제해결이 한 번씩 끝날 때마다 철자 연쇄체의 출현 및 위치 분포 확률이 어떻게 유연한 변화를 갖는지에 집중한다. 하나의 문제를 풀고 나면 본 모델은 그 전 문제를 풀었을 때의 상태 패턴으로부터 변화를 보인다. 이러한 분포 변화를 통해 하위기호적 활동의 영향이 문제해결에 있어서 학습구조의 유연한 변화에 중요한 영향을 끼친다는 점을 확인했다.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.2
• /
• pp.347-356
• /
• 1999

We present a method of thesaurus construction in order to produce semantic concept hierarchy of Korean nouns and implement a system for constructing the thesaurus in this paper. Multiple-step construction method is applied to this system which uses bottom-up and top-down method complementarily for solving the non-objectivity of word hierarchy, working speed, structural contradiction and incoherency of existing thesaurus. This system maintains objectivity using the meaning sentence of machine-readable dictionary and solves structural contradiction and incoherency with reference to existing thesaurus. We implement a developmental tool based on client/server system to construct thesaurus including massive entries as soon as possible and multiple client can work simultaneously.

• Journal of Aquaculture
• /
• v.16 no.2
• /
• pp.76-83
• /
• 2003

To improve the efficiency of removal of bacteria and virus with DynaSand Filters used for treatment of recycling wastewater from aquaculture, effect of biofilm formation on bacterial transport through coated sand was estimated. At the neutral pH (pH 7.0), the coated sand was positive of zeta potential (surface charge). Column experiments were also carried out to test the effect of uncoated sand as well as coated sand with Al and Fe. The coated sand influenced more significantly the surface properties, adsorption and transport than the uncoated sand. The leaching batch system investigated for synthetic water showed concentrations of 7.47, 4.80, 20.89 and 7.23 mg/L for the uncoated sand, coated sand with Al, Fe and Al+Fe, respectively. Hence there are significant differences among the tested coatings with reference to bacterial transport and surface properties.

• Journal of the Korea Computer Industry Society
• /
• v.4 no.12
• /
• pp.1103-1111
• /
• 2003

Information retrieval system with indexing system matching single keyword is simple and popular. But with single keyword matching it is very hard to represent the exact meaning of documents and the set of documents from retrieval is very large, therefore it can't satisfy the user of the information retrieval systems. This paper proposes a phrase-based indexing system based on the phrase, the larger syntax unit than a single keyword. Web documents include lots of syntactic errors, the natural language parser with high Quality cannot be expected in Web. Partial trees, even not a full tree, from fully bottom-up parsing is still useful for extracting phrases, and they are much more discriminative than single keyword for index. It helps the information retrieval system enhance the efficiency and reduce the processing overhead.

• Korean Journal of Cognitive Science
• /
• v.10 no.3
• /
• pp.1.2-1.2
• /
• 1999

시각적 선택에 대한 과거 정신물리학적, 신경 생리학적 연구결과를 토대로 Feature Gate 라는 신경 망 모형을 제안하였다. 이 모형에는 공간 배치도가 위계 적으로 구성되어 있으며, 정보의 흐름이 위계의 각 수준으로부터 그 다음 수준으로 넘어갈 때 주의 게이트에 의해 조절되도록 되어 있다. 주의 게이트들은 독특한 세부 특징을 가진 위치에 반응하는 상향식 시스템과 표적 세부 특징이 있는 위치에 반응하는 하향식 기제 모두에 의해 조절된다. 본 연구는 Feature Gate 모형의 하향식 기제에 초점을 맞추어 모형을 설명하고, 현재 다른 모형들이 설명하지 못하는 Moran & Desimone(1985)의 연구결과를 이 모형이 어떻게 설명하는지를 제시하고자 한다. Feature Gate 모형은 병렬 적인 세부특징 검색, 계열 적 접합표적 검색, 단서에 의한 주의의 점진적 감소 모형, 세부특징-주도적인 공간적 선택, 주의의 분할, 방해자극 위치의 억제, 주변 억제 등을 포함한 시각적 주의 연구의 여러 가지 많은 현상들을 설명하는데 하나의 일관적인 해석을 제공해 준다. 앞으로 이 모형을 더욱 확장, 발전 시켜 세부특징의 조합된 배열에 반응하는 상위 수준의 유닛을 사용한다면 시각적 선택과정이 포함된 형태 재인 모형으로 개발될 수 있다.

• Journal of Korean language education
• /
• v.23 no.1
• /
• pp.77-100
• /
• 2012

This study investigates issues in relation to text segmenting, in particular, line breaks in Korean language textbooks. Research on L1 and L2 reading has shown that readers process texts by chunking (grouping words into phrases or meaningful syntactic units) and, therefore, phrase-cued texts are helpful for readers whose syntactic knowledge has not yet been fully developed. In other words, it would be important for language textbooks to avoid awkward syntactic divisions at the end of a line, in particular, those textbooks for beginners and intermediate level learners. According to our analysis of a number of major Korean language textbooks for beginner-level learners, however, many textbooks were found to display line-breaks of awkward syntactic division. Moreover, some textbooks displayed frequent instances where a single word (or eojeol in the case of Korean) is split between different lines. This can hamper not only learners' learning of the rules of spaces between eojeols in Korean, but also learners' development in automatic word recognition, which is an essential part of reading processes. Based on the findings of our textbook analysis and of existing research on reading, this study suggests ways to overcome awkward line-breaks in Korean language textbooks.

• Journal of Korea Port Economic Association
• /
• v.39 no.3
• /
• pp.47-60
• /
• 2023

A thorough inventory of ship emissions, particularly ship's emission of in-port area is necessary to identify significant sources of exhaust gases such as NOx, SOx, PM, and CO₂ and trends in emission levels over time, and reduce their serious effects on the environment and human health. Therefore, the goal of this study is to assess the volume of emissions from ships in Mokpo port, which serves as a gateway to the southwest coast of Korea, using a bottom-up methodology and data from the automatic identification system (AIS) and the Korean Port Management Information System (Port-MIS). In this work, an analysis of ship movement utilizing AIS data and an actual set of data on ship specification were gathered. By examining ship movement using AIS data, We also proposed a new approach for identifying cruising/maneuvering mode. Finally, the results were classified by ship operating mode, by exhaust gas, by ship type, and by berth, which provides a thorough and in-depth analysis of the air pollution caused by ships in Mokpo port.

• The Transactions of the Korea Information Processing Society
• /
• v.1 no.2
• /
• pp.172-183
• /
• 1994

In the multidatabase systems(MDBS), local autonomy and global consistency are important issues. Global consistency could be maintained by a global concurrency control algorithm and a global recovery algorithm. In this thesis, we propose a global concurrency control algorithm to ensure local autonomy and to guarantee global serializability, and a global recovery algorithm which is possible to recover the multudatabase from any failures. The proposed global concurrency control algorithm uses bottom-up approach, based on three-level transaction processing model. It can produce a local history that the execution order of subtransactions is identical to their serialization order by using dummy-operations in the server when an indirect conflict is caused between subtransactions due to local transactions. At the global module, it can efficiently validate global serializability of global transactions by checking global serializability only for the global transactions which conflict with each other.

• The KIPS Transactions:PartB
• /
• v.17B no.2
• /
• pp.139-148
• /
• 2010

This paper shows a reinforcement post-processing method and feedback algorithm for improvement of assigning method in classification. Especially, we focused on complex documents that are generally considered to be hard to classify. A basis factors in traditional classification system are training methodology, classification models and features of documents. The classification problem of the documents containing shared features and multiple meanings, should be deeply mined or analyzed than general formatted data. To address the problems of these document, we proposed a method to expand classification scheme using decision boundary detected automatically in our previous studies. The assigning method that a document simply decides to the top ranked category, is a main factor that we focus on. In this paper, we propose a post-processing method and feedback algorithm to analyze the relevance of ranked list. In experiments, we applied our post-processing method and one time feedback algorithm to complex documents. The experimental results show that our system does not need to change the classification algorithm itself to improve the accuracy and flexibility.