• The Journal of the Korea Contents Association
• /
• v.17 no.4
• /
• pp.576-584
• /
• 2017

Unknown nouns which are not in a dictionary make problems not only morphological analysis but also almost all natural language processing area. This paper describes a recognition method for Korean unknown nouns using strings following nouns such as postposition, suffix and postposition, suffix and eomi, etc. We collect and sort words including nouns from documents and divide a word including unknown noun into two parts, candidate noun and string following the noun, by finding same prefix morphemes from more than two unknown words. We use information of strings following nouns extracted from Sejong corpus and decide unknown noun finally. We obtain 99.64% precision and 99.46% recall for unknown nouns occurred more than two forms in news of two portal sites.

• Journal of Korean Library and Information Science Society
• /
• v.39 no.2
• /
• pp.429-448
• /
• 2008

The number of public libraries is not good enough in Korea. The main reasons why there are not many libraries are: the lack of people's recognition of necessity and importance of public libraries, stingy investment by the central and local governments for building public libraries, and no policy on effective building of public libraries. Now, there should be a policy on building and operating small-size public libraries that the developed country adopts. Redevelopment of local areas and reconstruction of old apartment complex are the best chance of building small-size libraries in those areas. The apartment complex in which this study took place is going to be reconstructed in the near future. A small-size library can be built with a district office and a post-office as a multi-complex building. Building a small-size library with public institutions or business buildings in the place of redevelopment of local areas and reconstruction of old apartment complex is one of the best solution of solving the lack of public libraries in Korea.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2000.05a
• /
• pp.273-278
• /
• 2000

본 연구는 한국어 기능 범주(functional category)의 처리를 살펴보기 위하여 실문법증 환자를 대상으로 실시되었다. 일반적으로 수형도에서 기능 범주는 보문소구(Complementizer Phrase), 시제구(Tenxe Phrase), 일치소구(Agreement Phrase)의 순으로 구성되어 있다. 본 실험은 Grodzinsky(1997)의 수형도 가지치기 가설과 실문법증의 엄밀성 척도를 바탕으로 한국어 기능 범주의 처리에 있어서도 두 가지 가설의 적용이 타당한지 여부를 검증하기 위해 실시되었다. 실험 1은 일치소(Agr)의 처리를 살펴보기 위하여 실시되었는데, 국어의 존칭 선어말 어미를 중심으로 살펴보았다. 국어의 존칭선어말 어미는 크게 '-시-'삽입에 의한 존칭과 다른 어휘로 대체하는 보충법에 의한 존칭으로 구분된다. '-시-'삽입에 의한 존칭은 AgrP의 범주로 보충법에 의한 존칭은 동사구(Verb Phrase)의 범주로 설정하여 문법에 어려움을 보이는 실문법증의 경우 어휘부에 혹하는 보충법에 의한 존칭을 문법범주에 속하는 '-시-'삽입의 존칭보다 더 잘 할 것으로 가정하였다. 실험 2는 AgrP의 범주인 '-시-'선택의 장애가 TPqja주인 시제 선택에 역시 영향을 미칠 것이라는 가정아래 설계되었다. 수형도 상에서 '-시-'는 일치소(Agr)로서 시제(TP)의 아래에 위치하므로 실험 3은 '-시-'삽입 존칭과 시제에서 어려움이 보문소(CP)의 선택에 역시 영향을 미칠 것이라는 가정하에서 실시되었다. 실험 1의 결과 '-시-'삽입에 의한 존칭을 처리하는데 보충법에 의한 존칭보다 어려움을 보였다. 실험 2의 결과 '-시-'삽입 존칭에 어려움을 보인 환자는 시제 선어말 어미를 선택하는데도 어려움을 보임이 확인되었다. 실험 3 역시 실험 1과 실험2에서와 동일하게 처리의 어려움을 보였다. 이러한 실험 결과들은 국어의 존칭과 시제 선어말 어미가 통사부에서 구(XP)와 결합하여 새로운 구를 형성하는 통사적 접사로 해석할 수 있으며 Grodzinsky의 가설을 지지하는 결과를 보여 줌으로서 국어에서도 AgrP, TP, CP 사이의 통사적 위계가 있음을 뒷받침하는 증거가 된다.

• Korean Journal of Cognitive Science
• /
• v.22 no.3
• /
• pp.271-298
• /
• 2011

The lexical semantic system should be built to grasp lexical semantic information more clearly. In this paper, we studied a semantic clustering of predicates that is one of the steps in building the lexical semantic system. Unlike previous studies that used argument of subcategorization(subject and object), selectional restrictions and interaction information of adverb, we used sense tagged definition in dictionary for the semantic clustering of predicate, and also attempted hierarchical clustering of predicate using the relationship between the generic concept and the specific concept. Most of the predicates in the dictionary were used for clustering. Total of 106,501 predicates(85,754 verbs, 20,747 adjectives) were used for the test. We got results of clustering which is 2,748 clusters of predicate and 130 recursive definition clusters and 261 sub-clusters. The maximum depth of cluster was 16 depth. We compared results of clustering with the Sejong semantic classes for evaluation. The results showed 70.14% of the cohesion.

• Proceedings of the Korean Society for Bioinformatics Conference
• /
• 2006.02a
• /
• pp.109-114
• /
• 2006

최근 Human 지놈 프로젝트를 포함한 다양한 종의 지놈 프로젝트가 수행되고 수많은 지놈정보가 생산되고 있으며 이를 해석하고 서로 연관성를 찾기 위한 다양한 연구가 진행되고 있다. 즉 최신 생명공학과 관련된 연구방향이 DNA의 구조적 해석에서 기능 해석과 유전자들의 상호연관성을 규명하는 방향으로 변화하고 있으며 이를 위한 강력한 도구로서 DNA microarray (DNA chip)는 방대한 양의 지놈 정보를 이용하여 단시간에 대량으로 고속처리하여 효율적으로 유전자 기능을 분석할 수 있는 주목받고 있는 방법이다. DNA microarray 실험과 분석에 있어 데이터분석, 재현성, 종간의 비교, 확인실험 및 비용 등의 문제가 있지만 유전자발현양상 데이터로부터 정확한 환자의 예후를 예측할 수 있는 비교적 적은 유전자 그룹의 진단마커를 찾거나, 하나의 유전자가 아니라 mouse 전체 지놈의 유전자발현 패턴을 인간의 암을 위시한 각종 질병 연구를 위한 발현 신호나 변화 등을 발견하여 신약개발 등에 활용하고자 하는 시도가 활발히 진행되고 있다. 서로 다른 종간에 비슷한 phenotype의 유전자발현도 진화적으로 보존되었다는 전제 하에서 지놈 sequence의 비교연구가 가능하고 DNA microarray 발현 데이터에 근거하여 독립적으로 각 종간의 유전자발현패턴을 비교함으로써 난치병 등을 새롭게 분류할 수 있다. 즉, 암세포 등에서 유전자발현 양상은 유전학적, 환경적 alteration들이 잘 반영되어 있다고 간주하고, 이러한 양상을 바탕으로 인간의 암을 위시한 다양한 질병 연구를 위한 최적의 mouse 모델을 찾을 수 있고, 이는 결국 새로운 치료 방법 개발이나 맞춤의학 실현에 중요한 역할을 할 것으로 기대된다. 특히 pathway 타겟으로 하는 치료를 위해서는 Human-mouse 비교를 통한 발현 신호를 찾는 것이 진단에서는 매우 유용한 방법이다. 이를 위한 고성능의 분석방법이나 시스템의 개발이 중요하게 된다.. 관류의 정도와 조영증강정도를 중심으로 관류 MR 영상소견과 조직학적 소견을 관련지어 분석하였다. 결과: 조영증강 T1강조MR영상에서 환상조영증강을 보이는 다형성 교보세포종 2예에서는 변연부 외륜이 고관류를, 중심부의 괴사부위는 저관류로 나타났다. 저등급 교종은 경계가 불분명한 저관류부위로 보였다. 뇌농양 2예는 변연부 외륜이 경도의 고관류를, 중심부는 저관류로 나타났다. 뇌수막종은 미만성의 균일한 중등도 혹은 고도의 고관류로 보였으며, 임파종과 배아종은 경계가 명확한 저관류부위로 나타났다. 신경세포종은 종괴\ulcorner 일부에 중등도 혹은 고도의 고관류부위가 관찰되었고, 전이암은 다수병변중 일부에서 중등도의 고관류를 보였다. 방사선괴사는 저관류부위내에 국소적 고관류부위를 보였다. 결론: 관류 MR영상은 뇌종양의 관류상태를 비교적 잘 반영하며, 조직학적 특성을 예측하는데에 도움을 주 수 있을 것으로 기대된다. 뇌종야에서의 관류MR영상의 분명한 역할을 규명하기 위해서는 앞으로 더 많은 임상적 연구가 필요할 것으로 생각된다.조증 환자의 자극성 전타액내 lactobacilli양은 peroxidase system을 함유한 세치제를 사용한 군에서 대조군에 비해 상대적으로 낮게 나타났으나(p = 0.067) 통계학적 유의성은 없었다.같은 예에서 찾아 볼 수 있다. 첫째, 발음상으로 동사의 변화형에서 "porte[$p{\jmath}rte$](들다: 현재형), porte[$p{\jmath}rte$](과거분사형), porta[$p{\jmath}rte$](단순과거형)"등이 대립되며, 이휘 "Porto[$p{\jmath}rte$](포르토)"와도 대립된다. 둘째, 어휘적 대립 "le haut[$l{\partial}o$](위)/l'eau[lo](물)"와 형태론적 대립 "le[$l{\partial}$](정관사, 남성단수)/l

• Journal of The Korean Association of Information Education
• /
• v.25 no.5
• /
• pp.733-740
• /
• 2021

The purpose of this study is to present the direction of elementary school AI education by analyzing cases of classes related to AI education in actual school settings. For this purpose, 19 classes were collected as elementary school class cases based on AI education. According to the result of analyzing the class case, it was confirmed that the class was designed in a hybrid aspect of learning content and method using AI. As a result of analyzing the achievement standards and learning goals, action verbs related to memory, understanding, and application were found in 8 classes using AI from a tool perspective. When class was divided into introduction, development, and rearrangement stages, the AI education element appeared the most in the development stage. On the other hand, when looking at the ratio of learning content and learning method of AI education elements in the development stage, the learning time for approaching AI education as a learning method was overwhelmingly high. Based on this, the following implications were derived. First, when designing the curriculum for schools and grades, it should be designed to comprehensively deal with AI as a learning content and method. Second, to supplement the understanding of AI, in the short term, it is necessary to secure the number of hours in practical subjects or creative experience activities, and in the long term, it is necessary to secure information subjects.

• Journal of Intelligence and Information Systems
• /
• v.28 no.4
• /
• pp.309-328
• /
• 2022

This study empirically analyzed a Korean pre-trained language models (PLMs) designed for natural language generation. The performance of two PLMs - BART and GPT - at the task of abstractive text summarization was compared. To investigate how performance depends on the characteristics of the inference data, ten different document types, containing six types of informational content and creation content, were considered. It was found that BART (which can both generate and understand natural language) performed better than GPT (which can only generate). Upon more detailed examination of the effect of inference data characteristics, the performance of GPT was found to be proportional to the length of the input text. However, even for the longest documents (with optimal GPT performance), BART still out-performed GPT, suggesting that the greatest influence on downstream performance is not the size of the training data or PLMs parameters but the structural suitability of the PLMs for the applied downstream task. The performance of different PLMs was also compared through analyzing parts of speech (POS) shares. BART's performance was inversely related to the proportion of prefixes, adjectives, adverbs and verbs but positively related to that of nouns. This result emphasizes the importance of taking the inference data's characteristics into account when fine-tuning a PLMs for its intended downstream task.

• 한국정보교육학회:학술대회논문집
• /
• 2021.08a
• /
• pp.377-383
• /
• 2021

The purpose of this study is to present the direction of elementary school AI education by analyzing cases of classes related to AI education in actual school settings. For this purpose, 19 classes were collected as elementary school class cases based on AI education. According to the result of analyzing the class case, it was confirmed that the class was designed in a hybrid aspect of learning content and method using AI. As a result of analyzing the achievement standards and learning goals, action verbs related to memory, understanding, and application were found in 8 classes using AI from a tool perspective. When class was divided into introduction, development, and rearrangement stages, the AI education element appeared the most in the development stage. On the other hand, when looking at the ratio of learning content and learning method of AI education elements in the development stage, the learning time for approaching AI education as a learning method was overwhelmingly high. Based on this, the following implications were derived. First, when designing the curriculum for schools and grades, it should be designed to comprehensively deal with AI as a learning content and method. Second, to supplement the understanding of AI, in the short term, it is necessary to secure the number of hours in practical subjects or creative experience activities, and in the long term, it is necessary to secure information subjects.

• Information Systems Review
• /
• v.23 no.4
• /
• pp.45-67
• /
• 2021

As the transaction volume of the C2C second-hand market is growing, the number of frauds, which intend to earn unfair gains by sending products different from specified ones or not sending them to buyers, is also increasing. This study explores the model that can identify frauds in the online C2C second-hand market by examining the postings for transactions. For this goal, this study collected 145,536 field data from actual C2C second-hand market. Then, the model is built with the characteristics from postings such as the topic and the linguistic characteristics of the product description, and the characteristics of products, postings, sellers, and transactions. The constructed model is then trained by the machine learning algorithm XGBoost. The final analysis results show that fraudulent postings have less information, which is also less specific, fewer nouns and images, a higher ratio of the number and white space, and a shorter length than genuine postings do. Also, while the genuine postings are focused on the product information for nouns, delivery information for verbs, and actions for adjectives, the fraudulent postings did not show those characteristics. This study shows that the various features can be extracted from postings written in C2C second-hand transactions and be used to construct an effective model for frauds. The proposed model can be also considered and applied for the other C2C platforms. Overall, the model proposed in this study can be expected to have positive effects on suppressing and preventing fraudulent behavior in online C2C markets.

• Cross-Cultural Studies
• /
• v.53
• /
• pp.163-192
• /
• 2018

The purpose of this paper was to examine universal dependency UD application cases of Korean and Japanese with similar morphological characteristics. In addition, UD application and improvement methods of Korean were examined through comparative analysis. Korean and Japanese are very well developed due to their agglutinative characteristics. Therefore, there are many difficulties to apply UD which is built around English refraction. We examined the application of UPOS and DEPREL as components of UD with discussions. In UPOS, we looked at category problem related to narrative such as AUX, ADJ, and VERB, We examined how to handle units. In relation to the DEPREL annotation system, we discussed how to reflect syntactic problem from the basic unit annotation of syntax tags. We investigated problems of case and aux arising from the problem of setting dominant position from Korean and Japanese as the dominant language. We also investigated problems of annotation of parallel structure and setting of annotation basic unit. Among various relation annotation tags, case and aux are discussed because they show the most noticeable difference in distribution when comparing annotation tag application patterns with Korean. The case is related to both Korean and Japanese surveys. Aux is a secondary verb in Korean and an auxiliary verb in Japanese. As a result of examining specific annotation patterns, it was found that Japanese aux not only assigned auxiliary clauses, but also auxiliary elements to add the grammatical meaning to the verb and form corresponding to the end of Korean. In UD annotation of Japanese, the basic unit of morphological analysis is defined as a unit of basic syntactic annotation in Japanese UD annotation. Thus, when using information, it is necessary to consider how to use morphological analysis unit as information of dependency annotation in Korean.