• ETRI Journal
• /
• 제43권2호
• /
• pp.299-312
• /
• 2021

It is necessary to achieve high performance in the task of zero anaphora resolution (ZAR) for completely understanding the texts in Korean, Japanese, Chinese, and various other languages. Deep-learning-based models are being employed for building ZAR systems, owing to the success of deep learning in the recent years. However, the objective of building a high-quality ZAR system is far from being achieved even using these models. To enhance the current ZAR techniques, we fine-tuned a pretrained bidirectional encoder representations from transformers (BERT). Notably, BERT is a general language representation model that enables systems to utilize deep bidirectional contextual information in a natural language text. It extensively exploits the attention mechanism based upon the sequence-transduction model Transformer. In our model, classification is simultaneously performed for all the words in the input word sequence to decide whether each word can be an antecedent. We seek end-to-end learning by disallowing any use of hand-crafted or dependency-parsing features. Experimental results show that compared with other models, our approach can significantly improve the performance of ZAR.

• 인지과학
• /
• 제28권2호
• /
• pp.111-131
• /
• 2017

연결주의 모델은 계산주의적 관점에서 언어 처리를 연구하는 한 가지 접근법이다. 그리고 연결주의 모델 연구를 진행하는데 있어서 표상(representation)을 구축하는 것은, 모델의 학습 수준 및 수행 능력을 결정한다는 점에서 모델의 구조를 만드는 것만큼이나 중요한 일이다. 연결주의 모델은 크게 지역 표상(localist representation)과 분산 표상(distributed representation)이라는 두 가지 서로 다른 방식으로 표상을 구축해 왔다. 하지만 종래 연구들에서 사용된 지역 표상은 드문 목표 활성화 값을 갖고 있는 출력층의 유닛이 불활성화 하는 제한점을, 그리고 과거의 분산 표상은 표상된 정보의 불투명성에 의한 결과 확인의 어려움이라는 제한점을 갖고 있었으며 이는 연결주의 모델 연구 전반의 제한점이 되어 왔다. 본 연구는 이와 같은 과거의 표상 구축의 제한점에 대하여, 제한된 볼츠만 머신(restricted Boltzmann machine)이 갖고 있는 특징인 정보의 추상화를 활용하여 지역 표상을 가지고 분산 표상을 유도하는 새로운 방안을 제시하였다. 결과적으로 본 연구가 제안한 방법은 정보의 압축과 분산 표상을 지역 표상으로 역변환하는 방안을 활용하여 종래의 표상 구축 방법이 갖고 있는 문제를 효과적으로 해결함을 보였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 1998년도 하계종합학술대회논문집
• /
• pp.925-928
• /
• 1998

In this paper, we implemented a sign language communication system between korean and japanese using 3D model. Unitl now, we have developed the system through 2D image. The system, however, has some drawbacks base don the limitations of 2D model. Especially it is not comprehensive in the sign language gesture including forward or backward movements becasue it can only display a front view. Therefore, we employed a 3D model for the representation of sign language. The proposed communication system si implemented with windows 95 visual C++ 5.0 imported open inventor library on pentium 233 system. Experimented results show that the 3D system can solve the problems in 2D representation.

• ETRI Journal
• /
• 제44권5호
• /
• pp.794-804
• /
• 2022

Transformer models have had a great impact on natural language processing (NLP) in recent years by realizing outstanding and efficient contextualized language models. Recent studies have used transformer-based language models for various NLP tasks, including Persian named entity recognition (NER). However, in complex tasks, for example, NER, it is difficult to determine which contextualized embedding will produce the best representation for the tasks. Considering the lack of comparative studies to investigate the use of different contextualized pretrained models with sequence modeling classifiers, we conducted a comparative study about using different classifiers and embedding models. In this paper, we use different transformer-based language models tuned with different classifiers, and we evaluate these models on the Persian NER task. We perform a comparative analysis to assess the impact of text representation and text classification methods on Persian NER performance. We train and evaluate the models on three different Persian NER datasets, that is, MoNa, Peyma, and Arman. Experimental results demonstrate that XLM-R with a linear layer and conditional random field (CRF) layer exhibited the best performance. This model achieved phrase-based F-measures of 70.04, 86.37, and 79.25 and word-based F scores of 78, 84.02, and 89.73 on the MoNa, Peyma, and Arman datasets, respectively. These results represent state-of-the-art performance on the Persian NER task.

• 한국언어정보학회지:언어와정보
• /
• 제20권1호
• /
• pp.25-49
• /
• 2016

Honorific forms express the speaker's social attitude to others and also indicate the social ranks and level of intimacy of the participants in the discourse. In a cross-linguistic perspective of grammar engineering, modelling honorification has been regarded as a key strategy for improving language processing applications. Using the HPSG and MRS formalism, this article provides a multilingual grammar model of honorification. The present study incorporates the honorific information into the Meaning Representation System (MRS) via Individual Constraints (ICONS), and then conducts an evaluation to see if the model contributes to semantics-based language processing.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 춘계학술발표대회
• /
• pp.411-414
• /
• 2021

One of the elements of technology that has become extremely critical within the field of education today is Deep learning. It has been especially used in the area of natural language processing, with some word-representation vectors playing a critical role. However, some of the low-resource languages, such as Swahili, which is spoken in East and Central Africa, do not fall into this category. Natural Language Processing is a field of artificial intelligence where systems and computational algorithms are built that can automatically understand, analyze, manipulate, and potentially generate human language. After coming to discover that some African languages fail to have a proper representation within language processing, even going so far as to describe them as lower resource languages because of inadequate data for NLP, we decided to study the Swahili language. As it stands currently, language modeling using neural networks requires adequate data to guarantee quality word representation, which is important for natural language processing (NLP) tasks. Most African languages have no data for such processing. The main aim of this project is to recognize and focus on the classification of words in English, Swahili, and Korean with a particular emphasis on the low-resource Swahili language. Finally, we are going to create our own dataset and reprocess the data using Python Script, formulate the syllabic alphabet, and finally develop an English, Swahili, and Korean word analogy dataset.

• Korean Journal of Radiology
• /
• 제24권10호
• /
• pp.1038-1041
• /
• 2023

• ETRI Journal
• /
• 제43권6호
• /
• pp.1038-1048
• /
• 2021

We propose an end-to-end neural coreference resolution for the Korean language that uses an attention mechanism to point to the same entity. Because Korean is a head-final language, we focused on a method that uses a pointer network based on the head. The key idea is to consider all nouns in the document as candidates based on the head-final characteristics of the Korean language and learn distributions over the referenced entity positions for each noun. Given the recent success of applications using bidirectional encoder representation from transformer (BERT) in natural language-processing tasks, we employed BERT in the proposed model to create word representations based on contextual information. The experimental results indicated that the proposed model achieved state-of-the-art performance in Korean language coreference resolution.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2002년도 Proceedings of International Symposium on Remote Sensing
• /
• pp.860-879
• /
• 2002

Traditionally, Geographical Information Systems can only process spatial data in a procedure-oriented way, and the data can't be treated integrally. This method limits the development of spatial data applications. A new and promising method to solve this problem is the spatial structural query language, which extends SQL and provides integrated accessing to spatial data. In this paper, the theory of spatial structural query language is discussed, and a new geographical data model based on the concepts and data model in OGIS is introduced. According to this model, we implemented a spatial structural query language G/SQL. Through the studies of the 9-Intersection Model, G/SQL provides a set of topological relational predicates and spatial functions for GIS application development. We have successfully developed a Web-based GIS system-WebGIS-using G/SQL. Experiences show that the spatial operators G/SQL offered are complete and easy-to-use. The BNF representation of G/SQL syntax is included in this paper.

• Asia pacific journal of information systems
• /
• 제8권2호
• /
• pp.53-67
• /
• 1998

The explosive growth of the Web is providing end-users access to ever-increasing volumes of information. The resources of legacy systems and relational databases have also been made available to the Web browser, which has become an essential business tool. Recently, model management on the Internet/Web is also proposed with its conceptual design or prototypical system like DecisionNet and DSS Web. However, they are also suffering from the same symptoms as the Web, Although we can identify the elements of a page with HTML tags and (declare) the relationships among the various document elements, they are semantically opaque to computer systems and have no domain-specific meaning. However, HTML is not extensible, so developers are forced to invent convoluted, non-standard solutions for embedding and parsing data. Extensible Markup Language (XML) is a simplified subset of SGML that has many benefits for folks who want to improve structure, maintainability, searchability, presentation, and other aspects of their document management. This paper proposes a structured markup language for model representation and management on the Web as an XML application. The language is based on a conceptual modeling framework, Object-Oriented Structured Modeling (OOSM), which is an extension of the structured modeling.