• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.11
• /
• pp.481-492
• /
• 2023

This study explores how to build a Korean dataset to extract information from text using generative large language models. In modern society, mixed information circulates rapidly, and effectively categorizing and extracting it is crucial to the decision-making process. However, there is still a lack of Korean datasets for training. To overcome this, this study attempts to extract information using text-based zero-shot learning using a generative large language model to build a purposeful Korean dataset. In this study, the language model is instructed to output the desired result through prompt engineering in the form of "system"-"instruction"-"source input"-"output format", and the dataset is built by utilizing the in-context learning characteristics of the language model through input sentences. We validate our approach by comparing the generated dataset with the existing benchmark dataset, and achieve 25.47% higher performance compared to the KLUE-RoBERTa-large model for the relation information extraction task. The results of this study are expected to contribute to AI research by showing the feasibility of extracting knowledge elements from Korean text. Furthermore, this methodology can be utilized for various fields and purposes, and has potential for building various Korean datasets.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.12
• /
• pp.4467-4488
• /
• 2014

The bibliometric analysis is widely used for understanding research domains, trends, and knowledge structures in a particular field. The analysis has majorly been used in the field of information science, and it is currently applied to other academic fields. This paper describes the analysis of academic literatures for classifying research domains and for suggesting empty research areas in the telecommunications policy. The application software is developed for retrieving Thomson Reuters' Web of Knowledge (WoK) data via web services. It also used for conducting text mining analysis from contents and citations of publications. We used three text mining techniques: the Keyword Extraction Algorithm (KEA) analysis, the co-occurrence analysis, and the citation analysis. Also, R software is used for visualizing the term frequencies and the co-occurrence network among publications. We found that policies related to social communication services, the distribution of telecommunications infrastructures, and more practical and data-driven analysis researches are conducted in a recent decade. The citation analysis results presented that the publications are generally received citations, but most of them did not receive high citations in the telecommunications policy. However, although recent publications did not receive high citations, the productivity of papers in terms of citations was increased in recent ten years compared to the researches before 2004. Also, the distribution methods of infrastructures, and the inequity and gap appeared as topics in important references. We proposed the necessity of new research domains since the analysis results implies that the decrease of political approaches for technical problems is an issue in past researches. Also, insufficient researches on policies for new technologies exist in the field of telecommunications. This research is significant in regard to the first bibliometric analysis with abstracts and citation data in telecommunications as well as the development of software which has functions of web services and text mining techniques. Further research will be conducted with Big Data techniques and more text mining techniques.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.2
• /
• pp.139-147
• /
• 2012

In this paper, we discuss the structure of biological knowledge discovery system based on text mining and automatic inference. Given a set of biology documents, the system produces a new hypothesis in an integrated manner. The text mining module of the system first extracts the 'event' information of predefined types from the documents. The inference module then produces a new hypothesis based on the extracted results. Such an integrated system can use information more up-to-date and diverse than other automatic knowledge discovery systems use. However, for the success of such an integrated system, the precision of the text mining module becomes crucial, as any hypothesis based on a single piece of false positive information would highly likely be erroneous. In this paper, we propose a probabilistic filtering method that filters out false positives from the extraction results. Our proposed method shows higher performance over an occurrence-based baseline method.

• Journal of Korean Society for Quality Management
• /
• v.50 no.1
• /
• pp.77-89
• /
• 2022

Purpose: In order to improve the audit quality of a company, an in-depth analysis is required to categorize the audit report in the form of a text document containing the details of the external audit. This study introduces a systematic methodology to extract keywords for each group that determines the differences between groups such as 'audit plan' and 'interim audit' using audit reports collected in the form of text documents. Methods: The first step of the proposed methodology is to preprocess the document through text mining. In the second step, the documents are classified into groups using machine learning techniques and based on this, important vocabularies that have a dominant influence on the performance of classification are extracted. In the third step, the association rules for each group's documents are found. In the last step, the final keywords for each group representing the characteristics of each group are extracted by comparing the important vocabulary for classification with the important vocabulary representing the association rules of each group. Results: This study quantitatively calculates the importance value of the vocabulary used in the audit report based on machine learning rather than the qualitative research method such as the existing literature search, expert evaluation, and Delphi technique. From the case study of this study, it was found that the extracted keywords describe the characteristics of each group well. Conclusion: This study is meaningful in that it has laid the foundation for quantitatively conducting follow-up studies related to key vocabulary in each stage of auditing.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.4
• /
• pp.174-180
• /
• 2024

In this paper, we explore an emotion classification method through multimodal learning utilizing wav2vec 2.0 and KcELECTRA models. It is known that multimodal learning, which leverages both speech and text data, can significantly enhance emotion classification performance compared to methods that solely rely on speech data. Our study conducts a comparative analysis of BERT and its derivative models, known for their superior performance in the field of natural language processing, to select the optimal model for effective feature extraction from text data for use as the text processing model. The results confirm that the KcELECTRA model exhibits outstanding performance in emotion classification tasks. Furthermore, experiments using datasets made available by AI-Hub demonstrate that the inclusion of text data enables achieving superior performance with less data than when using speech data alone. The experiments show that the use of the KcELECTRA model achieved the highest accuracy of 96.57%. This indicates that multimodal learning can offer meaningful performance improvements in complex natural language processing tasks such as emotion classification.

• Genomics & Informatics
• /
• v.18 no.2
• /
• pp.14.1-14.11
• /
• 2020

Phenotyping is a major issue for wheat agriculture to meet the challenges of adaptation of wheat varieties to climate change and chemical input reduction in crop. The need to improve the reuse of observations and experimental data has led to the creation of reference ontologies to standardize descriptions of phenotypes and to facilitate their comparison. The scientific literature is largely under-exploited, although extremely rich in phenotype descriptions associated with cultivars and genetic information. In this paper we propose the Wheat Trait Ontology (WTO) that is suitable for the extraction and management of scientific information from scientific papers, and its combination with data from genomic and experimental databases. We describe the principles of WTO construction and show examples of WTO use for the extraction and management of phenotype descriptions obtained from scientific documents.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2015.01a
• /
• pp.19-22
• /
• 2015

본 논문에서는 문장 및 문단에서 키워드의 역할에 따른 가중치에 근거하여 도서 본문에서 주제어를 추출하는 방법을 제안한다. 기존의 주제어 추출 방식은 도서 본문이 아닌 신문이나 논문에 대한 방식이므로 도서 본문에서의 주제어 추출에 그대로 적용하기에는 어려움이 있다. 따라서 본 논문에서는 빈도수뿐만 아니라 문장 내 중요 요소에 대한 가중치와 중요 문장에 대한 가중치를 후보 키워드에 부여하는 방식을 제안하였다. 제안한 계산 방식을 비문학 도서에 대하여 실험한 결과, 빈도수만으로 주제어를 추출한 기존 방식보다 본 논문에서 제안한 방식의 주제어 추출 결과의 정확도가 향상되는 것을 확인하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.05a
• /
• pp.589-590
• /
• 2010

All these images consist a considerable database. What's more, the semantic meanings of images are well presented by the surrounding text and links. But only a small minority of these images have precise assigned keyphrases, and manually assigning keyphrases to existing images is very laborious. Therefore it is highly desirable to automate the keyphrases extraction process. In this paper, we first introduce WWW image annotation methods, based on low level features, page tags, overall word frequency and local word frequency. Then we put forward our method of multi-cues integration image annotation. Also, show multi-cue image annotation method is more superior than other method through an experiment.

• Journal of Information Processing Systems
• /
• v.11 no.2
• /
• pp.248-265
• /
• 2015

In this paper, we propose a maximum entropy-based model, which can mathematically explain the bio-molecular event extraction problem. The proposed model generates an event table, which can represent the relationship between an event trigger and its arguments. The complex sentences with distinctive event structures can be also represented by the event table. Previous approaches intuitively designed a pipeline system, which sequentially performs trigger detection and arguments recognition, and thus, did not clearly explain the relationship between identified triggers and arguments. On the other hand, the proposed model generates an event table that can represent triggers, their arguments, and their relationships. The desired events can be easily extracted from the event table. Experimental results show that the proposed model can cover 91.36% of events in the training dataset and that it can achieve a 50.44% recall in the test dataset by using the event table.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.3
• /
• pp.61-67
• /
• 2009

The proliferation of the Internet grows, according to electronic documents, along with increasing importance of technology in information retrieval. This research is possible to build a more efficient and accurate knowledge-base with unstructured text documents from the Web using to extract knowledge of the core meaning of LGG (Local Grammar Graph). We have built a ontology based on OWL(Web Ontology Language) using the areas of particular stocks up/down patterns created by the extraction and grammar patterns. It is possible for the user can search for meaning and quality of information about the user wants.