• Annual Conference of KIPS
• /
• 2022.11a
• /
• pp.583-585
• /
• 2022

최근 기술의 발전으로 인해 자연어 처리 모델의 성능이 증가하고 있다. 그에 따라 평문 지문이 아닌 KorQuAD 2.0 과 같은 웹 문서를 지문으로 하는 기계 독해 과제를 해결하려는 연구가 증가하고 있다. 최근 기계 독해 과제의 대부분의 모델은 트랜스포머를 기반으로 하는 추세를 보인다. 그 중 대표적인 모델인 BERT 는 문자열의 순서에 대한 정보를 임베딩 과정에서 전달받는다. 한편 웹 문서는 태그 구조가 존재하므로 문서를 이해하는데 위치 정보 외에도 태그 정보도 유용하게 사용될 수 있다. 그러나 BERT 의 기존 임베딩은 웹 문서의 태그 정보를 추가적으로 모델에 전달하지 않는다는 문제가 있었다. 본 논문에서는 BERT 에 웹 문서 태그 정보를 효과적으로 전달할 수 있는 HTML 임베딩 기법 및 이를 위한 전처리 기법으로 HTML 태그 스택을 소개한다. HTML 태그 스택은 HTML 태그의 정보들을 추출할 수 있고 HTML 임베딩 기법은 이 정보들을 BERT 의 임베딩 과정에 입력으로 추가함으로써 웹 문서 질의 응답 과제의 성능 향상을 기대할 수 있다.

• Annual Conference of KIPS
• /
• 2023.05a
• /
• pp.714-715
• /
• 2023

최근들어 청원 시스템은 사람들의 다양한 의견을 반영하고 대응하기 위한 중요한 수단으로 부상하고 있다. 그러나 많은 양의 청원 글들을 수작업으로 분류하는 것은 매우 시간이 많이 소요되며, 인적 오류가 발생할 수 있는 문제점이 존재한다. 이를 해결하기 위해 자연어처리(NLP) 기술을 활용한 청원 분류 시스템을 개발하는 것이 필요하다. 본 연구에서는 BERT(Bidirectional Encoder Representations from Transformers)[1]를 기반으로 한 텍스트 필터링 시스템을 제안한다. BERT 는 최근 자연어 분류 분야에서 상위 성능을 보이는 모델로, 이를 활용하여 청원 글을 분류하고 분류된 결과를 이용해 해당 글의 노출여부를 결정한다. 본 논문에서는 BERT 모델의 이론적 배경과 구조, 그리고 미세 조정 학습 방법을 소개하고, 이를 활용하여 청원 분류 시스템을 구현하는 방법을 제시한다. 우리가 제안하는 BERT 기반의 텍스트 필터링 시스템은 청원 글 분류를 자동화하고, 이에 따른 대응 속도와 정확도를 향상시킬 것으로 기대된다. 또한, 이 시스템은 다양한 분야에서 응용 가능하며, 대용량 데이터 처리에도 적합하다. 이를 통해 대학 청원 시스템에서 혐오성 발언 등 부적절한 내용을 사전에 방지하고 학생들의 의견을 효율적으로 수집할 수 있는 기능을 제공할 수 있다는 장점을 가지고 있다.

• Journal of the Korean Society of Industry Convergence
• /
• v.27 no.4_2
• /
• pp.925-933
• /
• 2024

Identification of potential technology convergence opportunities is crucial to drive innovation and growth in modern enterprises. In this study, we proposed a framework to explore technological convergence opportunities based on CPC code sequences from patents by utilizing the BERT model. We relied on the BERT architecture to train a new model using about 1.3 million patents registered at the Korean Intellectual Property Office, and achieved an accuracy of approximately 73% based on HitRate@10 metric. A case study using patents related to wearable technologies was conducted to demonstrate practicability and effectiveness of the proposed framework. The key contributions of this research include: (1) enabling in-depth analysis that takes into account the complex interactions between CPC codes and contextual variability; (2) enabling the exploration of diverse technology convergence scenarios beyond simple sequential patterns. This study is one of the first studies to apply the BERT model for exploring technology convergence opportunities, and is expected to contribute to the establishment of technology innovation and R&D strategies by providing a more accurate and practical tool for enhancing the speed and efficiency of technology opportunity-related decision-making processes.

• Journal of Internet Computing and Services
• /
• v.21 no.3
• /
• pp.123-131
• /
• 2020

Along with the rapid development of artificial intelligence technology, natural language processing, which deals with human language, is also actively studied. In particular, BERT, a language model recently proposed by Google, has been performing well in many areas of natural language processing by providing pre-trained model using a large number of corpus. Although BERT supports multilingual model, we should use the pre-trained model using large amounts of Korean corpus because there are limitations when we apply the original pre-trained BERT model directly to Korean. Also, text contains not only vocabulary, grammar, but contextual meanings such as the relation between the front and the rear, and situation. In the existing natural language processing field, research has been conducted mainly on vocabulary or grammatical meaning. Accurate identification of contextual information embedded in text plays an important role in understanding context. Knowledge graphs, which are linked using the relationship of words, have the advantage of being able to learn context easily from computer. In this paper, we propose a system to extract Korean contextual information using pre-trained BERT model with Korean language corpus and knowledge graph. We build models that can extract person, relationship, emotion, space, and time information that is important in the text and validate the proposed system through experiments.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.4
• /
• pp.169-178
• /
• 2022

Morphemes are most primitive units in a language that lose their original meaning when segmented into smaller parts. In Korean, a sentence is a sequence of eojeols (words) separated by spaces. Each eojeol comprises one or more morphemes. Korean morphological analysis (KMA) is to divide eojeols in a given Korean sentence into morpheme units. It also includes assigning appropriate part-of-speech(POS) tags to the resulting morphemes. KMA is one of the most important tasks in Korean natural language processing (NLP). Improving the performance of KMA is closely related to increasing performance of Korean NLP tasks. Recent research on KMA has begun to adopt the approach of machine translation (MT) models. MT is to convert a sequence (sentence) of units of one domain into a sequence (sentence) of units of another domain. Neural machine translation (NMT) stands for the approaches of MT that exploit neural network models. From a perspective of MT, KMA is to transform an input sequence of units belonging to the eojeol domain into a sequence of units in the morpheme domain. In this paper, we propose a deep learning model for KMA. The backbone of our model is based on the BERT-fused model which was shown to achieve high performance on NMT. The BERT-fused model utilizes Transformer, a representative model employed by NMT, and BERT which is a language representation model that has enabled a significant advance in NLP. The experimental results show that our model achieves 98.24 F1-Score.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.11
• /
• pp.57-66
• /
• 2022

In this paper, we propose a legal document search method that uses the Sentence-BERT model. The general public who wants to use the legal search service has difficulty searching for relevant precedents due to a lack of understanding of legal terms and structures. In addition, the existing keyword and text mining-based legal search methods have their limits in yielding quality search results for two reasons: they lack information on the context of the judgment, and they fail to discern homonyms and polysemies. As a result, the accuracy of the legal document search results is often unsatisfactory or skeptical. To this end, This paper aims to improve the efficacy of the general public's legal search in the Supreme Court precedent and Legal Aid Counseling case database. The Sentence-BERT model embeds contextual information on precedents and counseling data, which better preserves the integrity of relevant meaning in phrases or sentences. Our initial research has shown that the Sentence-BERT search method yields higher accuracy than the Doc2Vec or TF-IDF search methods.

• Journal of the Korean Society for information Management
• /
• v.39 no.3
• /
• pp.293-310
• /
• 2022

In this study, we analyzed the performance of the BERT-based document classification model by automatically classifying documents in the field of library and information science based on the KoBERT. For this purpose, abstract data of 5,357 papers in 7 journals in the field of library and information science were analyzed and evaluated for any difference in the performance of automatic classification according to the size of the learned data. As performance evaluation scales, precision, recall, and F scale were used. As a result of the evaluation, subject areas with large amounts of data and high quality showed a high level of performance with an F scale of 90% or more. On the other hand, if the data quality was low, the similarity with other subject areas was high, and there were few features that were clearly distinguished thematically, a meaningful high-level performance evaluation could not be derived. This study is expected to be used as basic data to suggest the possibility of using a pre-trained learning model to automatically classify the academic documents.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.5
• /
• pp.61-67
• /
• 2024

In recent years, artificial intelligence and deep learning technologies have made significant advances, and the BERT model has been recognized for its excellent contextual understanding in natural language processing based on the transformer architecture. This performance has the potential to take traditional recommendation systems to the next level. In this study, we adopt an approach that combines a collaborative filtering approach with a deep learning model to improve the performance of recommendation systems. Specifically, we implemented a system that uses BERT to analyze the sentiment of user reviews and embed users based on these review sentiments to find and recommend users with similar tastes. In the process, we also utilized Elasticsearch, an open-source search engine, for quick search and retrieval of recommended results. The approach of analyzing users' textual data to increase the accuracy and personalization of recommendations will play an important role in improving the user experience on various online services in the future.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.3
• /
• pp.125-132
• /
• 2022

Sentence compression is a natural language processing task that generates concise sentences that preserves the important meaning of the original sentence. For grammatically appropriate sentence compression, early studies utilized human-defined linguistic rules. Furthermore, while the sequence-to-sequence models perform well on various natural language processing tasks, such as machine translation, there have been studies that utilize it for sentence compression. However, for the linguistic rule-based studies, all rules have to be defined by human, and for the sequence-to-sequence model based studies require a large amount of parallel data for model training. In order to address these challenges, Deleter, a sentence compression model that leverages a pre-trained language model BERT, is proposed. Because the Deleter utilizes perplexity based score computed over BERT to compress sentences, any linguistic rules and parallel dataset is not required for sentence compression. However, because Deleter compresses sentences only considering perplexity, it does not compress sentences by reflecting the linguistic information of the words in the sentences. Furthermore, since the dataset used for pre-learning BERT are far from compressed sentences, there is a problem that this can lad to incorrect sentence compression. In order to address these problems, this paper proposes a method to quantify the importance of linguistic information and reflect it in perplexity-based sentence scoring. Furthermore, by fine-tuning BERT with a corpus of news articles that often contain proper nouns and often omit the unnecessary modifiers, we allow BERT to measure the perplexity appropriate for sentence compression. The evaluations on the English and Korean dataset confirm that the sentence compression performance of sentence-scoring based models can be improved by utilizing the proposed method.

• Journal of Korea Society of Industrial Information Systems
• /
• v.28 no.5
• /
• pp.15-30
• /
• 2023

Recently, generative models based on the Transformer architecture, such as ChatGPT, have been gaining significant attention. The Transformer architecture has been applied to various neural network models, including Google's BERT(Bidirectional Encoder Representations from Transformers) sentence generation model. In this paper, a method is proposed to create a text binary classification model for determining whether a comment on Korean movie review is positive or negative. To accomplish this, a pre-trained multilingual BERT sentence generation model is fine-tuned and transfer learned using a new Korean training dataset. To achieve this, a pre-trained BERT-Base model for multilingual sentence generation with 104 languages, 12 layers, 768 hidden, 12 attention heads, and 110M parameters is used. To change the pre-trained BERT-Base model into a text classification model, the input and output layers were fine-tuned, resulting in the creation of a new model with 178 million parameters. Using the fine-tuned model, with a maximum word count of 128, a batch size of 16, and 5 epochs, transfer learning is conducted with 10,000 training data and 5,000 testing data. A text sentiment binary classification model for Korean movie review with an accuracy of 0.9582, a loss of 0.1177, and an F1 score of 0.81 has been created. As a result of performing transfer learning with a dataset five times larger, a model with an accuracy of 0.9562, a loss of 0.1202, and an F1 score of 0.86 has been generated.