研究者詳細 - 嶋田　和孝

論文 - 嶋田　和孝

分割表示146 件中 1 - 20 件目／全件表示 >>

Share What You Already Know: Cross-Language-Script Transfer and Alignment for Sentiment Detection in Code-Mixed Data 査読有り国際誌

Pahari N., Shimada K.

ACM Transactions on Asian and Low-Resource Language Information Processing 23 ( 7 ) 2024年07月

担当区分：最終著者記述言語：英語掲載種別：研究論文（学術雑誌）

Code-switching entails mixing multiple languages. It is an increasingly occurring phenomenon in social media texts. Usually, code-mixed texts are written in a single script, even though the languages involved have different scripts. Pre-trained multilingual models primarily utilize the data in the native script of the language. In existing studies, the code-switched texts are utilized as they are. However, using the native script for each language can generate better representations of the text owing to the pre-trained knowledge. Therefore, a cross-language-script knowledge-sharing architecture utilizing the cross-attention and alignment of the representations of text in individual language scripts was proposed in this study. Experimental results on two different datasets containing Nepali-English and Hindi-English code-switched texts, demonstrate the effectiveness of the proposed method. The interpretation of the model using the model explainability technique illustrates the sharing of language-specific knowledge between language-specific representations.

DOI： 10.1145/3661307

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85199966141&origin=inward

Towards Human-Level Evaluation: Assessing the Potential of GPT-4 in Automated Evaluation and Feedback Generation on Japanese Essays 査読有り

Sayaka Nakamoto, Yoshiaki Okamoto, Takashi Nakakouchi, Kazutaka Shimada

Proceedings of the 16th International Conference on Learning Technologies and Learning Environments 2024年07月

Automatic Few-shot Selection on In-Context Learning for Aspect Term Extraction 査読有り

Koki Imazato, Kazutaka Shimada

Proceedings of the 18th International Conference on E-Service and Knowledge Management, 2024年07月

Dataset Construction and Verification for Detecting Factual Inconsistency in Japanese Summarization 査読有り

Keisuke Iwamoto, Kazutaka Shimada

Proceedings of the 15th International Conference on Smart Computing and Artificial Intelligence 2024年07月

Share What You Already Know: Cross-Language-Script Transfer and Alignment for Sentiment Detection in Code-Mixed Data 査読有り

Niraj Pahari, Kazutaka Shimada

ACM Transactions on Asian and Low-Resource Language Information Processing 2024年07月

An utterance is enough to the gaze? Gaze detection from utterance information in multiparty discussion 査読有り

Kensho Wakita, Kazutaka Shimada

Proceedings of the 6th International Conference on Activity and Behavior Computing 2024年05月

Simile Identification with Pseudo Data Acquisition and Re-labeling 査読有り

Jimi Jintaro, Shimada Kazutaka

International Journal of Smart Computing and Artificial Intelligence ( 特定非営利活動法人国際応用情報学研究機構 ) 8 ( 1 ) 1 2024年05月

Layer Configurations of BERT for Multitask Learning and Data Augmentation 査読有り

Pahari N., Shimada K.

Journal of Advanced Computational Intelligence and Intelligent Informatics ( 富士技術出版株式会社 ) 28 ( 1 ) 29 - 40 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（学術雑誌）

Multitask learning (MTL) and data augmentation are becoming increasingly popular in natural language processing (NLP). These techniques are particularly useful when data are scarce. In MTL, knowledge learned from one task is applied to another. To address data scarcity, data augmentation facilitates by providing additional synthetic data during model training. In NLP, the bidirectional encoder representations from transformers (BERT) model is the default candidate for various tasks. MTL and data augmentation using BERT have yielded promising results. However, a detailed study regarding the effect of using MTL in different layers of BERT and the benefit of data augmentation in these configurations has not been conducted. In this study, we investigate the use of MTL and data augmentation from generative models, specifically for category classification, sentiment classification, and aspect-opinion sequence-labeling using BERT. The layers of BERT are categorized into top, middle, and bottom layers, which are frozen, shared, or unshared. Experiments are conducted to identify the optimal layer configuration for improved performance compared with that of single-task learning. Generative models are used to generate augmented data, and experiments are performed to reveal their effectiveness. The results indicate the effectiveness of the MTL configuration compared with single-task learning as well as the effectiveness of data augmentation using generative models for classification tasks.

DOI： 10.20965/jaciii.2024.p0029

Scopus

CiNii Research

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85184850065&origin=inward

Dataset Construction and Evaluation for Aspect-Opinion Extraction in Bangla Fine-Grained Sentiment Analysis 査読有り

Al-Mahmud , Shimada K.

Lecture Notes in Networks and Systems 818 437 - 449 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

In the present age, increasingly more people and entities post their opinions on web-based platforms. With this increasing trend of web-based opinion content, it is quite impossible to digest it by hand. Consequently, a computerized system for opinion and sentiment analysis is required. Sentiment analysis builds systems that automatically extract useful information from massive online reviews. Conventional sentiment analysis simply classifies a document/text into positive/negative/neutral polarity on the basis of overall sentiment polarity. In this case, it is considered that one sentiment is carried for the whole text, which may not be the case in the practical scenario. It is possible to appear more than one aspect, opinion, and sentiment in a single text, especially when the text is long. This motivates the introduction of fine-grained sentiment analysis, in which one can identify what a person/entity is mentioning and how they feel about each aspect/target entity. Aspect-opinion term extraction is a crucial aspect-based sentiment analysis (ABSA). It is a kind of granular-level sentiment analysis. In this study, our target is aspect-opinion terms extraction in the Bangla language. However, no annotated Bangla dataset is available for this task. Hence, we obtained textual data from a publicly available Bangla restaurant dataset, annotated them, and prepared a novel dataset. We then performed the task as a sequence labeling technique by utilizing our dataset. For the experiment, we employed vanilla transformers-based models such as mBERT, BanglaBERT, and BanglishBERT with linear layers and conditional random field (CRF) on top of them. To enhance the models’ performance, we combined different feature embeddings of mBERT, BanglaBERT, and BanglishBERT. We then utilized the linear layers and CRF on top of them. The experimental results indicated that the combined feature embeddings technique significantly improves models’ performance, and (mBERT & BanglishBERT) with linear layer performed best: 0.5840 on the F1 score among all the tested models.

DOI： 10.1007/978-981-99-7862-5_33

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85187643421&origin=inward

Towards Human-Level Evaluation: Assessing the Potential of GPT-4 in Automated Evaluation and Feedback Generation on Japanese Essays 査読有り国際誌

Nakamoto S., Okamoto Y., Nakakouchi T., Shimada K.

Proceedings - 2024 16th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2024 156 - 161 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

In recent years, Automated Writing Evaluation (AWE) has been extensively researched within the field of AI in education. This paper explores generative AI, such as GPT-4, which has garnered significant attention for its ability to score essays and provide feedback to students. We designed prompts for GPT-4 to assign scores and rationales based on a given rubric and to generate feedback beneficial for students' development. We compared the evaluations produced by GPT-4 with those made by human evaluators. The results demonstrate GPT-4's potential to assist in generating evaluations at a human level. In addition, we analysed the consistency of the scoring and the quality of the rationales and feedback generated by GPT-4. In this paper, we will share our analysis and also describe the points that need to be improved for implementation in practice.

DOI： 10.1109/IIAI-AAI63651.2024.00039

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85208091336&origin=inward

Dataset Construction and Verification for Detecting Factual Inconsistency in Japanese Summarization 査読有り国際誌

Iwamoto K., Shimada K.

Proceedings - 2024 16th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2024 243 - 248 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Abstractive document summarization is one of the most important tasks in natural language processing. Many approaches based on large language models have been proposed. However, it is known that the output of LLM often includes hallucinations, such as factual inconsistency. Therefore, detecting factual inconsistencies in a summary is one important task for summarization. One solution for the detection is to utilize machine learning techniques. In general, machine learning approaches require a large number of training data to generate a robust model. However, it is difficult automatically to collect article-summary pairs with factual inconsistency from the Web because hand-written summaries on the Web are usually correct. Moreover, some existing datasets are written in English. In this paper, we propose some approaches to construct Japanese datasets with factual inconsistency automatically. For this purpose, we utilize two approaches from previous studies: FactCC and SumFC. In addition, we propose a new approach to construct summaries with exaggerated expressions, as a variety of factual inconsistencies. We call the datasets JFactCC, JSumFC, and JExnoS. For JExnoS, we utilize a two-stage approach based on GPT-4 and BART for the generation of summaries with exaggerated expressions from correct article-summary pairs. We also verify the usefulness of each constructed dataset through an experiment about factual inconsistency detection with BERT.

DOI： 10.1109/IIAI-AAI63651.2024.00054

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85208094978&origin=inward

Automatic Few-Shot Selection on In-Context Learning for Aspect Term Extraction 査読有り国際誌

Imazato K., Shimada K.

Proceedings - 2024 16th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2024 15 - 20 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

In this paper, we handle aspect term extraction with In-Context Learning (ICL) as the target task. ICL is a method for learning Large Language Models. Instead of updating the model's parameters, prompts are provided to guide it to perform a target task. While the strength of ICL is that the model does not need to undergo additional training, the prompt includes input-output instances that cause instability in accuracy. The instances are called few-shot. Hence, selecting appropriate instances has the most important role in ICL. For this purpose, we propose a selection method with active learning. Active learning is a method of selecting instances that are useful to the model for training from unlabeled data. We regard the active learning-based approach as a sub-task for the target task. We introduce two types of sub-tasks and evaluate the effectiveness of them in the target task.

DOI： 10.1109/IIAI-AAI63651.2024.00012

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85208144512&origin=inward

An Utterance is Enough to the gaze? Gaze Detection from Utterance Information in Multiparty Discussion 査読有り国際誌

Wakita K., Shimada K.

2024 International Conference on Activity and Behavior Computing, ABC 2024 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Gaze information has an important role in conversation understanding. Generally, gaze detection is performed using a camera. However, such systems often require expensive equipment. To solve this problem, we propose a method of eye detection using an utterance during conversations. We first annotate an existing discussion corpus with gaze information and then analyze the annotated data. We apply machine learning to construct a gaze detection model. We utilize BERT, a large-scale pre-Trained language model, and CART, a traditional machine-learning approach based on a decision tree. Experimental results show that the CART-based model was more accurate than the BERT-based model and show the potential possibilities of gaze detection from utterances.

DOI： 10.1109/ABC61795.2024.10651736

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85203829893&origin=inward

Dataset Construction and Evaluation for Aspect-Opinion Extraction in Bangla Fine-grained Sentiment Analysis 査読有り国際誌

Al-Mahmud, Kazutaka Shimada

Proceedings of the 4th International Conference on Data Science and Applications 2023年07月

Disfluency detection with context information from real utterances and generative utterances 査読有り国際誌

Hiroto Nakashima, Kazutaka Shimada

Proceedings of 13th International Conference on Smart Computing and Artificial Intelligence 2023年07月

Detecting speech recognition errors using topic information and BERT 査読有り国際誌

Masaaki Yokoyama, Kazutaka Shimada

Proceedings of 13th International Conference on Smart Computing and Artificial Intelligence 2023年07月

Combination and Knowledge Extension of Pre-trained Language Model for Offensive Language Detection 査読有り国際誌

Zhenming Li, Kazutaka Shimada

Proceedings of 16th International Conference on E-Service and Knowledge Management 2023年07月

Handling Data Scarcity in Code-Mixing Using Learning-Based Approaches 査読有り国際誌

Niraj Pahari, Kazutaka Shimada

Proceedings of 16th International Conference on E-Service and Knowledge Management 2023年07月

Automated Scoring of Logical Consistency of Japanese Essays 査読有り国際誌

Nakamoto S., Shimada K.

Communications in Computer and Information Science 1831 CCIS 652 - 658 2023年07月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Automated essay scoring (AES) is to estimate the scores of essays automatically. Two types of AES models are commonly used: handcrafted feature-based and neural-based models. In this paper, we introduce AES systems based on the two types for evaluating the logical consistency of Japanese essays. In addition, to enhance the performance of models, we integrate the neural-based model with the handcrafted features: a hybrid AES system. In the experiment, we show the effectiveness of our hybrid AES system. Besides, most of our AES models obtained higher QWK scores than human evaluators.

DOI： 10.1007/978-3-031-36336-8_101

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85164955884&origin=inward

Demonstration of Effectiveness of Nativeness in Stepwise Learning by Performing Sentiment Analysis 査読有り国際誌

Al-Mahmud, Kazutaka Shimada

Proceedings of 2023 International Conference on Next-Generation Computing, IoT and Machine Learning (NCIM) 2023年06月

　前のページ- 次のページ　

<嶋田 和孝>

論文 - 嶋田 和孝

<嶋田　和孝>

論文 - 嶋田　和孝