J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/pjsai.jsai2026.0_4yinb25

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Association for Natural Language Processing  

DOI： 10.5715/jnlp.33.407

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.5715/jnlp.33.308

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

researchmap

Other Link： https://dblp.org/db/conf/iwslt/iwslt2026.html

Publishing type：Research paper (international conference proceedings)  

researchmap

Other Link： https://dblp.org/rec/conf/iwslt/2026

Publishing type：Research paper (international conference proceedings)  

researchmap

Other Link： https://dblp.org/rec/conf/iwslt/2026

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/pjsai.jsai2026.0_3j2os15b01

CiNii Research

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Association for Natural Language Processing  

The ability to accurately interpret implied meanings plays a crucial role in human communication and language use, and language models are also expected to possess this capability. This study demonstrates that providing language models with pragmatic theories as prompts is an effective approach for tasks to understand implied meanings. Specifically, we propose an approach in which an overview of pragmatic theories, such as Gricean pragmatics and Relevance Theory, is presented as a prompt to the language model, guiding it through a step-by-step reasoning process to derive a final interpretation. To validate the effectiveness of our method, we conducted experiments on a benchmark that tests the implied meaning of utterances using four- to five-choice multiple-choice questions.Experimental results showed that, compared to the baseline, which prompts intermediate reasoning without presenting pragmatic theories (0-shot Chain-of-Thought), our methods enabled language models to achieve up to 9.6% higher scores on the benchmark. Furthermore, we show that even without explaining the details of pragmatic theories, merely mentioning their names in the prompt leads to a certain performance improvement (around 1–3%) in larger models compared to the baseline.

DOI： 10.5715/jnlp.33.880

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2511.22490

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Cascade implementations of multiple machine learning models allow individual modules to be trained independently. However, they face the challenge that some information other than the final hypothesis output by the preceding module becomes partially lost at that stage. To address this issue, N-best training methods are generally employed, but they suffer from increased training and inference costs depending on N. This research proposes a framework for speech dialogue state tracking that addresses this issue. By inputting a vector composed of the probability values for each hypothesis output by the ASR model into a quantum machine learning model and simultaneously processing multiple hypotheses, it suppresses the increase in training and inference costs associated with conventional N-best training methods. We applied the proposed method to the DSTC2 dataset for speech dialogue tracking tasks and confirmed that it enables a significant reduction in the number of parameters while maintaining accuracy.

DOI： 10.11517/jsaislud.105.0_58

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I034421214

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In target-guided conversation, it is crucial to improve the user experience by leading the conversation toward the system's own goal without making the user feel guided and without making them aware of the system's objective.In this study, we propose SBIS-TGC (Surprisal-Based Induction Score for Target-Guided Conversation), an automatic evaluation metric designed to assess the degree of induction in system utterances, with the objective of achieving conversation goals without the user noticing either the system's goal or its guiding behavior.SBIS-TGC quantifies the sense of induction by calculating surprisal between utterances using an external language model.Through dialogue experiments employing a system that selects utterances based on SBIS-TGC, we demonstrate that the proposed method can reduce the perceived induction in target-guided dialogue and enable conversations where users remain unaware of the system's intended target.

DOI： 10.11517/jsaislud.105.0_152

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I034426467

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In collaborative problem-solving, particularly in technical domains like mathematics, discussions often combine spoken dialogue with a shared visual space, such as a whiteboard. A critical challenge for comprehending these interactions is resolving the reference between ambiguous expressions in dialogue (e.g., pronouns) and the specific symbols or equations written on the board.To address this, and drawing inspiration from research in Visually-Grounded Dialogue, we propose a new annotation schema for capturing the discourse structure of these multimodal discussions by explicitly linking dialogue utterances to their corresponding element on the whiteboard.

DOI： 10.11517/jsaislud.105.0_150

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

When the same linguistic content carries different acoustic nuances, particularly in terms of expressed emotions, the corresponding dialogue system response must align with the given nuance. However, existing SLMs such as Qwen2-Audio are not necessarily robust against such differences. In this work, we define a task that detects the consistency or inconsistency between the emotional label of an utterance and the system's response, and build a model to perform this prediction. We hypothesize that emotion labels are a control signal that modulates text interpretation, and we construct a prediction model based on Feature-wise Linear Modulation (FiLM).

DOI： 10.11517/jsaislud.105.0_31

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I034420487

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In communication, gestures play an important role in aiding the understanding of speech and conveying social information. In particular, human gestures are consistent with individual personality traits, such as extraversion and introversion, and gestures that consider these personality traits are necessary to send appropriate social signals. Therefore, this research proposes a method for generating co-speech gestures that reflect a robot's personality traits (e.g., extraversion/introversion) by utilizing a large language model (LLM).

DOI： 10.11517/jsaislud.105.0_130

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2510.21761

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2510.26253

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：ISCA  

DOI： 10.21437/interspeech.2025-595

researchmap

Other Link： https://dblp.org/db/conf/interspeech/interspeech2025.html#MoriKLICY25

Publishing type：Research paper (international conference proceedings)   Publisher：ISCA  

DOI： 10.21437/interspeech.2025-600

researchmap

Other Link： https://dblp.org/db/conf/interspeech/interspeech2025.html#MoriKCY25

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2508.04403

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2506.13956

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2505.11726

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/icassp49660.2025.10888775

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2504.07698

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/qcnc64685.2025.00092

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-96-2074-6_2

CiNii Research

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/mmm/mmm2025-5.html#YoshiharaYKIY25

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-96-2074-6_21

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/mmm/mmm2025-5.html#KawanishiNSIKYMM25

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-96-2074-6_28

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/mmm/mmm2025-5.html#ContrerasCKCY25

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/2025.acl-long.1526

researchmap

Other Link： https://dblp.org/rec/conf/acl/2025-1

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/2025.acl-long.547

researchmap

Other Link： https://dblp.org/rec/conf/acl/2025-1

Publishing type：Research paper (scientific journal)   Publisher：Association for Natural Language Processing  

DOI： 10.5715/jnlp.32.598

researchmap

Publishing type：Research paper (scientific journal)   Publisher：Institute of Electrical and Electronics Engineers (IEEE)  

DOI： 10.1109/access.2025.3563982

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Association for Natural Language Processing  

Situated conversations, which refer to visual information as visual question answering (VQA), often contain ambiguities caused by reliance on directive information. Word ellipsis exacerbates the problem on some languages such as Japanese, which happens on central contents of the questions such as subjects or objects. Contexts in conversational situations, such as joint attention between a user and a system, or user gaze can clarify such ambiguities in questions. In this study, we propose the Gaze-grounded VQA dataset (LookVQA), which corresponds to directive or ellipsis and gaze information by focusing on a clarification process complemented by gaze information. We also propose a question answering model that utilizes gaze target estimation results to improve the accuracy of LookVQA tasks. Our experimental results showed that our model improved the performance in some question types of LookVQA and outperformed a VQA baseline.

DOI： 10.5715/jnlp.32.3

CiNii Research

researchmap

Publishing type：Research paper (international conference proceedings)  

researchmap

Other Link： https://dblp.org/rec/conf/sigdial/2025

Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/IROS60139.2025.11246615

researchmap

Other Link： https://dblp.org/db/conf/iros/iros2025.html#AtuhurraKWY25

Publishing type：Research paper (international conference proceedings)  

DOI： 10.18653/v1/2025.findings-emnlp.131

researchmap

Other Link： https://dblp.org/rec/conf/emnlp/2025f

Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/ASRU65441.2025.11434793

researchmap

Other Link： https://dblp.org/db/conf/asru/asru2025.html#KomatsuOTKY25

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In recent years, advancements in bioinformatics have significantly improved epitope prediction. Notably, the IEDB database enables the prediction of both T-cell epitopes and linear B-cell epitopes with a certain level of accuracy using only antigen sequence data. Furthermore, tools for evaluating the toxicity and allergenicity of these epitopes have also been well-developed. Leveraging these tools, multi-epitope vaccines, which align multiple epitopes in sequence, have been developed for various antigens and have demonstrated effectiveness. In the case of influenza, numerous in silico vaccine designs have been published, and some studies have shown superior efficacy compared to the quadrivalent influenza vaccine (QIV) in mouse infection experiments. This suggests that multi-epitope design is a viable approach for effective antigen design in influenza. In this study, we focused on H5N1 influenza and performed antigen design using the in silico tools.

DOI： 10.11517/pjsai.jsai2025.0_3h6os10d04

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (international conference proceedings)  

researchmap

Other Link： https://dblp.org/rec/conf/sigdial/2025

Language：English   Publishing type：Research paper (international conference proceedings)  

Non-verbal behaviors, such as body movements, play a crucial role in enhancing a robot’s speech to elevate engagement in human-robot dialogue. Many existing approach based on rules offered natural and engaging motions aligned with the robot’s utterances but required significant resources to maintain. Recent methods leveraging large language models (LLMs) offer a promising alternative to reduce these costs. However, there is a trade-off between flexibility and safety when determining whether the language model should generate motions based on joint angle parameters or action primitives. In this study, we evaluated two LLM-based motion control models: one for motion generation based on joint angle parameters (LLM-GJA) and the other for motion generation based on primitive actions (LLM-GPA). Our human evaluations indicated that directly generating joint angles outperformed generating action primitives in naturalness, timing consistency, and overall engagement, even achieving performance comparable to rule-based systems. This work highlights the potential of LLMs in generating expressive and contextually appropriate robot motions at the joint angle level.

DOI： 10.1109/RO-MAN63969.2025.11217871

CiNii Research

researchmap

Other Link： https://dblp.org/db/conf/ro-man/ro-man2025.html#BaihaqiCKY25

Publishing type：Research paper (international conference proceedings)  

DOI： 10.21437/Interspeech.2025-1117

researchmap

Other Link： https://dblp.org/db/conf/interspeech/interspeech2025.html#BaihaqiCKY25

Publishing type：Research paper (international conference proceedings)  

researchmap

Other Link： https://dblp.org/rec/conf/interspeech/2025

Publishing type：Research paper (international conference proceedings)  

DOI： 10.18653/v1/2025.ijcnlp-long.132

researchmap

Other Link： https://dblp.org/rec/conf/ijcnlp/2025-l

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_121

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830173

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_53

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830133

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_177

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830191

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_96

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830161

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_109

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830167

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_34

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830110

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_152

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.102.0_59

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033830138

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/smc54092.2024.10831793

researchmap

Language：Japanese  

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033522694

Language：Japanese  

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033522699

Language：Japanese   Publisher：東京 : 日本ロボット学会 ; 1983-  

CiNii Books

CiNii Research

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033494719

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Patent retrieval refers to the process of searching within patent databases for information on technologies, inventions, inventors, and applicants. Particularly, since recognized patent infringement in court could result in substantial damages or licensing fees, conducting thorough prior art searches is crucial. However, patent documents are composed of unique vocabularies and the number of documents is vast, making the research process costly. While there are several methods aimed at conducting exhaustive searches by expanding search queries, they generally struggle to address complex vocabularies present in only a small number of patents.Therefore, this study proposes query expansion combining thesauruses and large language models (LLMs). It focuses on the output tendencies of LLMs and the independence and co-occurrence rates of new words generated by existing thesauruses and LLMs, conducting a foundational analysis of the method. As a result, new words generated by large language models had low co-occurrence with existing thesauruses. The success in generating new vocabularies through large language models suggests the potential for comprehensive patent searches that can accommodate the unique vocabularies and complex expressions of patent documents.

DOI： 10.11517/pjsai.jsai2024.0_4a1gs605

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2406.09839

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/BigData62323.2024.10825172

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/bigdataconf/bigdataconf2024.html#KawanoNY24

Publishing type：Research paper (scientific journal)  

DOI： 10.1109/ACCESS.2024.3350174

Scopus

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：ELRA and ICCL  

CiNii Research

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/coling/2024

Publishing type：Research paper (international conference proceedings)   Publisher：ELRA and ICCL  

DOI： 10.48550/arXiv.2403.19259

CiNii Research

J-GLOBAL

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/coling/2024

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2403.17545

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Association for Natural Language Processing  

Understanding the situation in the physical world is crucial for robots assisting humans in the real world. Especially, when a robot is required to collaborate with humans through verbal interactions, such as dialogues, the verbal information that appears in user interactions must be grounded in the visual information observed in egocentric views. To this end, we proposed a multimodal reference resolution task and constructed a Japanese Conversation dataset for Real-world Reference Resolution (J-CRe3). Our dataset contains egocentric videos and dialogue audio from real-world conversations between two people acting as a master and an assistant robot at home. It is annotated with crossmodal tags between phrases in the utterances and object bounding boxes in the video frames. These tags include indirect reference relations, such as predicate-argument structures and bridging references as well as direct reference relations. We also constructed an experimental model that combined an existing textual reference resolution model with a phrase grounding model. Our experiments with this model showed that crossmodal reference resolution is significantly more challenging than textual reference resolution in the proposed task.

DOI： 10.5715/jnlp.31.1107

CiNii Research

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Dialog systems need not only to generate natural responses in the dialogue context but also to transit the topic naturally, which the user is more interested in. In this study, we propose the method for constructing a topic transition graph generated from word association based on large language model, which can naturally transit the dialogue topic according to the user's preference. We evaluate the topic transition graphs generated from several methods for investigating better topic transition of dialogue systems.

DOI： 10.11517/pjsai.jsai2024.0_4xin294

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Large language models (LLMs) are flexible and can handle various natural language processing tasks. Many spoken dialogue systems are realized by linking a dialogue model built using an LLM with other modules, such as speech recognition or synthesis systems. However, such a cascaded model with multiple modules is complicated and tends to propagate errors from the previous module. The model can also not consider sensitive expressions in the non-verbal representation of dialogue because the discrete representation, such as texts, is used to connect modules. This research aims to solve these problems by converting the input speech into a vector of continuous expressions and connecting it to a dialogue model. The experimental results show that the generated sentences do not fully take the dialogue context into account, and there is room for improvement, but the natural sentence generation is learned, suggesting that a dialogue model using continuous expressions is feasible.

DOI： 10.11517/pjsai.jsai2024.0_1g3gs604

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In this study, we developed a dataset for the development of image-language models of text-drawing pairs in patent documents. Specifically, we created a large image-language dataset by mapping patent drawings to explanatory text using standardized expressions in patents.

DOI： 10.11517/pjsai.jsai2024.0_3xin2103

CiNii Research

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2407.03963

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：ISCA  

DOI： 10.21437/Interspeech.2024-1406

researchmap

Other Link： https://dblp.org/db/conf/interspeech/interspeech2024.html#BaihaqiCKY24

Publishing type：Research paper (scientific journal)  

DOI： 10.1109/TASLP.2023.3293030

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.5715/jnlp.31.1107

researchmap

Publishing type：Research paper (scientific journal)  

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.48550/arXiv.2410.05575

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Association for Natural Language Processing  

DOI： 10.5715/jnlp.31.809

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.3389/frobt.2023.949600

Web of Science

researchmap

Publishing type：Research paper (scientific journal)   Publisher：Elsevier BV  

DOI： 10.1016/j.csl.2022.101463

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Association for Natural Language Processing  

DOI： 10.5715/jnlp.30.1278

CiNii Research

researchmap

Language：Japanese   Publisher：The Robotics Society of Japan  

Analyzing time-series changes of the contact surface by touching is important to elucidate the touching skills of Humanitude as one of the pervasive multimodal comprehensive care methodologies. However, the transparent flat plates to capture contact surfaces used in the previous studies are far from the actual surfaces of care receivers because the surfaces of humans consist of curved shapes. In this paper, we propose a Frustrated Total Internal Reflection (FTIR) sensing system using a transparent curved plate for the purpose to capture contacting state by using a surface more similar to the human body. We evaluate the equivalent performance between data from the proposed FTIR sensing system and an FTIR sensing system using a transparent flat plate by calculating time-series contact areas and the quantitative indices representing theHumanitude's touching motion. Through these experiments, we conclude that the proposed system can obtain the same data quality for the contact areas and the indices as the FTIR system using the flat plate.

DOI： 10.7210/jrsj.41.893

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/2023.findings-emnlp.531

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/emnlp/2023f

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/ICRA48891.2023.10161545

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/icra/icra2023.html#NakamuraKYKY23

Publishing type：Research paper (scientific journal)  

Narratives include a rich source of events unfolding over time and context.
Automatic understanding of these events provides a summarised comprehension of
the narrative for further computation (such as reasoning). In this paper, we
study the Information Status (IS) of the events and propose a novel challenging
task: the automatic identification of new events in a narrative. We define an
event as a triplet of subject, predicate, and object. The event is categorized
as new with respect to the discourse context and whether it can be inferred
through commonsense reasoning. We annotated a publicly available corpus of
narratives with the new events at sentence level using human annotators. We
present the annotation protocol and study the quality of the annotation and the
difficulty of the task. We publish the annotated dataset, annotation materials,
and machine learning baseline models for the task of new event extraction for
narrative understanding.

DOI： 10.48550/arXiv.2302.07748

arXiv

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：Acoustical Society of Japan  

DOI： 10.20697/jasj.79.9_477

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I033041574

Language：Japanese  

DOI： 10.11517/jsaislud.96.0_20

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I032570587

Language：Japanese  

DOI： 10.11517/jsaislud.96.0_13

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I032570566

Language：Japanese  

DOI： 10.11517/jsaislud.96.0_08

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I032570551

Language：Japanese   Publishing type：Research paper (scientific journal)  

DOI： 10.11517/jjsai.37.4_490

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I032266799

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.94.0_17

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.94.0_16

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I032014069

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Conversational robots with human-like appearances, such as android robots, are expected to appropriately combine verbal and nonverbal expressions to realize intelligent and natural behavior in communication. One of the most important perspectives for the robots is generating appropriate nonverbal motions based on the intent expressed in their utterance. In this paper, we implement nonverbal motions (facial expressions) using interest levels to the current dialogue topics and then evaluate how the robot's intentions were conveyed to subjects. Specifically, we analyze a multi-modal dialogue corpus, which is annotated with levels of interest, to implement the motions of the facial area and then implement the extracted typical behavior on the android robot. We evaluate the implemented behavior by a subjective evaluation with a scenario-based conversation to investigate the intention conveying performance.

DOI： 10.11517/pjsai.jsai2022.0_3yin233

CiNii Research

researchmap

Publishing type：Research paper (scientific journal)  

Such human-assisting systems as robots need to correctly understand the
surrounding situation based on observations and output the required support
actions for humans. Language is one of the important channels to communicate
with humans, and the robots are required to have the ability to express their
understanding and action planning results. In this study, we propose a new task
of operative action captioning that estimates and verbalizes the actions to be
taken by the system in a human-assisting domain. We constructed a system that
outputs a verbal description of a possible operative action that changes the
current state to the given target state. We collected a dataset consisting of
two images as observations, which express the current state and the state
changed by actions, and a caption that describes the actions that change the
current state to the target state, by crowdsourcing in daily life situations.
Then we constructed a system that estimates operative action by a caption.
Since the operative action's caption is expected to contain some state-changing
actions, we use scene-graph prediction as an auxiliary task because the events
written in the scene graphs correspond to the state changes. Experimental
results showed that our system successfully described the operative actions
that should be conducted between the current and target states. The auxiliary
tasks that predict the scene graphs improved the quality of the estimation
results.

DOI： 10.48550/arXiv.2210.02735

arXiv

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：ISCA  

DOI： 10.21437/Interspeech.2022-565

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/interspeech/interspeech2022.html#KawanoAYYIK0Y22

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE / ACM  

DOI： 10.1109/HRI53351.2022.9889605

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/hri/2022

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/2022.dialdoc-1.4

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/acl-dialdoc/2022

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Entrainment is a phenomenon observed in human-human conversation, which is a synchronization of speaking style according to dialogue progress. In this study, we propose a method to build an entrainable chitchat system by predicting the ideal entrainment score for the given dialogue history. The proposed method reranks existing neural conversation model outputs based on the predicted entrainment score. We conducted automatic and human-subjective evaluations to investigate the effect of the proposed method by comparing it with the system response without using the reranking system. The experimental results showed that our proposed method achieves ideal entrainment while maintaining the naturalness of the generated responses compared to the baseline method.

DOI： 10.11517/pjsai.jsai2022.0_3yin248

CiNii Research

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Robots working in a symbiotic space with a user must be trusted and smoothly generate cooperative behavior. The ability to express a robot's observation, memorization, and intention in a human-understandable manner is essential for trust. Description by language is an effective way to convey various information to users. In this study, we studied the robot's describing ability by comparing different description rules in a user guide task. Specifically, we used an autonomous mobile interactive robot platform to describe its observation and internal states for comparison and evaluated them by human subjective evaluation using crowdsourcing.

DOI： 10.11517/pjsai.jsai2022.0_2n5os7a03

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Association for Natural Language Processing  

DOI： 10.5715/jnlp.29.718

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.93.0_50

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：Japanese  

DOI： 10.20729/00213185

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： http://id.nii.ac.jp/1001/00213185/

Language：Japanese   Publisher：Acoustical Society of Japan  

DOI： 10.20697/jasj.77.8_521

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I031643136

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

Building a controllable neural conversation model (NCM) is an important task. In this paper, we focus on controlling the responses of NCMs using dialogue act labels of responses as conditions. We introduce a reinforcement learning framework involving adversarial learning for conditional response generation. Our proposed method has a new label-aware objective that encourages the generation of discriminative responses by the given dialogue act label while maintaining the naturalness of the generated responses. We compared the proposed method with conventional methods that generate conditional responses. The experimental results showed that our proposed method has higher controllability conditioned by the dialogue acts even though it has higher or comparable naturalness to the conventional models.

DOI： 10.1527/tjsai.36-4_e-kc9

CiNii Research

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

Human-assisting systems such as dialogue systems must take thoughtful,
appropriate actions not only for clear and unambiguous user requests, but also
for ambiguous user requests, even if the users themselves are not aware of
their potential requirements. To construct such a dialogue agent, we collected
a corpus and developed a model that classifies ambiguous user requests into
corresponding system actions. In order to collect a high-quality corpus, we
asked workers to input antecedent user requests whose pre-defined actions could
be regarded as thoughtful. Although multiple actions could be identified as
thoughtful for a single user request, annotating all combinations of user
requests and system actions is impractical. For this reason, we fully annotated
only the test data and left the annotation of the training data incomplete. In
order to train the classification model on such training data, we applied the
positive/unlabeled (PU) learning method, which assumes that only a part of the
data is labeled with positive examples. The experimental results show that the
PU learning method achieved better performance than the general
positive/negative (PN) learning method to classify thoughtful actions given an
ambiguous user request.

DOI： 10.18653/v1/2021.sigdial-1.9

arXiv

researchmap

Other Link： http://arxiv.org/pdf/2106.07999v1

Publishing type：Research paper (scientific journal)  

researchmap

Other Link： https://dblp.uni-trier.de/db/journals/corr/corr2106.html#abs-2106-07999

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-19-5538-9_10

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/iwsds/iwsds2021.html#AsaiYSS021

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Association for Natural Language Processing  

Dialogue continuity is a dialogue system-evaluation metric that indicates how well the system engages its users. Response coherence considered an important factor in determining dialogue continuity. Herein, we propose novel methods of selecting coherent responses for a given dialogue context. These methods improve coherence and dialogue continuity using related-event pairs such as “be stressed out” and “relieve stress.” Two re-ranking methods are proposed. The first method estimates the coherence of event pairs in a dialogue by matching event-causality pairs that are statistically extracted from web texts. The second method estimates the coherence of response candidates for the dialogue context using a coherence model. The results of an automatic evaluation show that objective coherence at the word level was improved by these re-ranking methods. In contrast, in the case of human evaluation, subjective coherence at the response level was not improved by the first method; however, the dialogue continuity was improved. These results seem to be contradictory. We conducted a correlation analysis and a case analysis to clarify the relationship between subjective coherence and dialogue continuity. The results indicate that subjective coherence does not have a strong correlation with dialogue continuity. Further, dialogue continuity may be improved if an event related to the dialogue context is selected.

DOI： 10.5715/jnlp.28.26

CiNii Research

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Reusing policies in a new domain, which is trained on the existing domain, is an important problem of dialogue management research based on reinforcement learning. This work defines action-relation probabilities between the action spaces of the new and the target domains using mixture density networks for the reuse of policies. Experimental results showed that the proposed modeling of action-relation probabilities based on component matching using regression realized the effective policy reuse.

DOI： 10.11517/pjsai.jsai2021.0_4e1os11a01

CiNii Research

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：一般社団法人 人工知能学会  

DOI： 10.11517/jsaislud.90.0_07

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I030789230

Publishing type：Research paper (scientific journal)   Publisher：Association for Computational Linguistics  

Word embeddings, which often represent such analogic relations as king - man
+ woman = queen, can be used to change a word's attribute, including its
gender. For transferring king into queen in this analogy-based manner, we
subtract a difference vector man - woman based on the knowledge that king is
male. However, developing such knowledge is very costly for words and
attributes. In this work, we propose a novel method for word attribute transfer
based on reflection mappings without such an analogy operation. Experimental
results show that our proposed method can transfer the word attributes of the
given words without changing the words that do not have the target attributes.

DOI： 10.18653/v1/2020.acl-srw.8

arXiv

researchmap

Other Link： http://arxiv.org/pdf/2007.02598v2

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

Bridging robot action sequences and their natural language captions is an
important task to increase explainability of human assisting robots in their
recently evolving field. In this paper, we propose a system for generating
natural language captions that describe behaviors of human assisting robots.
The system describes robot actions by using robot observations; histories from
actuator systems and cameras, toward end-to-end bridging between robot actions
and natural language captions. Two reasons make it challenging to apply
existing sequence-to-sequence models to this mapping: 1) it is hard to prepare
a large-scale dataset for any kind of robots and their environment, and 2)
there is a gap between the number of samples obtained from robot action
observations and generated word sequences of captions. We introduced
unsupervised segmentation based on K-means clustering to unify typical robot
observation patterns into a class. This method makes it possible for the
network to learn the relationship from a small amount of data. Moreover, we
utilized a chunking method based on byte-pair encoding (BPE) to fill in the gap
between the number of samples of robot action observations and words in a
caption. We also applied an attention mechanism to the segmentation task.
Experimental results show that the proposed model based on unsupervised
learning can generate better descriptions than other methods. We also show that
the attention mechanism did not work well in our low-resource setting.

DOI： 10.1007/978-981-15-8395-7_17

arXiv

researchmap

Other Link： http://arxiv.org/pdf/2003.10066v1

Publishing type：Research paper (scientific journal)  

DOI： 10.1109/ACCESS.2020.3027099

CiNii Research

researchmap

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：International Committee on Computational Linguistics  

DOI： 10.18653/v1/2020.coling-main.234

researchmap

Other Link： https://dblp.uni-trier.de/conf/coling/2020

Publishing type：Research paper (international conference proceedings)   Publisher：European Language Resources Association  

researchmap

Other Link： https://dblp.uni-trier.de/conf/lrec/2020

Publishing type：Research paper (scientific journal)  

researchmap

Other Link： https://dblp.uni-trier.de/db/journals/corr/corr2003.html#abs-2003-10066

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

Word embeddings, which often represent analogic relations such as king − man + woman ≈ queen, can be used to change an attribute of a word, including its gender. To transfer the gender attribute of king to obtain queen in this analogy, we subtract a difference vector man − woman from king based on the knowledge that king is male. However, developing such knowledge is significantly costly for words and attributes. In this work, we propose a novel method for word attribute transfer based on reflection mapping without an analogy-based operation. Experimental results show that our proposed method can transfer the word attributes of the given words without changing the words that are invariant with respect to the target attributes.

DOI： 10.18653/v1/2020.acl-srw.8

CiNii Research

researchmap

Other Link： https://dblp.uni-trier.de/db/journals/corr/corr2007.html#abs-2007-02598

Publishing type：Research paper (scientific journal)  

DOI： 10.1527/tjsai.DSI-C

Scopus

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.1109/ACCESS.2020.2997012

CiNii Research

researchmap

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.1109/ACCESS.2020.3017780

CiNii Research

researchmap

Authorship：Lead author   Publishing type：Research paper (scientific journal)  

DOI： 10.1016/j.csl.2020.101068

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Expressing emotion is known as an efficient way to persuade one's dialogue partner to accept one's claim or proposal. In this paper, we built a spoken dialogue robot in a persuasive scenario that uses emotional expressions. We collected a speech dialogue corpus in persuasion and then built a spoken dialogue robot based on the corpus. Experimental results showed the system based on emotional expressions performed better than the general system based on neutral emotion and worked well by using several modalities including robot gestures.

DOI： 10.11517/pjsai.jsai2020.0_4rin171

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Language generation is a task to generate sentences corresponding to a given intention. Existing researches of neural language generation models gave the intention to encoder; however, there was no strong constraint to contain the intention in generation results. In this research, we propose a learning method for the language generation system to ensure the generated sentence contains the given intention. We evaluated the effect of the proposed method in both automatic and human subjective evaluation.

DOI： 10.11517/pjsai.jsai2020.0_3rin494

CiNii Research

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.87.0_15

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I030099511

Language：Japanese   Publisher：一般社団法人 人工知能学会  

DOI： 10.11517/jsaislud.87.0_23

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.87.0_07

CiNii Books

CiNii Research

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I030099309

Language：Japanese   Publisher：一般社団法人 人工知能学会  

DOI： 10.11517/jsaislud.87.0_17

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I030099557

Language：Japanese   Publisher：東京 : 日本工業出版  

CiNii Books

CiNii Research

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I030084856

Publishing type：Research paper (scientific journal)  

We propose a novel method for selecting coherent and diverse responses for a
given dialogue context. The proposed method re-ranks response candidates
generated from conversational models by using event causality relations between
events in a dialogue history and response candidates (e.g., ``be stressed out''
precedes ``relieve stress''). We use distributed event representation based on
the Role Factored Tensor Model for a robust matching of event causality
relations due to limited event causality knowledge of the system. Experimental
results showed that the proposed method improved coherency and dialogue
continuity of system responses.

arXiv

researchmap

Other Link： http://arxiv.org/pdf/1906.09795v1

Publishing type：Research paper (international conference proceedings)   Publisher：ISCA  

In a human-machine dialog scenario, deciding the appropriate time for the
machine to take the turn is an open research problem. In contrast, humans
engaged in conversations are able to timely decide when to interrupt the
speaker for competitive or non-competitive reasons. In state-of-the-art
turn-by-turn dialog systems the decision on the next dialog action is taken at
the end of the utterance. In this paper, we propose a token-by-token prediction
of the dialog state from incremental transcriptions of the user utterance. To
identify the point of maximal understanding in an ongoing utterance, we a)
implement an incremental Dialog State Tracker which is updated on a token basis
(iDST) b) re-label the Dialog State Tracking Challenge 2 (DSTC2) dataset and c)
adapt it to the incremental turn-taking experimental scenario. The re-labeling
consists of assigning a binary value to each token in the user utterance that
allows to identify the appropriate point for taking the turn. Finally, we
implement an incremental Turn Taking Decider (iTTD) that is trained on these
new labels for the turn-taking decision. We show that the proposed model can
achieve a better performance compared to a deterministic handcrafted
turn-taking algorithm.

DOI： 10.21437/Interspeech.2019-1826

arXiv

researchmap

Other Link： http://arxiv.org/pdf/1905.11806v3

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.85.0_07

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：東京 : 人工知能学会  

CiNii Books

CiNii Research

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I029547283

Authorship：Lead author   Publishing type：Research paper (scientific journal)  

This paper introduces the Seventh Dialog System Technology Challenges (DSTC),
which use shared datasets to explore the problem of building dialog systems.
Recently, end-to-end dialog modeling approaches have been applied to various
dialog tasks. The seventh DSTC (DSTC7) focuses on developing technologies
related to end-to-end dialog systems for (1) sentence selection, (2) sentence
generation and (3) audio visual scene aware dialog. This paper summarizes the
overall setup and results of DSTC7, including detailed descriptions of the
different tracks and provided datasets. We also describe overall trends in the
submitted systems and the key results. Each track introduced new datasets and
participants achieved impressive results using state-of-the-art end-to-end
technologies.

arXiv

researchmap

Other Link： http://arxiv.org/pdf/1901.03461v1

Publishing type：Research paper (scientific journal)  

DOI： 10.1109/TASLP.2019.2900910

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：ACM  

DOI： 10.1145/3351529.3360657

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/icmi/icmi2019a.html#KinoshitaTY019

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/W19-8627

researchmap

Other Link： https://dblp.uni-trier.de/conf/inlg/2019

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-15-9323-9_12

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/iwsds/iwsds2019.html#YoshinoMLSTST019

Publishing type：Research paper (scientific journal)  

researchmap

Other Link： https://dblp.uni-trier.de/db/journals/corr/corr1905.html#abs-1905-11806

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

researchmap

Other Link： https://aclanthology.org/W19-5900/

Publishing type：Research paper (scientific journal)  

DOI： 10.1609/aimag.v40i3.4981

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.1016/j.csl.2018.08.003

researchmap

Publishing type：Research paper (scientific journal)  

DOI： 10.1016/j.csl.2018.09.004

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

Empathy has been studied intensively in the field of human-machine interaction and human-human interaction.In many of these studies, empathy is subjectively measured by questionnaire. We propose a method to directly measure human empathy from neural potential activity. In this research, EEG signals were recorded simultaneously from two subjects engaged in emotional communication. After time frequency analysis was performed on the recorded Electroencephalogram (EEG) signals, a cross correlation functions between EEG signals of two subjects were calculated.To investigate the significance of cross correlations, paired t-test and false discovery rate correction were performed.As a result, there was a significant cross correlation in the gamma frequency band in the joy condition.In addition, there was a marginally significant correlation in the sadness condition.From these results we showed the possibility to measure empathy from cross correlation of EEG signals recorded from two subjects.

DOI： 10.11517/pjsai.jsai2019.0_4f2os11a03

CiNii Research

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Japanese Society for Artificial Intelligence  

Argumentation-based dialogue systems, which can handle and exchange arguments
through dialogue, have been widely researched. It is required that these
systems have sufficient supporting information to argue their claims
rationally; however, the systems often do not have enough of such information
in realistic situations. One way to fill in the gap is acquiring such missing
information from dialogue partners (information-seeking dialogue). Existing
information-seeking dialogue systems are based on handcrafted dialogue
strategies that exhaustively examine missing information. However, the proposed
strategies are not specialized in collecting information for constructing
rational arguments. Moreover, the number of system's inquiry candidates grows
in accordance with the size of the argument set that the system deal with. In
this paper, we formalize the process of information-seeking dialogue as Markov
decision processes (MDPs) and apply deep reinforcement learning (DRL) for
automatically optimizing a dialogue strategy. By utilizing DRL, our dialogue
strategy can successfully minimize objective functions, the number of turns it
takes for our system to collect necessary information in a dialogue. We
conducted dialogue experiments using two datasets from different domains of
argumentative dialogue. Experimental results show that the proposed
formalization based on MDP works well, and the policy optimized by DRL
outperformed existing heuristic dialogue strategies.

DOI： 10.1527/tjsai.DSI-D

CiNii Research

arXiv

researchmap

Other Link： http://arxiv.org/pdf/1811.10728v1

Publishing type：Research paper (scientific journal)  

Although generation-based dialogue systems have been widely researched, the
response generations by most existing systems have very low diversities. The
most likely reason for this problem is Maximum Likelihood Estimation (MLE) with
Softmax Cross-Entropy (SCE) loss. MLE trains models to generate the most
frequent responses from enormous generation candidates, although in actual
dialogues there are various responses based on the context. In this paper, we
propose a new objective function called Inverse Token Frequency (ITF) loss,
which individually scales smaller loss for frequent token classes and larger
loss for rare token classes. This function encourages the model to generate
rare tokens rather than frequent tokens. It does not complicate the model and
its training is stable because we only replace the objective function. On the
OpenSubtitles dialogue dataset, our loss model establishes a state-of-the-art
DIST-1 of 7.56, which is the unigram diversity score, while maintaining a good
BLEU-1 score. On a Japanese Twitter replies dataset, our loss model achieves a
DIST-1 score comparable to the ground truth.

arXiv

researchmap

Other Link： http://arxiv.org/pdf/1811.08100v2

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.84.0_25

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.84.0_27

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.11517/jsaislud.84.0_15

CiNii Research

J-GLOBAL

researchmap

Language：Japanese  

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：English   Publisher：Information Processing Society of Japan  

<p>Protein production in plants is a hot topic because there are many benefits relative to bacteria, yeasts, and animals, but the amount of protein expression in plants is less. It is argued that editing 5'UTRs increases the amount of translated proteins. However, obtaining such 5'UTRs is difficult due to the cost, time and effort required in experiments. To solve this, we predict the amount of translated proteins by machine learning. In this paper, we propose a method, named "R-STEINER, " that generates 5'UTRs that increase the amount of proteins of a given gene. The proposed process involves building a model for predicting the amount of translated proteins, generating 5'UTRs, selecting them and increasing the proteins according to the model. This method enables us to obtain 5'UTRs that increase the amount of translated proteins without real synthesis experiments, resulting in reduced cost, time and effort. In our study, we built a prediction model for Oryza sativa and synthesized the 5'UTRs generated by R-STEINER. We confirmed that the model can predict the amount of translated proteins with a correlation coefficient of 0.89.</p>

DOI： 10.2197/ipsjtbio.11.14

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Association for Natural Language Processing  

<p>In this paper, we propose a new method for calculating the output layer in neural machine translation systems with largely reduced computation cost based on binary code. The method is performed by predicting a bit array instead of actual output symbols to obtain word probabilities, and can reduce computation time/memory requirements of the output layer to be logarithmic in vocabulary size in the best case. In addition, since learning proposed model is more difficult than softmax models, we also introduce two approaches to improve translation quality of the proposed model: combining softmax and our models and using error-correcting codes. Experiments on English-Japanese bidirectional translation tasks show proposed models achieve that their BLEU approach the softmax, while reducing memory usage on the order of one tenths, and also improving decoding speed on CPUs by x5 to x10. </p>

DOI： 10.5715/jnlp.25.167

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-13-9443-0_17

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/iwsds/iwsds2018.html#NguyenYS018

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

DOI： 10.1527/tjsai.DSH-D

Scopus

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-3-319-98539-8_17

researchmap

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：European Language Resources Association (ELRA)  

researchmap

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：European Language Resources Association (ELRA)  

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/w18-5017

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/sigdial/sigdial2018.html#conf/sigdial/LubisSYN18

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：The Institute of Electronics, Information and Communication Engineers  

We propose an interactive image-manipulation system with natural language
instruction, which can generate a target image from a source image and an
instruction that describes the difference between the source and the target
image. The system makes it possible to modify a generated image interactively
and make natural language conditioned image generation more controllable. We
construct a neural network that handles image vectors in latent space to
transform the source vector to the target vector by using the vector of
instruction. The experimental results indicate that the proposed framework
successfully generates the target image by using a source image and an
instruction on manipulation in our dataset.

DOI： 10.14923/transinfj.2018iut0001

CiNii Research

arXiv

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

DOI： 10.1527/tjsai.DSH-B

Scopus

CiNii Research

J-GLOBAL

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.1109/SLT.2018.8639613

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/slt/slt2018.html#LubisSY018

Authorship：Lead author   Publishing type：Research paper (scientific journal)  

researchmap

Other Link： https://dblp.uni-trier.de/db/journals/shonan-reports/shonan-reports2018.html#YoshinoMA18

Language：English   Publishing type：Research paper (scientific journal)   Publisher：The Institute of Electronics, Information and Communication Engineers  

The problem with distributed representations generated by neural networks is that the meaning of the features is difficult to understand. We propose a new method that gives a specific meaning to each node of a hidden layer by introducing a manually created word semantic vector dictionary into the initial weights and by using paragraph vector models. We conducted experiments to test the hypotheses using a single domain benchmark for Japanese Twitter sentiment analysis and then evaluated the expandability of the method using a diverse and large-scale benchmark. Moreover, we tested the domain-independence of the method using a Wikipedia corpus. Our experimental results demonstrated that the learned vector is better than the performance of the existing paragraph vector in the evaluation of the Twitter sentiment analysis task using the single domain benchmark. Also, we determined the readability of document embeddings, which means distributed representations of documents, in a user test. The definition of readability in this paper is that people can understand the meaning of large weighted features of distributed representations. A total of 52.4% of the top five weighted hidden nodes were related to tweets where one of the paragraph vector models learned the document embeddings. For the expandability evaluation of the method, we improved the dictionary based on the results of the hypothesis test and examined the relationship of the readability of learned word vectors and the task accuracy of Twitter sentiment analysis using the diverse and large-scale benchmark. We also conducted a word similarity task using the Wikipedia corpus to test the domain-independence of the method. We found the expandability results of the method are better than or comparable to the performance of the paragraph vector. Also, the objective and subjective evaluation support each hidden node maintaining a specific meaning. Thus, the proposed method succeeded in improving readability.

DOI： 10.1587/transinf.2017DAP0019

CiNii Research

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：AAAI Press  

DOI： 10.1609/aaai.v32i1.11955

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/aaai/aaai2018.html#conf/aaai/LubisSYN18

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-13-9443-0_2

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：IEEE  

DOI： 10.23919/APSIPA.2018.8659614

researchmap

Other Link： https://dblp.uni-trier.de/db/conf/apsipa/apsipa2018.html#HosomiSY018

J-GLOBAL

researchmap

Language：Japanese   Publisher：一般社団法人 人工知能学会  

DOI： 10.11517/jsaislud.81.0_33

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：ACM  

DOI： 10.1145/3106426.3106521

Scopus

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/webi/webi2017.html#conf/webi/KeshiSYN17

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Institute of Electrical and Electronics Engineers Inc.  

DOI： 10.1109/ICSDA.2016.7918991

Scopus

CiNii Research

researchmap

Language：Japanese   Publishing type：Research paper (bulletin of university, research institution)  

CiNii Books

CiNii Research

researchmap

Other Link： http://hdl.handle.net/10191/49565

Authorship：Lead author   Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/w17-5542

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/sigdial/sigdial2017.html#conf/sigdial/YoshinoSN17

Publishing type：Research paper (international conference proceedings)   Publisher：International Workshop on Spoken Language Translation  

researchmap

Other Link： https://dblp.uni-trier.de/rec/conf/iwslt/2017

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/P17-1079

Scopus

arXiv

researchmap

Authorship：Lead author   Publishing type：Research paper (scientific journal)  

This paper investigates and analyzes the effect of dependency information on
predicate-argument structure analysis (PASA) and zero anaphora resolution (ZAR)
for Japanese, and shows that a straightforward approach of PASA and ZAR works
effectively even if dependency information was not available. We constructed an
analyzer that directly predicts relationships of predicates and arguments with
their semantic roles from a POS-tagged corpus. The features of the system are
designed to compensate for the absence of syntactic information by using
features used in dependency parsing as a reference. We also constructed
analyzers that use the oracle dependency and the real dependency parsing
results, and compared with the system that does not use any syntactic
information to verify that the improvement provided by dependencies is not
crucial.

arXiv

researchmap

Publishing type：Research paper (scientific journal)   Publisher：Association for Computational Linguistics  

DOI： 10.18653/v1/w17-3208

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：IEEE Computer Society  

DOI： 10.1109/ACII.2017.8273582

Scopus

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics (ACL)  

DOI： 10.18653/v1/P17-1079

Scopus

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Association for Computational Linguistics  

Training of neural machine translation (NMT) models usually uses mini-batches
for efficiency purposes. During the mini-batched training process, it is
necessary to pad shorter sentences in a mini-batch to be equal in length to the
longest sentence therein for efficient computation. Previous work has noted
that sorting the corpus based on the sentence length before making mini-batches
reduces the amount of padding and increases the processing speed. However,
despite the fact that mini-batch creation is an essential step in NMT training,
widely used NMT toolkits implement disparate strategies for doing so, which
have not been empirically validated or compared. This work investigates
mini-batch creation strategies with experiments over two different datasets.
Our results suggest that the choice of a mini-batch creation strategy has a
large effect on NMT training and some length-based sorting strategies do not
always work well compared with simple shuffling.

DOI： 10.18653/v1/w17-3208

arXiv

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/aclnmt/aclnmt2017.html#conf/aclnmt/MorishitaONYSN17

Publishing type：Research paper (international conference proceedings)   Publisher：Asian Federation of Natural Language Processing  

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/ijcnlp/ijcnlp2017-1.html#conf/ijcnlp/PragstYMNU17

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-3-319-92108-2_15

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-3-319-65521-5_93

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Publishing type：Research paper (scientific journal)  

CiNii Research

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (international conference proceedings)   Publisher：言語処理学会  

<p>Through using knowledge bases, question answering (QA) systems have come to be able to answer questions accurately over a variety of topics. However, knowledge bases are limited to only a few major languages, and thus it is often necessary to build QA systems that answer questions in one language based on an information source in another language (cross-lingual QA: CLQA). Machine translation (MT) is one tool to achieve CLQA, and it is intuitively clear that a better MT system improves QA accuracy. However, it is not clear whether an MT system that is better for human consumption is also better for CLQA. In this paper, we investigate the relationship between manual and automatic translation evaluation metrics and CLQA accuracy by creating a data set using both manual and machine translation, and performing CLQA using this created data set. As a result, we find that QA accuracy is closely related with a metric that considers frequency of words, and as a result of manual analysis, we identify two factors of translation results that affect CLQA accuracy. One is mistranslation of content words and another is lack of question type words. In addition, we show that using a metric which has high correlation with CLQA accuracy can improve CLQA accuracy by choosing an appropriate translation result from translation candidates. </p>

DOI： 10.5715/jnlp.23.437

CiNii Books

CiNii Research

researchmap

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.1587/transinf.2016SLP0018

Web of Science

researchmap

Other Link： http://dblp.uni-trier.de/db/journals/ieicet/ieicet99d.html#journals/ieicet/NioSNYN16

Language：Japanese   Publishing type：Research paper (scientific journal)  

コンピュータから人間へ情報を伝達するためには，コンピュータが処理しやすい表現を，人間に理解しやすい自然言語文に変換する技術が重要である．本研究では，構造化された情報であるフローグラフから自然言語文を自動生成する手法を提案する．手順書のドメインとしてレシピを扱う．まず，実際の手順書に対して重要語とその種類を表すラベルを自動認識し，固有表現に置き換えることで，統計的なテンプレートの自動生成を行う．次に，フローグラフを1つの動作を含む最小手順部分グラフに分割する．最後に，最小手順部分グラフに最もマッチするテンプレートを選択して文生成を行う．テンプレート選択の指標として，負対数確率，内容語残存コスト，エントロピーを用いる．実験では，正解として用意されたフローグラフを表現する自然言語文章と生成された文章を比較評価した．具体的にはBLEU値による自動評価と主観評価を行い，提案手法が精度向上に寄与し，BLEUが人手による主観評価と相関があることを確認した．Algorithms that convert a computational representation to human-readable natural language sentences are essential for communications. In this paper, we propose a method that automatically generates natural language sentences from a structured flow graph. We demonstrate this method in the domain of cooking recipes. Firstly, we automatically generate templates from real recipe texts by recognizing keywords and their type label, and replacing the words with named entity labels. Secondly, we divide the flow graph into minimum procedural subgraphs, each of which has one action. Finally, we select the most suitable template for each subgraph. When selecting a template, we use three measures: a negative logarithmic probability, a remaining content word cost, and a sentence entropy. In the experiments we generate recipe texts from flow graphs and evaluated them through both an objective metric BLEU and a subjective evaluation by human examines. The results showed that our method improves both metrics and BLEU is correlated to the subjective metrics.

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

DOI： 10.1527/tjsai.DSF-517

Scopus

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)   Publisher：Japanese Society for Artificial Intelligence  

DOI： 10.1527/tjsai.C-G31

Scopus

CiNii Research

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.21437/Interspeech.2016-919

Web of Science

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-981-10-2585-3_5

Scopus

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：European Language Resources Association (ELRA)  

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/lrec/lrec2016.html#conf/lrec/LubisGSNYNN16

Publishing type：Research paper (international conference proceedings)   Publisher：European Language Resources Association (ELRA)  

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/lrec/lrec2016.html#conf/lrec/YoshinoHMTIO16

Publishing type：Research paper (international conference proceedings)   Publisher：The Association for Computer Linguistics  

DOI： 10.18653/v1/w16-3610

CiNii Research

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/sigdial/sigdial2016.html#conf/sigdial/MiehleYPUNM16

Publishing type：Research paper (international conference proceedings)   Publisher：The Association for Computer Linguistics  

DOI： 10.18653/v1/w16-3640

CiNii Research

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/sigdial/sigdial2016.html#conf/sigdial/MizukamiYNTN16

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/SLT.2016.7846242

Web of Science

CiNii Research

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/SLT.2016.7846311

Web of Science

CiNii Research

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.1016/j.csl.2015.01.003

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)   Publisher：Springer  

DOI： 10.1007/978-3-319-19291-8_2

Scopus

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

In dialogue system, dialogue modeling is one of the most important factors contributing to user satisfaction. Especially in the Example-Based Dialogue Modeling (EBDM), effective method for selecting response utterances from examples improve dialogue quality. We describe an EBDM framework that predicts user satisfaction to select the best system response for the user from multiple response candidates. This framework select satisfactory response for user preference while prediction using user feedback can be predicts user satisfaction for system responses.

DOI： 10.11517/jsaislud.75.0_23

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：The Japanese Society for Artificial Intelligence  

We propose a dialogue breakdown detector based on the Recurrent Neural Netwoek (RNN). RNN is an extension of the neural network, which can consider old context information in sequential input. We construct a RNN network that receives utterance pairs of user and dialogue system as inputs, and outputs the breakdown label for each system's utterance. The input utterances are separated into vectors of words,cooccurrence words,and represented distributional sentence vectors. Finally, we combined these features and detector evaluated the breakdown detection accuracy.

DOI： 10.11517/jsaislud.75.0_11

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Other Link： https://ndlsearch.ndl.go.jp/books/R000000004-I026828647

Language：Japanese   Publisher：日本産科婦人科学会  

収集根拠 : NII-ELS
資料形態 : テキストデータ
コレクション : 国立国会図書館デジタルコレクション > 電子書籍・電子雑誌 > 学術機関 > 学協会
著者所属: 大阪労災病院
レポート・講演番号: P1-21-3

CiNii Books

CiNii Research

researchmap

Other Link： https://dl.ndl.go.jp/pid/10732064

Language：Japanese   Publisher：日本産科婦人科学会  

収集根拠 : NII-ELS
資料形態 : テキストデータ
コレクション : 国立国会図書館デジタルコレクション > 電子書籍・電子雑誌 > 学術機関 > 学協会
著者所属: 大阪労災病院
レポート・講演番号: P3-42-6

CiNii Books

CiNii Research

researchmap

Other Link： https://dl.ndl.go.jp/pid/10733098

Language：Japanese   Publisher：日本産科婦人科学会  

収集根拠 : NII-ELS
資料形態 : テキストデータ
コレクション : 国立国会図書館デジタルコレクション > 電子書籍・電子雑誌 > 学術機関 > 学協会
著者所属: 大阪労災病院
レポート・講演番号: P1-13-6

CiNii Books

CiNii Research

researchmap

Other Link： https://dl.ndl.go.jp/pid/10732007

Language：Japanese   Publisher：日本産科婦人科学会  

収集根拠 : NII-ELS
資料形態 : テキストデータ
コレクション : 国立国会図書館デジタルコレクション > 電子書籍・電子雑誌 > 学術機関 > 学協会
著者所属: 大阪労災病院
レポート・講演番号: P2-24-7

CiNii Books

CiNii Research

researchmap

Other Link： https://dl.ndl.go.jp/pid/10732445

Language：English   Publishing type：Research paper (scientific journal)  

DOI： 10.1109/TASLP.2014.2387414

Web of Science

researchmap

Language：Japanese   Publishing type：Research paper (international conference proceedings)   Publisher：一般社団法人 言語処理学会  

&lt;p&gt;In syntax-based machine translation, it is known that the accuracy of parsing greatly affects the translation accuracy. Self-training, which uses parser output as training data, is one method to improve the parser accuracy. However, because automatically generated parse trees often include errors, these parse trees do not always contribute to improving accuracy. In this paper, we propose a method for removing noisy incorrect parse trees from the training data to improve the effect of self-training by using automatic evaluation metrics of translations. Specifically, we perform syntax-based machine translation using &lt;i&gt;n&lt;/i&gt;-best parse trees, then we re-scoring parse trees based on the automatic evaluation score of translations. By using the parse trees that have higher score among the candidates for self-training, we can improve parsing and machine translation accuracy by using parallel corpora that are not annotated syntax structure. In experiments, using higher score parse trees for self-training, we found that our self-trained parsers significantly improve a state-of-the-art syntax-based machine translation system in two language pairs, and self-trained parsers significantly improve the accuracy of the parsing itself. &lt;/p&gt;

DOI： 10.5715/jnlp.23.353

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：Japanese   Publisher：情報処理学会  

雑談型音声対話システムの一つにユーザの話を聞く傾聴システムがある。従来の傾聴システムは質問を行う、相槌を打つなどといった基本的な行為が実装されているが、人間の対話でみられるような相槌の形態や韻律を調整することは行われていない。一方、相槌生成のタイミング予測の研究はあるが多様な形態の相槌をうつような枠組みは検討されていない。そこで本研究では、傾聴対話システムの実現を目標として相槌の形態パターンと先行発話の構文的な区切り、節境界の強弱との関係の分析を行う。

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/APSIPA.2015.7415301

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/ASRU.2015.7404867

Web of Science

researchmap

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/ASRU.2015.7404868

Web of Science

researchmap

Publishing type：Research paper (international conference proceedings)   Publisher：The Association for Computer Linguistics  

DOI： 10.18653/v1/w15-3057

J-GLOBAL

researchmap

Other Link： http://dblp.uni-trier.de/db/conf/wmt/wmt2015.html#conf/wmt/SugiyamaMNYSTN15

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

J-GLOBAL

researchmap

Language：Japanese   Publishing type：Research paper (scientific journal)  

本論文では，様々な方言の混合に対応する音声認識システムを構築する．まず，単一方言音声認識の言語モデルを，大規模共通語言語コーパスから擬似生成した方言言語コーパスで学習する．擬似生成には，共通語-方言対訳コーパスからWFST（重み付き有限状態トランスデューサ）によって学習されたルールを用いる．次に，構築された各方言言語モデルを混合し，発話ごとに最適な混合比を推定しながら認識を行う．これは，実際に話される方言が純粋な単一方言ではなく，人の移動やテレビ，ラジオなどの放送の影響を受けた様々な方言の混合であると考えられるからである．この推定には，音声認識用言語モデルにおける対数尤度の値を用いる．実験により，方言音声認識用言語モデルを用いて方言音声の認識精度が向上することを確認した．また，対数尤度と音声認識精度に強い相関があること，対数尤度を最大化する混合比を発話ごとに選択することで，固定混合比の場合と比較して音声認識精度が向上することを確認した．This paper designs and implements an automatic speech recognition (ASR) system that accepts a mixture of various kinds of dialects.The language model for a particular dialect is trained on a dialect language corpus simulated from a large common language corpus.The simulation is carried out with a weighted finite-state transducer (WFST)trained on a parallel corpus of a dialect and common language.The resulting system recognizes dialect utterances with a mixture of dialect language models by estimating the optimal dialect mixing proportion for each utterance.Since actually-spoken dialect is not a purely single dialectbut a mixture of various dialects,influenced by communication in daily lives and broadcasting such as television and radio,estimating optimal dialect mixing proportion, that is,what maximizes the value of log-likelihood forthe input utterance, is critical in ASR.Experiments showed that recognition accuracy improves by usingthe dialect language model,that log-likelihood and recognition accuracyare highly correlated, and that recognition accuracy improvesby choosing the dialect mixing proportion that maximizes log-likelihoodfor each utterance,compared to a fixed dialect mixing proportion.

CiNii Books

CiNii Research

J-GLOBAL

researchmap

Language：English   Publishing type：Doctoral thesis   Publisher：The Japanese Society for Artificial Intelligence  

DOI： 10.14989/doctor.k18614

CiNii Books

CiNii Research

researchmap

Authorship：Lead author   Language：Japanese  

J-GLOBAL

researchmap