Papers by Koh Mitsuda
Generating Responses that Reflect Meta Information in User-Generated Question Answer Pairs (2020.lrec-1)
Copied to clipboard
Takashi Kodama, Ryuichiro Higashinaka, Koh Mitsuda, Ryo Masumura, Yushi Aono, Ryuta Nakamura, Noritake Adachi, Hidetoshi Kawabata
| Challenge: | Existing approaches to realize consistent personalities require expensive data collection. |
| Approach: | They propose to collect question-answer pairs for particular characters from online users . meta information such as emotion and intimacy was also collected . |
| Outcome: | The proposed method can be used to train neural conversational models with high quality questions and meta information. |
Dialogue Collection for Recording the Process of Building Common Ground in a Collaborative Task (2022.lrec-1)
Copied to clipboard
| Challenge: | Existing studies on the process of building common ground have not been well conducted. |
| Approach: | They propose a method for recording the process of building common ground through a dialogue by using the intermediate result of a task. |
| Outcome: | The proposed method can record the building common ground process by using the intermediate result of a task and can be estimated quite accurately. |
Investigating person-specific errors in chat-oriented dialogue systems (2022.acl-short)
Copied to clipboard
| Challenge: | Errors in general chatbots and chatbot that follow a rough persona have been studied . but those in chatbot based on real people have not been thoroughly investigated . |
| Approach: | They analyze dialogue data of a generation-based chatbot trained from dialogue data . they find errors in attributes and relations can be divided into two levels: self and other . |
| Outcome: | The results show that errors in chatbots can be divided into two types . the correspondence with an existing taxonomy of errors was also investigated . |
Combining Argumentation Structure and Language Model for Generating Natural Argumentative Dialogue (2022.aacl-short)
Copied to clipboard
| Challenge: | Argumentative dialogue is important process where speakers discuss a specific theme for consensus building or decision making. |
| Approach: | They propose a method to generate argumentative dialogues by combining argumentation structure and language model. |
| Outcome: | The proposed method significantly improves the naturalness of arguments without losing consistency. |
Integrating Pre-Trained Speech and Language Models for End-to-End Speech Recognition (2024.findings-acl)
Copied to clipboard
| Challenge: | Mainstream of automatic speech recognition (ASR) has shifted from pipeline methods to end-to-end (E2E) methods. |
| Approach: | They propose to integrate a pre-trained speech representation model and a large language model (LLM) for automatic speech recognition in an end-to-end manner. |
| Outcome: | The proposed model achieves comparable performance to modern E2E ASR models by utilizing powerful pre-training models with the proposed integrated approach. |
Dialogue Corpus Construction Considering Modality and Social Relationships in Building Common Ground (2022.lrec-1)
Copied to clipboard
| Challenge: | Several studies have examined the process of building common ground in text chat, but none have investigated the process in depth. |
| Approach: | They constructed a dialogue corpus to investigate the process of building common ground with a particular focus on the modality of dialogue and the social relationship between workers. |
| Outcome: | The results suggest that adding the modality or developing the relationship between workers speeds up the building of common ground. |
A Speculative and Tentative Common Ground Handling for Efficient Composition of Uncertain Dialogue (2022.lrec-1)
Copied to clipboard
| Challenge: | a study explores how the grounding process is composed and adapts to human cognitive processes . common ground is a set of information shared among participants that serves as a precondition for understanding individual utterances . |
| Approach: | a study investigates how the grounding process is composed by participants . it suggests that common ground may not necessarily be formed bottom-up through analytic expressions . |
| Outcome: | a new approach to human-like dialogue may be more suitable for natural human communication, the authors say . they show that common ground is mutually accepted among participants through holistic expressions . |
Release of Pre-Trained Models for the Japanese Language (2024.lrec-main)
Copied to clipboard
Kei Sawada, Tianyu Zhao, Makoto Shing, Kentaro Mitsui, Akio Kaga, Yukiya Hono, Toshiaki Wakatsuki, Koh Mitsuda
| Challenge: | democratization of AI aims to create a world where everyone can use AI . pre-trained models with high performance in Japanese are lagging in non-English-speaking communities . |
| Approach: | et al. released large-scale pre-trained models trained on large-data to improve access to AI . authors say the models are more accurate and more accurate than those trained in the English language . e-mail protected: email protected. |
| Outcome: | a new study shows that pre-trained models specialized for Japanese can achieve high performance in Japanese tasks. |
PSLM: Parallel Generation of Text and Speech with LLMs for Low-Latency Spoken Dialogue Systems (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models that process both text and speech face problems in response generation latency. |
| Approach: | They propose to extend the input and output sequences of the language model to support the parallel generation of text and speech. |
| Outcome: | The proposed model improves latency while maintaining quality of response content while maintaining the quality of the response content. |