研究課題/領域番号 |
18K18085
|
研究機関 | 東京大学 |
研究代表者 |
ハウタサーリ アリ 東京大学, 大学院情報学環・学際情報学府, 特任講師 (70752236)
|
研究期間 (年度) |
2018-04-01 – 2022-03-31
|
キーワード | socio-emotional / multilingual / communication |
研究実績の概要 |
Third iteration of the development of a software tool for supporting foreign students’ socio-emotional communication was completed. The approach taken in the prototype development was to design an automated method to generate unique speech balloons with ACGAN that match and convey a message sender's level of emotional arousal to the receiver. The speech balloons represent an abstraction of the level of arousal in emotional text-based messages to non-native speakers who may have difficulties detecting the emotional nuances in second language messages.
The system design and evaluation results were submitted to a top international conference and are presently under review.
Development of a mobile application to collect native and non-native English speakers' valence and arousal evaluations of English emotional words was completed. This application is used to collect data via crowdsourcing for the creation of a multilingual corpus to improve the accuracy of emotional valence detection based on users' first and second language backgrounds in the developed socio-emotional communication support system. An application for conserving nuances conveyed in typefaces or fonts based on contextual information when translating text from one language to another was also developed.
|
現在までの達成度 (区分) |
現在までの達成度 (区分)
1: 当初の計画以上に進展している
理由
The socio-emotional communication support system development is proceeding smoothly thanks to effective collaboration with other researchers and developers. In the second iteration of the prototype, a method for capturing message senders’ level of arousal from voice-input was developed to overcome challenges in using human-annotated text as training data. In the third iteration, a system to automatically generate unique speech balloons with ACGAN corresponding to the level of emotional arousal in a message was developed to display more nuanced emotional information in text-based communication. In addition, a prototype system for conserving the nuances in typefaces during translation of text was developed.
However, due to the COVID-19 pandemic, evaluation of the system in on-site controlled experiments was difficult, and hence the system was implemented as a web application to enable remote experiment participation. Presently, the process of combining the system of displaying emotional valence in fonts developed in FY2018 with the system for displaying emotional arousal in speech balloons in one automated system, as well as evaluation through controlled experiments is underway. Development of a mobile application with a novel interface for gathering emotional evaluations via crowdsourcing was completed. In order to create a reliable human-annotated corpus, evaluations from both native and non-native English speakers will be consolidated on two crowdsourcing platforms.
|
今後の研究の推進方策 |
- Conduct a multilingual communication experiment with the developed systems using fonts for valence and speech bubbles for arousal abstraction in order to evaluate the efficacy of the proposed approach on interpersonal relationship building in a controlled laboratory setting. - Continue the creation of the proposed non-native sentiment corpora for further system development by utilizing a mobile application to gather human evaluations, and analyze the discrepancies in emotional word evaluations between native and non-native speakers of English. - Naturalistic and longitudinal user studies on the effect of the proposed systems on relationship management are planned to be conducted. - Development of a cross-cultural training program based on the research results.
|
次年度使用額が生じた理由 |
Travel costs were reduced due to the COVID-19 pandemic. The next publications are planned to be presented in international conferences and journals, and the travel costs will be allocated for the purpose of travel, participation and publication costs. Personnel expenditure was reduced due to postponement of a laboratory experiments in FY2020, and are allocated to be used for controlled laboratory experiments, crowdsourcing costs, as well as software development, conducted in FY2021.
|