Servicenavigation

Project Details

Back

Projekt Print View

learning Cyclotron

Subject Area Data Management, Data-Intensive Systems, Computer Science Methods in Business Informatics
Term from 2020 to 2025
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 442581111
 
Final Report Year 2023

Final Report Abstract

(1) Research Summary Learning Cyclotron is a mechanism to support human learning by three functions, “Perceiving”, “Mastering”, and “Transferring”. “Perceiving” is intelligent sensing technologies for learners’ internal and knowledge states. “Mastering” is an AI-assisted learning to acquire knowledge and skills effectively and efficiently. “Trasferring” is an intelligent framework of “learning-by-teaching”, and its fundamental technologies. We also develp “nudging strategies for all three functions. The Japanese team has a background of learning augmentation worked with the German team, with a special focus on “Learning English as a second language.” The German team, in addition, has its own advantage of analyzing the process of learning programming. The French team consists of specialists of speech and language analysis. Thus, we assign the following tasks to each team. The Japanese team works on English learning focusing on visual modality with the German team, and speech modality with the French team. The German team is focusing on the analysis of programming learning with visual modality. The French team works mainly on speech modality of English learning together with the Japanese team. The following important results were obtained by the contributions of each team. The Japanese team analyzed fundamental aspects of learning in relation to sleep and inthe-wild data acquisition. It also developed mobile and stationary methods of sensing learners’ internal states, methods for “mastering” to boost learning with various mechanisms, and edutainment methods with Mangas. The German team developed two important software systems called “TrackThink” and “EnGauge”, which are employed for learning programming and analyzing communication during the “transferring,” respectively. The French team established a protocol for analyzing speech for L2 English learners, and the tool called “Eye-got-it” which allows simultaneous acquisition of speech and eye gaze data. (2) Remarkable Research Achievements 1. Summary Description for 1st result In order for the developed technology to be useful under real-world conditions, the original data for technology development must reflect the real-world environment (i.e., data acquired under so-called in-the-wild conditions). In addition to recording data of various modalities in a relevant manner, labels for machine learning are required. For this purpose, the Japanese and German teams developed a data labeling mechanism and a plug-in to obtain web search logs. The French team developed Eye-Got-It, a mechanism for synchronous acquisition of voice and gaze. 2. Summary Description for second result For learning to be efficient and effective, the environment in which the learning takes place is important. Specifically, the following factors are important: the state of the learner (sleep quality, respiration, exercise, etc.), the environment in which learning takes place (context of learning, other media (context, pictures, sound, etc.) that assist understanding, and relationships with others (peer pressure, etc.)). In this study, we conducted various experiments on these matters and obtained knowledge on what, when, how much, and to whom these factors affect. 2 3. Summary Description for third results The most important result mainly achieved by the French team are the design of a multimodal EFL (English as foreign language) protocol, the evaluation involving the development of a multimodal data tool, and finally the acquisition and constitution of a database of 75 speakers. This database enables us to identify and model L2 mastering relevant cues and propose hierarchy of features and preliminary propositions of cross L2 levels and cross-cultural distinctive features relevant to French and Japanese speakers. We also worked on a comparison between text read with and without nudging strategies such as the use of Manga. 1. Summary Description for 1st result One of the features of this research is that we have conducted extensive research to develop the process of “Matering,” which has not been well explored in the past. Specifically, these include confidence estimation during question answering, adaptive generation of questions and materials according to the learner, learning support using smartphone sensors, edutainment using Mangas as learning materials, and gamification. The possibilities for the generation of learner-specific questions and teaching materials have expanded with the advent of generative AI such as ChatGPT. We are currently conducting experiments to evaluate the effectiveness of this technology. 2. Summary Description for second result Another feature of this research is that it examines knowledge transfer methods and nudging strategies for knowledge transfer. This is a major feature of this study that is not found in other studies. Specifically, we developed the following methods and verified their effectiveness experimentally. (1) DiscussionJocky: a method to intervene in a multi-person conversation to prevent one person from speaking out of turn, (2) EnGauge: a method to read Engagement in dialogue from upper body video data, (3) Analysis of problems in knowledge sharing communities such as Stack Exchange, and (4) interventions using making-it-a-habit and peer pressure. Since many of these methods can be used beyond English and programming learning, they can be expected to be applied to other fields. 3. Summary Description for third result The main achievements of the French team consist of the conception of an innovative protocol to acquire and further process multimodal data in order to cover 2 out of the 3 objectives of the project: (I) perceiving and (ii) mastering with focus on English as foreign language for native French and Japanese learners. The second main result is the acquisition of a database including 75 speakers. 1. Summary Description Andrew Vargo, Motoi Iwata, Mathilde Hutin, Sofiya Kobylyanskaya, Ioana Vasilescu, Olivier Augereau, KoWatanabe, Shoya Ishimaru, Benjamin Tag, Tilman Dingler, Koichi Kise, Laurence Devillers and Andreas Dengel. “Learning Cyclotron: An Ecosystem of Knowledge Circulation”. In Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct Publication (UbiComp ’22 Adjunct), pp. 308–312, 2022. 3 The circulation of knowledge is an important function for enriching our social life. Knowledge circulation, which has traditionally been conducted manually, is undergoing a transformation in the age of artificial intelligence. In this paper, we describe our project called “Learning Cyclotron (LeCycl),” which was initiated to accelerate the circulation of knowledge for building a knowledge ecosystem based on artificial intelligence technology. The three functions of sensing, mastering, and transferring knowledge are effectively operated through the power of AI-empowered digital nudging strategies. We outline what has been accomplished to date and summarize future directions for the ultimate goal of LeCycl. 2. Summary Description Andrew Vargo, Shoya Ishimaru, Md. Rabiul Islam, Benjamin Tag, Koichi Kise, Obtaining Labels for In-the-Wild Studies: Using Visual Cues and Recall, IEEE Pervasive, Vol.21, Issue 1, pp.9-17, 2021.12, DOI: 10.1109/MPRV.2021.3129500 The observer effect challenges laboratory research. In-the-wild studies mitigate this but struggle with accurate data labeling for algorithm training. Manual labeling can be obtrusive, time-intensive, and raise privacy concerns. We introduce a labeling workflow from an inthe-wild study examining cognitive states via eye-gaze in real-world settings. Our method, employing J!NS MEME electrooculography glasses, Narrative Clip 2 cameras, and a unique data tagging software, allows participants to label data swiftly and discreetly. It ensures data quality and privacy, is replicable for field research, and is apt for the pandemic and postpandemic era. 3. Summary Description Hutin, M., Kobylyanskaya, S., Devillers, L. : Nudges in technology-mediated knowledge transfer : Two experimental designs. In : Adjunct Proceedings of the 2022 ACM International Joint Conference on Pervasive and Ubiquitous Computing and the 2022 ACM International Symposium on Wearable Computers. pp. 267–273 (2022) Recent advances in technologies now allow us to learn almost anything in virtual environments, be it via Internet forums or websites, telephone apps, video games, and many more. Such technology-mediated learning can be enhanced with the use of embedded nudges, i.e., devices in the architecture of choice to encourage (nudge) the users towards one choice rather than the other without limiting their freedom of choice. This paper presents an overview of how nudges can help improve knowledge acquisition, as well as a two ongoing projects. Ethical issues are also highlighted.

Publications

 
 

Additional Information

© 2025 DFG Disclaimer / Copyright / Privacy Policy

Textvergrößerung und Kontrastanpassung