| 研究開始時の研究の概要 |
Data is the new oil in the intelligence era. While artificial intelligence (AI) models can learn human-like intelligence from historical data, it is common for companies to lack sufficient data to train those models. On the other hand, mountains of data are generated in the world every second, but most of them do not circulate in society and thus cannot be fully exploited. Hence, to promote the circulation and use of data for AI applications, we aim to develop fundamental technologies for building a privacy-preserving, trustworthy, and user-friendly data marketplace.
|
| 研究実績の概要 |
Our contributions for this fiscal year are twofold. First, we conducted an interdisciplinary study on data protection in data markets. This study discusses the ambiguous boundaries among different categories of user data as defined in the GDPR from a legal perspective and proposes a computational method to delineate these boundaries clearly. Second, we developed a computer simulation framework to simulate data trading in practice. This framework employs large language model agents to represent the various parties in data markets and to simulate their interactions during data trading. Based on the simulation results, we can more accurately assess the performance of data trading mechanisms in society, thereby fostering trustworthy data trading.
|
