Real-time Support for Solution Search in Human-based Evolutionary Computation
Takato Utsumi, Hiroki Muta, Chen-Kun Tsung, Kei Ohnishi · 2024
We propose a method to extract important words common to good solution candidates in human-based evolutionary computation, like building blocks in genetic algorithms, and present them to the participants in a realtime manner. We also propose a method to obtain and present words similar to the important ones. Furthermore, we propose a method to measure the similarity between participants using the important words for each participant and present it to the participants. All of the proposed methods help participants solve problems by reducing the cognitive load of finding key words and persons. We implement those methods into a human-based evolutionary computation system and evaluate the proposed methods through human subject experiments using the system. The experimental results show that although the evaluation result to the outputs from the system incorporating the methods by a third party is not so good, the system is shown to be useful for participants to create solutions.