[4Xin2] Poster session 2

[4Xin2-80] Quantitative evaluation of the degree of consecutive success extraction of the hot-streak model

〇Yuta Tomokiyo¹, Takahiro Miura¹, Noriyuki Higashide¹, Kimitaka Asatani¹, Ichiro Sakata¹ (1.The University of Tokyo)

Keywords:Computational Social Science, Science of Science, Hot streaks

It is known that there are periods of success in an individual researcher's career, known as hot streaks, in which there is a series of extraordinary performances that are better than usual. Although [Liu 18] et al. proposed a hot-streak model that captures consecutive success, it is a model that aggregates time-series data on performance using moving averages and fits it with a piecewise function using optimization methods, which does not necessarily guarantee that it captures strict consecutive success. Hot streaks can indicate periods of creativity and productivity prominence in a researcher's career. Going into the context of these periods may provide helpful insights for the research community and science and technology policy. It is meaningful to examine the robustness of the hot-streak model that underlies these periods. The results show that the hot-streak model considers a few out-of-the-ordinary one-shot successes as hot streaks in two respects: the moving average and the interpretation of the data based on the assumption of the researcher's potential. The results suggest that hot streaks need to be detected, for example, in combination with a model that focuses on serial successes observed in the number of citations of the papers used to validate this study.