• Digital Humanities and Cultural Heritage
• Publications close
• JP close
• HU close

This study presents a groundbreaking approach to preserving the Ainu language, recognized as critically endangered by UNESCO, by developing a bi-directional neural machine translation (MT) system between Ainu and Japanese. Utilizing the Marian MT framework, known for its effectiveness with resource-scarce languages, the research aims to overcome the linguistic complexities inherent in Ainu's polysynthetic structure. The paper delineates a comprehensive methodology encompassing data collection from diverse Ainu text sources, meticulous preprocessing, and the deployment of neural MT models, culminating in the achievement of significant SacreBLEU scores that underscore the models' translation accuracy. The findings illustrate the potential of advanced MT technology to facilitate linguistic preservation and educational endeavors, advocating for integrating such technologies in safeguarding endangered languages. This research not only underscores the critical role of MT in bridging language divides but also sets a precedent for employing computational linguistics to preserve cultural and linguistic heritage.

I analyze factory worker households in the early 1920s in Osaka to examine idiosyncratic income shocks and consumption. Using the household-level monthly panel dataset, I find that while households could not fully cope with idiosyncratic income shocks at that time, they mitigated fluctuations in indispensable consumption during economic hardship. In terms of risk-coping mechanisms, I find suggestive evidence that savings institutions helped mitigate vulnerabilities and that both using borrowing institutions and adjusting labor supply served as risk-coping strategies among households with less savings.

The weak correlation between spatiotemporal fluctuations in nonequilibrium complex systems is shown to govern the fluctuation distribution, maximizing the conditional entropy associated with such fluctuations. The result is illustrated in diffusion phenomena observed in living cells. A generic feature of the weak correlation is briefly mentioned. Comment: 16 pages, no figures. For the proceedings of the 12th International Conference on Mathematical Modeling in Physical Sciences (August 28-31, 2023, Belgrade, Serbia)

ABSTRACTThe grain‐size distributions of sediments can yield important information about sediment provenance and sedimentary processes; however, grain‐size distributions are frequently polymodal, rendering analyses difficult. To improve analyses of polymodal grain‐size data, the present study decomposed the grain‐size distributions of bottom surface sediments from Lake Kitaura, a shallow lake in Japan, into lognormal distributions and performed logratio analysis of their mixing proportions. The polymodal grain‐size distributions of the studied samples were separated into four common components at most sites. This logratio analysis revealed clear differences in the characteristics of the spatial distributions of the separated grain‐size components. The logratio values indicated that the three finer components were uniformly deposited within the lake, whereas the coarsest component was spatially diverse, reflecting differences in their sources and sedimentary processes. These results demonstrate the effectiveness of decomposition and logratio analysis of polymodal grain‐size distributions for estimating sedimentary processes. This method can be applied to modern sediments and for palaeoenvironmental reconstructions using sediment cores.

Many Western-style paintings by Japanese artists in the early 1900s, though maintaining a unique quality, were greatly inspired by the works of Western artists. In this article, we employ machine learning to identify relationships and classify the works of Japanese and Western artists. The relationships are of significant interest to numerous art historians, as they can reveal how Western art was introduced to Japan. Historically, art historians have manually annotated these correspondences, which is a time-consuming and labor-intensive process. In this article, we introduce a new method for finding correspondences between related artworks by comparing their overall outline information. This technique is based on Siamese neural networks (SNNs) and a self-supervised learning approach. Additionally, we have compiled a dataset of illustrations from Japanese artists such as Seiki Kuroda and Western artists such as Raphaël Collin, complete with correspondence annotations. However, to exhibit the unique quality of works by Japanese artists, we demonstrate that machine learning can classify between artworks created by Japanese artists and those created by Western artists.