CODH Seminar Shows LLMs Judging Historical Earthquake Intensity from Japanese Archives
tkasasagi · x · 2026-09-10
Japan's ROIS-DS Center for Open Data in the Humanities (CODH) will host its 27th seminar on building and using historical big data on Sept 18, 2026, hybrid at NII, free with registration.
- Context: Eight years after CODH first outlined the historical big data vision, data releases and tooling have matured, and generative AI opens new possibilities — but speakers stress AI needs a solid manually-built data foundation.
- Highlights:
- The "rekichizu" historical map service: 35,000+ km of road data and 6,300+ historical POIs built by hand from primary sources with QGIS, with 8,000 km of major highways released as open data
- Late-Edo village boundary data reconstructed entirely from open data and OSS tools, showing 400 years of rice-yield (kokudaka) distribution
- A dataset of damage and intensity for the 1855 Ansei Edo earthquake
- Using LLMs to judge seismic intensity from historical earthquake records, with batch text processing to reduce misses and self-checking first-pass judgments so researchers can verify evidence (Univ. of Tokyo Earthquake Research Institute)
- Text repository construction and structuring of disaster archives (National Museum of Japanese History)
More from AGI Musings
- Agents can't prove humans are real: a fascinating new argument about simulation — CatAstro_Piyush · 2026-09-10
- Anthropic models extreme AI scenario: +32% US GDP by 2030, 17.9% knowledge-worker unemployment — robleclerc · 2026-09-10
- Shermer & Pinker mock media's fixation on AI doom over real progress — sapinker · 2026-09-10
- Most likely AI risk: adversaries injecting harmful instructions into pervasive AI — tristanbob · 2026-09-10
- A third of published ML researchers say AI extinction risk exceeds 10%, 2023 survey shows — ben_j_todd · 2026-09-10
- Slowing AI isn't realistic — accelerating safety research is the better bet — tristanbob · 2026-09-10