https://en.people.cn/n3/2025/1030/c90000-20384344.html

Oracle bone inscriptions, or 甲骨文 Jiaguwen, are an ancient Chinese language named for their inscriptions on tortoise shells and animal bones. They are a primitive form of Chinese characters and the oldest fully-developed characters in China. More than 160,000 oracle bone inscriptions have been excavated in Anyang since the Yinxu ruins were discovered in 1899. These relics are now in collections scattered across both China and abroad.

Chinese tech giant Tencent and the AI research institute of Xiamen University have now teamed up to analyse these inscriptions using AI, integrating two main models of DeepSeek and Tencent’s Hunyuan, as well as nearly 10 other auxiliary models. This AI tool can answer user queries by recognizing an image of an oracle bone character and interpreting it.

The database is claimed to be by far the world’s largest multimodal dataset of oracle bone inscriptions, which includes data for 1.43 million ideographic characters, 15,000 pieces of multimodal oracle bone data and 3,000 research papers. The platform is not only able to help enhance research efficiency for those who do serious research concerning these ancient characters, but can also assist ordinary users keen to obtain interesting knowledge related to oracle bone scripts, via a mini-program on the instant messaging platform WeChat.