It is an irreplaceable important issue in the artificial intelligence (AI) development of the 21st century to establish a database, to recover the related memory, and even to extract big data from written texts in the Chinese characters cultural circle. Theories about the “transitional forms” and “intermediary graphs” of Chinese characters provide theoretical basis for the big data excavation of the written texts in Chinese characters cultural circle. Chinese characters have been deconstructed twice during its long-time development history from the Warring States period to the Six Dynasties period. Experts tend to focus on the first time which is known as “libian” (transformation of the graphic shapes of the zhuanshu into clerical script forms) so there are lots of investigation and research on it. Regard to the second time, as the cursive scripts prevailed, the components of a graph can be reorganized or distorted and some strokes’ order can change the direction in hand-writing process, making a new distinguishing mechanism to produce batches of “transitional forms”. A very few of the “transitional forms” were evolved into “fixed standard scripts”, and a bulk of them were conversely developed as “intermediary graphs” gradually. The first priority for establishing a database for the written texts in the Chinese characters cultural circle is to exclave massive amounts of related information based on big data construction. It is imperative to collect and sort the ancient written letters which are of great value in Japan, Korea and China. This database acts as not only the foundation of the research on history of Chinese writing systems and new historical data, but also the “firsthand data” for revealing the rules of the development of cognitive behavior and intellectual inheritance of human being.