Records of conventional scientific research are generally left in the form of laboratory notebooks, academic papers, patents, and the like.In order to construct a database for analysis, a researcher must read and decipher the experiment notebook.In addition, the literature often omits detailed conditions such as temperature and humidity during experiments and the results of failed experiments, and it is difficult to say that the factors involved in the results of experiments are completely covered.

　Therefore, in this research, as an AI to manage this information in an integrated manner, we constructed a system that records the state of daily material experiments in a data format called a graph structure and automatically analyzes it with AI.In addition to experimental procedures and results, various information that can affect material properties such as the date, temperature, equipment and sample conditions, etc., is recorded as data that connects points and lines, and is called a fingerprint. Automatic analysis by algorithmic method.

　This AI system was actually used in the experimental research on polymer solid electrolytes that the researchers are working on.Data from more than 500 experiments, including failures, were digitally recorded and analyzed by an AI system.

　The AI ​​system, which can be called an "electronic experiment notebook system" constructed this time, can be deployed in various materials research, and will be a stepping stone to promote the digital transformation (DX) of experimental research.In addition, it is expected that the flow of open science through data science will be strengthened by making it easier to publish raw experimental data.

Paper information:[npj Computational Materials] Exploration of organic superionic glassy conductors by process and materials informatics with lossless graph database