Materials Database Group
Change Storage to Knowledge!
Materials Database Group progresses: 1) Creation and quality control of materials database applicable in the bigdata generation, 2) Classification and integration of the accumulated data, 3) Interactive data-driven by direct connections between laboratories and database, and 4) Reliable data-sharing through objective analyses and authorization systems. We contribute to materials developments and analyses by construction of database advanced in efficiency, quality, and quantity.
Specialized field and Research target
After the World War Ⅱ, the number of scientific journal is continuously increasing. For example, the number of active journals in 1945 which are accessible with Scopus was 1,000, while those at present are roughly 22,000. The growth rate is 10 times per 50 years except for the abrupt growth just after the WWII. Moreover, it is notable that page of each journal also increases 10 times per 50 years. Namely, the scientific document information has increased by ~100 times in the past 50 years. Imagine that the journals share your bookshelves every month!
The scientific journal is one of the most valuable information sources. We must accumulate and put in order the document information with 100 times efficiency than 50 years ago. In the serious situation that the same or higher efficiency is required for the data accumulation in future, we should reconsider the techniques for database formation.
The scientific journal is one of the most valuable information sources. We must accumulate and put in order the document information with 100 times efficiency than 50 years ago. In the serious situation that the same or higher efficiency is required for the data accumulation in future, we should reconsider the techniques for database formation.
1) Sustainable creation and quality control of materials database
We progress mechanical database-creation, namely text data mining (TDM). TDM should be performed under quality control of the accumulated data.
We also find high quality datasets in scientific societies and create practically usable databases.
We also find high quality datasets in scientific societies and create practically usable databases.
2) Classification and integration of the accumulated data
We research techniques of classification and integration of scattered data in the long history of science. Background of the solo data is summarized as metadata, and careful comparison of the metadata should be performed before data integration.
3) Direct connections between laboratories and materials database
We accumulate hot data obtained at laboratories through online services such as 24-hour data analyses and verifications using our database. For this purpose, we develop advanced database creator (artificial intelligence database, AIDB). Researchers who provide hot data may use related data in our database – It is the open science policy.
4) Reliable data-sharing of laboratory data
We establish an authorization system by using objective evaluation using AI. After machine learning, AI suggests a lot of possible structural models to explain the experimental data in spectroscopy. The models are in ranking in their probability. All the possible models and original spectrum are authorized with DOI (digital object identifier) and stored into a database.
