The invention relates to an atom pair
distribution function-oriented
database construction method, which comprises the following steps of: receiving actually-measured atom pair
distribution function data. Gr, a corresponding selectable structure file. Cif and related original scattering data. Iq / . Sq / . Fq, analyzing experimental parameters from a PDF
data file header. Gr, carrying out consistency check on the parameters, and determining whether the experimental parameters are consistent or not; the parameters passing the inspection, the corresponding PDF data. Gr, the related original scattering data. Iq / . Sq / . Fq and the standardized
metadata are uniformly stored in a
database; when the
database receives a
verification task, I (Q)-S (Q)-F (Q)-G (r) reprocessing is executed again according to the uploaded
metadata to obtain a verified PDF, a
quality score is generated according to
cosine similarity, PDF data with the
quality score lower than a threshold value are marked, and a difference positioning report is generated. Compared with the prior art, the method has the advantages that the
data management and query efficiency is improved, the data difference source is favorably positioned, and the requirements of
local structure analysis,
artificial intelligence and other applications on high-quality PDF data are met.