Preserved Specimen MIDS scores for GBIF data 2016-2026 (excl 2017)

These Parquet files were generated using code found on https://github.com/AgentschapPlantentuinMeise/mids-bigcalc-parquet, based on Darwin Core archives acquired for each listed year, each encompassing all GBIF occurrences at the time. The (huge) Darwin Core archives were downloaded, unzipped and their occurrence.txt (and multimedia.txt) components were subset for Preserved Specimen (through dwc:basisOfRecord) records only, using DuckDB commands that can be found on the above GitHub repository. The subsetted parquet files were processed using an R script that re-used the calculation code from the MIDSCalculator tool (https://github.com/AgentschapPlantentuinMeise/MIDSCalculator). The resulting files are archived in this Zenodo record. They consist of Parquet files with up to 100s of millions of rows, representing each Preserved Specimen record in GBIF at the time. For each row, boolean values are stored for each MIDS information element. Values for gbif:gbifID, dwc:year, dwc:phylum, gbif:datasetKey, dwc:countryCode and mids:MIDS_level are also retained. The Parquet files were used for an analysis of digitized specimen data in GBIF throughout the last 10 years. The results were presented in Oslo at the TDWG 2026 conference by Elspeth Haston. An extended abstract is forthcoming.