Finished! Looks like this project is out of data at the moment!
The classifications of the largest set of artworks (previously the English set) is nearly finished. Thank you all for your contributions!!! We have now made the small (originally Dutch) set also available in English.
Results
PRELIMINARY RESULTS
We have computed an interim score of the annotations that have been made up to January 29. A lot of paintings have been annotated by volunteers, but we need more annotations of the same painting in order to have more reliable results. Therefore, the results below should be interpreted with caution. It is just a first impression of which species can be seen within our selection of paintings.
The Results below are based on the 1692 classifications of 1054 unique paintings. On 455 of these paintings at least one of the species of the list was identified. The remaining paintings were either determinable up to family level, involved species that did not occur on the list, fantasy fish, fish that were not detailed enough or paintings that did not contain any fish.
Remarkably, the first three species concern freshwater species, which apparently occurred most frequently on paintings between 1500 and 1900. These are followed by the herring, which was already a very important fish in the 17th century for the Dutch fisheries. Also the numbers for the Sturgeon are remarkable. It used to be a common species in Europe, but nowadays has become rare. Within the Dutch waters, this species has considerably dropped in numbers in the 18th and 19th century and disappeared from the Netherlands.
After we have collected multiple annotations for each of the paintings, we will calculate a consensus for each of the pictured fish. We will analyze whether answers from the “expert” version are similar to the answers from the “general” versions. If they are similar, we will aggregate the answers from both versions and calculate the consensus for each of the pictured fish. If there are significant differences between the two versions we will investigate how it differs and how we can combine the two. We will determine the reliability of the results by calculating Pielou's evenness index for each fish. This index is a measure which combines how much agreement there is among people that contributed to the determination of a single fish.
After the project is finished we will make the information publicly available. We are currently looking for existing online open access databases where we could add our data. Additionally, we will give the information to the RKD and Rijksmuseum. They may use it to inform the public which species are on the paintings.
We plan to report the description of the database and how it was collected in a scientific journal, such as Scientific data.