Integração do HCF-web a herbários virtuais brasileiros

Abstract

The Herbarium of the Federal University of Technology - Paraná, on the Campo Mourão Campus, also known as HCF, shares its collection throught its web system (HCF-Web) with the Reflora Virtual Herbarium and the National Institute of Science and Technology Virtual Herbarium of Flora and Fungi (INCT-HVFF). The two virtual herbariums accept suggestions on sample re-identification elaborated by experts. However, suggestions are not automatically updated into the HCF-Web system and, therefore, such data are manually compared with equivalents on virtual herbariums. Since HCF has more than twenty-eight thousand exsiccates, the identification of inconsistencies demands a long time and is errorprone. This work presents the development of an HCF-Web extension that compare data to their equivalents in virtual herbariums so that divergences can be highlighted. In order to implement this solution, it was used: (i) the Reflora’s API for immediate or programmed updating; (ii) and a CSV file created by the speciesLink system from INCT-HVFF. To test this approach, nine HCF samples, ranging from five to twenty-five thousand specimens were randomly selected to validate functions. Each sample was tested three times to: (i) calculate mean execution time, in minutes; (ii) and verify the number of divergences. It was also gathered the number of fails during Reflora’s API execution. As expected, time for the divergence identification varied linearly according to the samples’ size. Regarding the INCT-HVFF, the Internet connection is not required to compare HCF and speciesLink data. Therefore, identification of divergences is faster than in the Reflora process. Finally, this research contributions are: (i) a solution to issues on data inconsistency between HCF and virtual herbariums; (ii) increase in precision concerning the identification of divergences; (iii) enhancment of the workflow of the HCF personnel, since manual data comparison, which usually took weeks to be performed, can now be automatically made in a shorter time.