Скачать с ютуб A Catalogue of Databases в хорошем качестве

A Catalogue of Databases

Presentation given by Johannes Englisch at the Language Documentation and Archiving conference #LDA2022 With the growing popularity of the CLDF standard, researchers are constantly adding and updating datasets in all kinds of shapes and sizes. At the time of writing, Zenodo already hosts hundreds of CLDF databases. Keeping track of what's out there becomes more and more challenging – especially considering the decentralised way in which datasets are created. This makes it increasingly difficult to identify areas or language families where data is sparse. To help with this I am making a database catalogue of the different CLDF datasets on Zenodo. This catalogue is populated by crawling Zenodo's repository and extracting metadata from the datasets, collecting information such as: * What type of data does the dataset contain (structural data, word list, dictionary, etc.)? * How many parameters or concepts are in the dataset and how many values are defined? * Which languages/families/areas are covered by the dataset and how densely? This information can easily be drawn onto a map to see what parts of the world might be under-documented. Additionally, one can also shine some light on imbalances across different dataset types and, for instance, find regions that might be covered by an extensive wealth of word lists but lack data on the grammatical structure of the languages. This would make it easier for people working on language documentation to make more informed decisions on where to go next and what kind of data to collect.