Paper short abstract:

The Dutch Folktale Database underwent technical changes that make the addition of metadata easier. The interpretation of data is facilitated by means of visualisation: geographical maps, timelines, a network of similar tales, and wordclouds. The database is ready to interact with similar databases.

Paper long abstract:

In 1994 the Dutch Folktale Database started as a stand-alone database and came online in 2004: www.verhalenbank.nl. After two large projects (FACT and Tunes & Tales), since 2016, all kinds of metadata can be added automatically and semi-supervised: languages, names, keywords, summaries, subgenres, motifs and tale types. For this, the database went over to a new platform called Omeka that fits the needs of many databases in the humanities, and that can handle all kinds of plug-ins. The following techniques have been used: n-grams, language detection, named entity recognition, keyword extraction, summarisation, bag of words, machine learning and natural language processing. Furthermore MOMFER, a search engine for MOTIFS has been added as well (www.Momfer.ml). The search capabilities of Omeka turned out to be a little disappointing, and have been further enhanced by adding Solr. The interpretation of data is facilitated by new means of visualisation: geographical maps, timelines, a network of similar tales, and wordclouds. It is now possible to ask for tales told in the 20th century in Rotterdam and its surroundings for 15 kilometers. Since the database meets the requirements of Dublin Core, a connection with similar databases or a data harvester is made possible. Recently, a Trans-Atlantic Digging into Data application has been sent in to build a harvester called ISEBEL: Intelligent Search Engine for Belief Legends. The harvester should be able to search into a Dutch, Danish and German database simultaneously. Lateron, other databases can join in.