Harvesting metadata from the correspondence of Edvard Munch

At NorKorr we are currently building a script to harvest and consolidate metadata from the correspondence of Edvard Munch.

Edvard Munch’s correspondence comprises approximately 12 000 letters to and from Edvard Munch, where 10 000 are in the Munch Museum (around 6 000 incoming letters addressed to Munch and 4 000 letters and letter drafts by Munch) and 2 000 in other institutions or on private hands. We still discover new letters at auctions or on private hands, so the number of letters is still growing.

MM.N.00830, Munchmuseet. Letter to Karen Bjølstad dated 8.5.1899.

The museum has set up the website emunch.no where we aim to publish not only the correspondence, but all of Munch’s writings which include diaries, notes on art, literary sketches and fragments, prose poems, various lists and so on. We have published 80 % of Munch’s own texts and 60 % of the letters he received.

Unfortunately, today we are in a situation where we have conflicting sources for the metadata. The website is built on XML and XSLT technologies, and all register files containing metadata were originally created through export-to-XML from the project’s databases. When the museum’s databases were consolidated into a single database and migrated to a new software, we lost the export-to-XML functionality. Since then, there have been two sets of metadata for Munch’s writings, where both have been updated. The intention was, of course, that the two sets were to be identical, but as often happens we have forgotten to update one or the other. Today, the two sets have many differences, and it is not given which source has the correct data. A third source is a spreadsheet we use in our ongoing work of updating and revising the dating of Munch’s letters and letter drafts. We work in a spreadsheet because it is more time-efficient than using our database and because the database cannot easily handle the order of the dated letters.

The data from the spreadsheet will eventually be incorporated into the database, and the differences between the XML register files and the database will also be consolidated. We are also working on a solution where we once again can publish directly from the database to the website. For the current process of creating a CMIF XML file we need a script which does the consolidation for us. The script is being developed by research assistant Loke Sjølie and can be found at GitHub (https://github.com/Loke-git/MXML).

As a result of the script development process, we have discovered duplicated letter entries, and errors and invalid date formats. We revise and correct our metadata, but we also need to look at how we deal with non-existing dates, e.g. November 31st or February 29th in non-leap years.

MM.K.00242, Munchmuseet. Letter from Gunnar Heiberg dated 29.02.1895.

Search OpenEdition Search

You will be redirected to OpenEdition Search