Evaluation data for subordinate clauses from Swedish literary text from 1800–1930. The dataset is described in:
Sara Stymne, Carin Östman, and David Håkansson (2023) Parser Evaluation for Analyzing Swedish 19th-20th Century Literature. Accepted to The 24th Nordic Conference on Computational Linguistics (Nodalida'23). Tórshavn, Faroe Island. May 22nd-24th, 2023.
The dataset is available in the file "literature-sample-annotated.csv". Each line contains the type of relation, the class (1: the relation occur with the marked word, or 0: the relation does no occur with the marked word), and the annotated sentence, where the a focus word is marked with asterisks. The possible types are:
- relcl: the focus word is the head of a relatvie clause, acl:relcl
- cleft: the focus word is the head of a cleft construction, acl:cleft
- ccomp: the focus word is the head of a clausal complement, ccomp
- no-aux: the focus word is the head of a clausal complement where the verb is infinite with a dropped auxilliary verb
All sentences are sampled from works in Litteraturbanken. A list of works from which sentences are smapled, is available in the file "list-of-works.csv". Before sampling, the texts were split into speech segment and narrative, based on the occurence of quotation marks.
The data is released under the Creative Commons license, CC BY-NC-SA
