Corpus of Slovak Legislative Documents

Abstract

Abstract The article describes the construction of the corpus of Slovak legislative documents. By analyzing several statistical values of the source metadata and documents, we efficiently improve corpus quality. We describe the methods used to clean up small variations in metadata, length based discrimination of document and examine the effectiveness of several strategies of deduplication. The corpus is a part of a comparable corpus of legislative documents of seven languages, created in the Multilingual Resources for CEF.AT in the Legal Domain (MARCELL) project.