Error Correction Environment for the Polish Parliamentary Corpus
Maciej Ogrodniczuk | Michał Rudolf | Beata Wójtowicz | Sonia Janicka
Proceedings of the Workshop ParlaCLARIN III within the 13th Language Resources and Evaluation Conference
The paper introduces the environment for detecting and correcting various kinds of errors in the Polish Parliamentary Corpus. After performing a language model-based error detection experiment which resulted in too many false positives, a simpler rule-based method was introduced and is currently used in the process of manual verification of corpus texts. The paper presents types of errors detected in the corpus, the workflow of the correction process and the tools newly implemented for this purpose. To facilitate comparison of a target corpus XML file with its usually graphical PDF source, a new mechanism for inserting PDF page markers into XML was developed and is used for displaying a single source page corresponding to a given place in the resulting XML directly in the error correction environment.