Organizing an in-class hackathon to correct PDF-to-text conversion errors of Genomics & Informatics 1.0

Area of Science:

• Biomedical Informatics
• Corpus Linguistics
• Digital Humanities

Background:

• Extracting text from multi-column biomedical journals like Genomics & Informatics presents significant challenges.
• Previous versions of the full-text corpus contained errors from PDF-to-text and optical character recognition (OCR) conversion.
• Accurate and clean text data is crucial for computational linguistic analysis and research.

Purpose of the Study:

• To describe a community-driven initiative to enhance the Genomics & Informatics full-text corpus.
• To address and correct PDF-to-text and OCR errors within the corpus.
• To foster knowledge creation in corpus linguistics and develop practical skills among researchers and students.

Main Methods:

• A hackathon event, the Genomics & Informatics Annotation Hackathon (GIAH), was organized.
• Projects involved semi-automatic detection and correction of conversion errors.
• An internal database of corpus versions and annotations was utilized.

Main Results:

• Successful identification and correction of PDF-to-text and OCR errors in the Genomics & Informatics corpus.
• Improved data quality for the full-text corpus.
• Community engagement in corpus improvement and annotation.

Conclusions:

• Community hackathons are effective for improving specialized biomedical corpora.
• The GIAH event facilitated knowledge sharing and skill acquisition in corpus annotation.
• Enhanced corpus data supports further research in genomics, informatics, and corpus linguistics.