Lost in migration: document quality for batch conversion to PDF/A

Abstract

Purpose Changes in file format specifications challenge long-term preservation of digital documents. Digital archives thus often focus on specific file formats that are well suited for long-term preservation, such as the PDF/A format. Since only few customers submit PDF/A files, digital archives may consider converting submitted files to the PDF/A format. The paper aims to discuss these issues. Design/methodology/approach The authors evaluated three software tools for batch conversion of common file formats to PDF/A-1b: LuraTech PDF Compressor, Adobe Acrobat XI Pro and 3-HeightsTM Document Converter by PDF Tools. The test set consisted of 80 files, with 10 files each of the eight file types JPEG, MS PowerPoint, PDF, PNG, MS Word, MS Excel, MSG and “web page.” Findings Batch processing was sometimes hindered by stops that required manual interference. Depending on the software tool, three to four of these stops occurred during batch processing of the 80 test files. Furthermore, the conversion tools sometimes failed to produce output files even for supported file formats: three (Adobe Pro) up to seven (LuraTech and 3-HeightsTM) PDF/A-1b files were not produced. Since Adobe Pro does not convert e-mails, a total of 213 PDF/A-1b files were produced. The faithfulness of each conversion was investigated by comparing the visual appearance of the input document with that of the produced PDF/A-1b document on a computer screen. Meticulous visual inspection revealed that the conversion to PDF/A-1b impaired the information content in 24 of the converted 213 files (11 percent). These reproducibility errors included loss of links, loss of other document content (unreadable characters, missing text, document part missing), updated fields (reflecting time and folder of conversion), vector graphics issues and spelling errors. Originality/value These results indicate that large-scale batch conversions of heterogeneous files to PDF/A-1b cause complex issues that need to be addressed for each individual file. Even with considerable efforts, some information loss seems unavoidable if large numbers of files from heterogeneous sources are migrated to the PDF/A-1b format.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 3 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.