Manuscript checker and PDF converter / Document checks and bulk conversion
10,000 contest manuscripts
turned into judging PDFs
A Windows program that checks and sorts manuscripts submitted as Hangul and Word files, then turns them into PDFs that judges can print right away.
- 10,000Manuscripts processed
- 4File formats: hwp, hwpx, docx, doc
- 5Judging categories
Judges had to read 10,000 contest manuscripts on paper. The files came in four Hangul and Word formats, and some were locked or damaged. Every printout had to be portrait A4 at the original type size with the entry number at the top. The manuscripts held personal information, so no outside service could be used.
How we planned it
- 1
Split what people check from what the machine does
Instead of one automatic run, the work was split into checking and sorting, human review, PDF conversion, post processing, PDF verification, and a final human check. A person confirms the result at every step so that exceptions in 10,000 files do not slip through.
- 2
Use the file name as data
Each file name holds the category, entry date, entry method, personal number, entry number, and title in a fixed order. The tool reads that rule to sort files by category and groups them in batches of 200 by personal number, a size people can review.
- 3
Drop the rule that did not fit
At first we measured empty documents and planned to flag files under a set size as near empty. Real submissions did not separate cleanly by size, so the rule was dropped and only zero byte files are treated as damaged.
- 4
Turn real conversion problems into post processing
Converting real manuscripts produced PDFs with two pages on one sheet, shrunken body text, or landscape paper. We added five post processing steps that find these, split pages, restore the size, and set landscape documents aside for a person to save again.
- 5
Check for missing text with numbers
Finally the tool compares character counts between each original and its PDF and flags anything under 97 percent as possibly missing. People start from a list sorted by risk instead of reading everything again.
How it works
- Step 1, check and sort: locked and damaged files are set aside and valid manuscripts are copied into category folders. Originals are never changed.
- Step 2, convert: each file opens in Hangul or Word, is saved as PDF, and gets the entry number at the top center. If the number would overlap the text, it is left out and the file goes to a separate folder.
- Post processing: two up PDFs are split, shrunken PDFs are restored, and landscape documents go to a manual list.
- Step 3, verify: character counts, broken characters, and page differences point to files that may be missing text.
How we built it
- It runs only on a Windows PC and never sends manuscripts anywhere.
- Running it again skips PDFs that already exist and picks up where it stopped.
- Every result log is named by time, task, scope, and counts, so the result is clear without opening the file.
- A manual and a checklist per category let a new staff member go from setup to troubleshooting alone.
Results
- 10,000 manuscripts were checked by category and converted to judging PDFs.
- People review only the exceptions and the flagged list, not every file.
- The PDFs made here become the source for the next tool, the pass entry tool.
See if it fits
how your team works
Tell us how you operate today and we will propose where ilnetta fits. Reach us on KakaoTalk @ilnetta.