Skip to content
Documentation
View as Markdown

Optimising files for full-text search

Full-text search lets your colleagues find the relevant specialist literature quickly. For it to work, the uploaded files must contain searchable text.

Librario searches the text of PDF files, EPUB e-books, Word, Excel and PowerPoint files, OpenDocument files and plain-text files (such as .txt, .md or .csv). We recommend PDF files, though, because they:

  • Work on all common devices (laptop, tablet, smartphone)
  • Display in high quality
  • Compress well
  • Are easy to share with others
  • Are suitable for long-term archiving as PDF/A

Common problems and solutions

Problem 1: Scanned documents without text recognition

Symptom: The file contains only scanned page images, with no searchable text. Librario shows the message “This file contains no recognisable text. Full-text search is not possible.” If the file contains only a little text, such as headers or stamps, the message reads “Recognised text is short. Please check the file.”

Solution: Run text recognition (OCR):

  1. Use an OCR program such as Adobe Acrobat Pro or ABBYY FineReader (there is a step-by-step guide for Adobe Acrobat Pro)
  2. Check these settings:
    • Choose the correct document language
    • Use high recognition accuracy
    • Turn on automatic image preprocessing
  3. Spot-check the result for errors
  4. Upload the searchable file to Librario again

Problem 2: Faulty text recognition

Symptom: The recognised text contains many errors or garbled characters. You notice this when you copy text from the PDF or when the search returns unexpected results.

Solution: Depending on how many errors there are, you have two options:

  1. With a few errors:

    • Use your program’s OCR correction
    • Correct problem words by hand

  2. With many errors:

    • Remove the existing OCR text layer
    • Run text recognition again
    • Make sure the original is scanned in good quality

Problem 3: DRM-protected files

Symptom: The file is protected by digital rights management (DRM). Librario cannot extract the text.

Solution:

  1. Ask your provider for a DRM-free version
  2. Check whether your licence allows other forms of use
  3. For e-books: some providers offer special library licences

Problem 4: Password-protected PDF files

Symptom: The PDF file is protected by a password.

Solution:

  1. Open the file with the correct password
  2. Export it as a new, unprotected PDF
  3. Most PDF programs offer a “remove security restrictions” option for this

Best practices for scanning

When you scan new documents, follow these tips:

  1. Scan quality:

    • At least 300 dpi for text documents
    • Greyscale or colour, depending on the original
    • Pages straight, not skewed
    • Use clean originals

  2. OCR optimisation:

    • Choose the correct document language
    • Turn on image preprocessing
    • Use existing OCR profiles for your document type
    • For multilingual documents, select all relevant languages

  3. File optimisation:

    • Compress the files after text recognition
    • Check the quality of the recognised text
    • Save important metadata such as title and author
    • Remove blank pages

OCR software

There are several OCR programs on the market. For example:

Quality check

Spot-check the quality of the files for full-text search:

  1. Pick a few distinctive terms from the document
  2. Search for them in the PDF file
  3. Check that the matches are correct
  4. Also test terms with accented characters (such as ä, ö, ü or é) and other special characters

If there are problems, run text recognition again with adjusted settings.