# Turn a Scanned Book into Markdown Research Notes You Can Verify

Convert a scanned book to Markdown, separate chapters from your notes, retain page references, and verify quotations before using extracted text in research.

A folder full of recognized text is not yet a set of research notes. The useful next step is to preserve the distinction between what the book says, what OCR recovered, and what you think about it.

For a long scanned book, Markdown makes this practical: you can edit text, divide it into manageable files, and link notes to source passages. Keep the PDF as the authority for quotations, page references, tables, and uncertain words.

## Start with an untouched source and an untouched extraction

Use [PDF to Markdown](https://pdfcraft.ai/en/pdf-craft/pdf-to-markdown/) to upload the scanned PDF, choose Markdown output, and download the completed conversion. If you are researching the original text, leave translation off. You can translate material separately when needed, but mixing translated wording into a source transcription makes later citation harder.

Save the downloaded result before editing it. A simple local folder can distinguish the stages:

```text
book-project/
  source/book-scan.pdf
  extraction/original-output.md
  working/chapter-01.md
  working/chapter-02.md
  notes/themes.md
  notes/quotation-log.md
```

This is a suggested folder arrangement in your own storage, not a PDF Craft export feature. Keep any downloaded image resources with the working Markdown and check relative paths when moving files.

First inspect headings, paragraph order, a difficult page, and a passage crossing a page boundary. OCR may recognize the words while reconstructing the order incorrectly. Do this before distributing the text among dozens of notes.

## Split by chapters before splitting by ideas

Use the source book's chapter boundaries as your first divisions. Keep chapter titles and section headings so you can navigate back to the argument surrounding a quotation. Do not immediately turn every paragraph into an isolated note: isolated passages lose qualifications and make repeated claims harder to interpret.

In each chapter file, add a short header with the book title, edition, chapter, and source page range. The range can be provisional until checked. Then write a separate note summarizing the chapter in your own words, with links or clear references to the relevant working files.

For example, a theme note on railway finance might refer to chapter 3 for the author's claim, chapter 6 for the counterexample, and an appendix for figures. Those references let you challenge your summary later instead of treating your first interpretation as settled.

## Preserve page clues without inventing precision

Printed page numbers and PDF page positions are often different. A scan may include a cover and unnumbered front matter before printed page 1. Record both when possible: “printed p. 73; PDF page 89.” Check the mapping again at later chapters because missing pages or inserts can break a single offset.

If the extraction does not contain reliable page markers, add verified references manually at passages you actually intend to use. Do not insert a page number by estimating from paragraph length. Reflowed Markdown is not paginated like the PDF.

A quotation log can use four columns:

| Passage                                     | Source location                     | Verification                             | Your use                             |
| ------------------------------------------- | ----------------------------------- | ---------------------------------------- | ------------------------------------ |
| Exact quotation or short identifying phrase | Edition, printed page, PDF position | Checked against scan, or not yet checked | Claim it supports and any limitation |

Keep the wording “not yet checked” visible. A copied extract should not become a verified quotation merely because it has been moved into a polished note.

## Review before quoting or calculating

For a quotation, compare every word, punctuation mark, and omission against the page image. Check names, dates, negations, and line-ending hyphens especially carefully. Mark illegible material as uncertain rather than filling it in from what would make sense.

For a table, retain its title, row and column labels, units, and notes. Reconstructing a list of numbers without those labels destroys the information needed to interpret them. For a formula, check operators, subscripts, superscripts, and equation references before using it in a calculation.

Footnotes deserve the same treatment. Confirm which marker belongs to which note and whether a citation continues onto another page. A correct note attached to the wrong sentence can misrepresent the author's evidence.

## Keep your interpretation separate from the author's words

Use clear labels such as “Source extract,” “My summary,” and “Open question.” Put exact quotations in quotation marks or blockquotes, with their verified location. Write paraphrases in your own words and still record the source.

When a note changes your interpretation, update your summary rather than silently changing the retained extraction. This small separation makes later review much easier: you can distinguish an OCR correction from a change in your argument.

For a broader document workflow, see [PDF to Markdown for research](https://pdfcraft.ai/en/blog/pdf-to-markdown-for-research/). If your immediate goal is reading the book rather than editing notes, [scanned PDF to EPUB](https://pdfcraft.ai/en/pdf-craft/scanned-pdf-to-epub/) may be a better output. Review current [Pricing](https://pdfcraft.ai/en/pricing/) before processing or repeating a large conversion.