LINGUAX ARC / TEXT PREPARATION

A clean start for your corpus.

Clean books, theses, articles, newspapers and OCR text with explicit rules, keyword controls and a visible audit.

✧

Add files, paste text or try the sample to begin. Everything is processed on this device.

02 / PREPARE

Choose what to clean.

Start with a preset

Normalize Unicode (NFKC) Join hyphenated line breaks Remove page-number-only lines Remove repeated PDF page edges (3+ pages) Remove trailing references / bibliography Remove numeric / author-year citations Remove front matter before Abstract Remove standalone news boilerplate Strip common HTML markup Remove exact duplicate paragraphs within each file Remove email addresses Lowercase text Remove URLs Remove digits Remove punctuation / symbols

Books, layout and repetitive text

Remove contents headings / dotted-leader entries Remove numbered chapter / part headings Remove leading line or list numbers Remove standalone bylines / date labels Reflow wrapped prose lines Collapse adjacent repeated English words

Heading, byline and line-number removal can remove real content. Reflow is for wrapped prose, not poetry, tables or transcripts. Deliberate repetitions can carry meaning.

OCR damage and encoding repair

Remove invisible soft hyphens Expand Latin ligatures and long s (ſ → s) Remove replacement characters and box placeholders Remove non-printing control characters Remove zero-width / BOM noise Repair known broken UTF-8 punctuation Remove symbol-only scan-noise lines Remove private-use font codes (missing glyphs)

Repairs known encoding patterns and removes selected placeholder characters. It cannot recover missing letters, guess damaged words or fix a PDF's font mapping. Box symbols and private-use codes can be meaningful: review the audit and original.

Remove your own words or phrases.

Keywords to remove · one per line

Whole words / phrases Remove entire matching lines

Literal matches, never a custom regular expression. Up to 200 entries, 200 characters each. Case sensitivity applies to this list and the find/replace fields below.

Find literal text

Replace with

Whitespace and paragraph spacing are normalized on every run. Reference, citation, front-matter and boilerplate removal use English heuristics. Review every result: these rules can match legitimate content. Duplicate removal works within each file; the separate duplicate check also compares files.

Your results will appear here.
Review first. Download when you are ready.

03 / REVIEW & DOWNLOAD

Your cleaned copies.

Output document

Download as

Original extracted text

Output preview

Preview shows the first 12,000 characters per document. Downloads contain the full output. Word exports contain text and paragraphs only.

REVIEW / REPEATED PASSAGES

Duplicate paragraph report.

Working with an image-only book? Open Arc English OCR first, then import the extracted text here.

All toolkits Corpus analysis Privacy Feedback