KeibaTech Originality Check (Beta) — free similarity checker

What are you checking?

Title or file name

Author / student (optional, stays on this page)

Upload one or more files (Word .docx, PDF with selectable text, PowerPoint .pptx, .txt, .md, code files — up to 100 MB each)

Batch (0) — every file you add is kept here; click one to view it 0 words

Text or code to check (paste here, or it fills in from the file you click above)

Optional: exclude the assignment prompt / starter code

Paste the instructions, question text or starter code that every student receives. Matching text is ignored so it does not count as copying.

Also look for matches online (Wikipedia and OpenAlex research abstracts)

Off by default. If you switch it on, up to 6 short phrases (10 words each) from your text are sent to wikipedia.org and api.openalex.org when you press Run check. Nothing else leaves your browser.

Settings

Ignore properly cited quotations

Only a quotation in “double quotes” with a citation next to it, like (Reyes, 2020) or [3], is ignored. A quotation with no citation is still counted and flagged.

Ignore the reference list

Skips everything after a heading such as References, Bibliography or Works Cited.

Shortest match to report 8 words

Matches shorter than this many words in a row are ignored. For code the length is counted in tokens (×3).

Also find re-worded passages

Looks for passages that reuse most of a source’s distinctive words even when order, filler words or a few synonyms change. English only; it can produce false alarms, so always read the match.

Compare with the other items in the batch

Treats the other files you added as sources, so copying between classmates is found.

Sources switched off: 0

Paste some text, upload a file, or open one of the demos above, then press Run check.

Overall similarity

—Not checked

Share of your words that match a source

Identical wording

—

Same words in a row

Re-worded

—

Near match, review it

In quotation marks

—

Matched words, cited / uncited

Sources checked

—

With a match: —

Words counted

—

Excluded: —

Your document with matches highlighted

Click (or press Enter on) a highlighted passage to compare it with its source.

Identical Re-worded Ignored In quotes

Side-by-side

Select a highlighted passage

Ready

Click a highlighted passage on the left to see it next to the passage it matches.

Matched sources

Switch a source off to leave it out of the score.

Privacy notice

What happens to your text

In plain words

This page has no server-side code that receives your documents. Files and text are read and compared in your browser’s memory only.
Nothing is saved in cookies, local storage or a database. Closing or reloading the page discards everything — use the exports if you want to keep something.
KeibaTech does not sell, share or train any model on submitted text, because it never receives it.
The web server that delivers this page keeps ordinary access logs (IP address, time, page requested), like any website. These logs do not contain your text.

The only times anything leaves your browser

When	What is sent	To
You tick “look for matches online” and press Run check	Up to 6 phrases of 10 words from your text	wikipedia.org, api.openalex.org
You press “Check references online”	Up to 20 reference lines (as typed in your reference list)	api.crossref.org
You open a Word, PDF or PowerPoint file	Nothing — the reader code is served from this site	—

Those services have their own privacy policies and see your IP address. If your text is confidential (an unpublished manuscript, a thesis under review), leave both options off.

Using this for a class

Under the Philippine Data Privacy Act (RA 10173), schools remain responsible for how student work is handled. Tell students that their work will be checked with this tool, keep the checks to the legitimate purpose of assessing authenticity, and do not keep reports longer than your school policy allows. This page does not claim to make a school “compliant” by itself.

Clear everything now

Removes all files, text, sources, notes, snapshots and the activity log from this page.

Activity log (this session only)

A simple record of what was done on this page since it opened. It lives in memory only and can be edited by anyone using the page, so it is not a tamper-proof audit trail.

How it works — and what it cannot do

Method and honest limits

KeibaTech Originality Check is an independent beta tool. It is not Turnitin, iThenticate or Crossref Similarity Check and is not connected to them.

1 · IDENTICAL WORDING

Every run of 5 words is turned into a fingerprint. Where the same fingerprints appear in a source in the same order, the run is extended into the longest identical stretch. Stretches shorter than your “shortest match” setting are ignored.

2 · RE-WORDED (NEAR MATCH)

Common words are dropped and the rest are reduced to a root form (with a short built-in English synonym list). If at least 7 of the 10 distinctive words of a source passage — including at least 4 uncommon ones — appear close together in your text, the stretch is marked as a near match. This catches inserted filler words and light synonym swaps. It is a heuristic and can flag two texts that merely discuss the same topic.

3 · CODE

Comments are removed; names, numbers and strings are replaced by placeholders so renamed variables still match; then runs of 10 tokens are compared the same way as text. Starter code you paste in the optional box is ignored.

Known limits

It only finds overlap with the sources it was given. “0% match” never means “original”.
It cannot find text translated from another language. For Filipino, Spanish, French or German text it warns you that only same-language matches are found.
Scanned PDFs (pictures of pages) have no selectable text and cannot be read; there is no OCR.
Word files: text boxes, headers, footnotes and comments are not read. Old .doc files are not supported — save as .docx.
Heavy paraphrasing by hand, or ideas copied without the words, will not be found.
Online lookup covers Wikipedia and research abstracts only. It is not a web search engine.
Very large batches are slow because everything runs in your browser. The limit is 150,000 words per document.

Built-in self-test

Runs the checker on known cases (copy, filler-word evasion, partial copy, quotations, references, code, empty input) and shows whether it behaved correctly. This is a small sanity test, not an accuracy benchmark.

Planned, not available yet

A server-side repository of past submissions (with student consent, retention rules and deletion).
Login, roles and an audit log on the server.
Connection to the KeibaTech class portal.
Matching of translated text.

Credits and licences

Engine v0.9.0-beta — written for KeibaTech.
Word reader: mammoth.js (BSD-2-Clause). PDF reader: PDF.js (Apache-2.0). PowerPoint reader: JSZip (MIT). All served from this site.
Wikipedia text is available under CC BY-SA; OpenAlex data under CC0; Crossref metadata is free to use.

KeibaTech Originality Check

Settings

Your document with matches highlighted

Select a highlighted passage

Matched sources

Who copied from whom?

Comments and rubric score

Quick comments (attached to the passage you last selected, if any)

Comments (0)

Rubric (entered by the reviewer)

Writing profile — descriptive only, no “AI score”

This document

Compare with an earlier sample by the same author

Citation check for the text on the Check tab

Do the references exist? (Crossref lookup)

Write here and keep snapshots

Snapshots (0)

Your sources

Add a source

Sources in this session (0)