Anti-Plagiarism Software Detection?

[Master Razor]

How do they work exactly? I know they check for similar if not the same paragraphs but what about:

• does it also translates the document (language to another language)?
• is it based on a percentage system (say 70% of this paragraph was found)?
• based only on exact paragraphs?
• can pictures be scanned?

I really need your help here. I'm awfully scared about this.

[v6ph1]

A clear answer: It depends!

Some of the programs offer an automatic translation - but not all.
Most of the tools work using a heuristic:

For example:
Searching for sentences with a very small difference.
Take these hits together and if some sentences next to each other are the same like in one of the origin sources, the paragraph will be marked.

Whether pictures are scanned ore not, depends on the professionality and price of the software.

So if I would like to have one, I would choose one with automatic translation, an heuristic approve and a scan function for pictures or image texts.

If you are scared: All programs will offer a score for a closer view.
No one will care if one or two sentences are identical in a large document.

-- v6ph1

[Master Razor]

Right, but the translation is crappy, correct? I mean, if google cannot get it right, neither will them, right?

Say the document is 80 pages with 17.000 words. It would mean a large amount of paragraphs to be marked in order to declare it, correct?

[v6ph1]

Right, but the translation is crappy, correct? I mean, if google cannot get it right, neither will them, right?

Google has not the best translation, but it is quite understandable. You get the relevant words combined to a sentence.
Especially for plagiarism detection, this is enough.
With a good translation you are able to make it harder for the software - not more or less.

Say the document is 80 pages with 17.000 words. It would mean a large amount of paragraphs to be marked in order to declare it, correct?

No - only the most significant paragraphs will be presented.
I meant the possibility of identical or similar sentences raises with the length of the document.

Shot fill sentences without content will be detected more often. - But this is the part, where the manual comes in.

-- v6ph1

[anon]

Or you can ask Strong Bad to write your paper.

Compy 386!

[joe1982]

I believe that the program selects a number of sentences and expressions and looks for them in the other text. For instance - it takes 50 complete sentences and 50 combinations of 4-5 words and scans the other text. If there's a match - there's your plagiarism proof. If it hasn't got a match, it'll take other parts of the source text and rescan for them in the target text. The thing is that proveded you've got a powerful computer and the time to do it - you can scan every sentence from the source text in the target text.

I don't think it could be done in two different languages because a machine would produce a google-translate-like version of the text, losing some context in the process.

I wonder how was it done before the computer age.....

[anon]

I wonder how was it done before the computer age.....

Good question. I imagine the cost of such a service would be much higher.

[v6ph1]

Good question. I imagine the cost of such a service would be much higher.

Indeed.
But in that times, most of professors knew all the literature in their subject. So if they remembered sentences from other books, they just looked into.

Also plagiarism was even harder: Try to copy pages of text without CTRL-C, CTRL-V.

Today, you have more literature then a single person could keep in mind. You can easily copy pages of text.

But you should not be a German politician - they all get busted

-- v6ph1

[anon]

Also plagiarism was even harder: Try to copy pages of text without CTRL-C, CTRL-V.

On the other side, you might actually learn something that way.

Where I live schools still make you do this at times. "Look up data on X and bring it here tomorrow". This worked before the 00s, because you had to read and copy from a book, and would retain some information in the process. But nowadays the purpose is defeated because it's very easy to print tons of text you don't even know what it says and be done with the task.

At this point, I feel compelled to share an experience of mine about how some schools don't give a fuck:

I once had to write a P.E. essay about drug usage in professional sports. They asked me to look up news articles on the matter, summarize, and comment them, then give my overall thoughts on the matter. Since my 14-year-old self was a fan of the "last minute panic" tactic, the day before the deadline I looked up articles vaguely related to the matter (including one about alcohol and road safety), padded the thing with three pages of pictures no one asked me for, then wrote some "drugs are bad, mmhkay" shit. Grade received? 8/10. Success.

[Nobody]

Also keep in mind that larger universities are beginning to keep digital copies of papers in their repositories to check for plagiarism within the school itself.

You seemed panicked, and haven't given us many specific details about your situation. Tell us exactly what you did, then we can more accurately tell you how screwed you are

[Master Razor]

First, I'd like to thank all of you for your valuable input in this thread.
Second, I'll turn in my paper Thursday.

Last year, I've found a paper on the internet about networking. It was pretty advanced stuff (at that time) for me. I started translating it word for word, my own words of course but same images, chapter names and even paper title.

Later on, I became attached of it and started to learn it. I could of made a new one from scratch but I was lazy to do so. Conscience caught up to me and 2 weeks ago I started it from scratch. No worries but I'm very glad I've learned it.

I'm building network typologies now in a matter of minutes and my understanding on the matter greatly increased.

[anon]

I became attached of it and started to learn it. I could of made a new one from scratch but I was lazy to do so. Conscience caught up to me and 2 weeks ago I started it from scratch. No worries but I'm very glad I've learned it.

I'm building network typologies now in a matter of minutes and my understanding on the matter greatly increased.

This looks like a win-win.

Once again I'm sorry I couldn't help with the subject at hand. There's a lot about these things I don't know.

[Master Razor]

Don't worry anon. If you knew, I'm sure you would've helped.

I'm hoping to pass my final exam next week.

[slikrapid]

basically, your (or just about any) paper is a combination of self-written/original material & material from other writers/sources (c/p, should include references) - the more it contains own material, the better and vice versa - if you just have to c/p a lot then do it in such a way that it is hard to be traced back to the original text(s) by using common sense in the process - though as people mentioned before, usually these rules are rather loose, especially if this paper is used in lower levels of education (elementary school,...), whereas if the paper is meant to be used for higher levels (like a scientific paper), the rules are strict, the examiners/reviewers much more careful and your reputation/job/license/scholarship/grants/student position may be at stake - as for any kind of advanced software detection, i'd say most of the checking is done simply via google or searching some internal digitized database or searching manually through written material, but you can probably forget translation matching, image matching, advanced heuristics, maybe also access to global repositories, etc.

I'm hoping to pass my final exam next week.

judging from what you already wrote, it shouldn't be a problem (seems you were careful enough about the paper contents)
anyways, good luck with that