Tortured conference series thriving in computer science | Hacker News Reader

Tortured conference series thriving in computer science

hal.science

72 pointswjb325 comments

Apparently much of the IEEE trouble is ICERECT[1], which as best I can tell, is primarily for papers rejected elsewhere. Filtering the dataset from the article[2] down to just this one paper[3] from that conference helps illustrate the point.

The suggesting elsewhere on thread that this could be LLM driven seems unlikely -- "bosom disease" is not a common pairing of words in the English language corpus, but "breast cancer" is.

[1]: http://www.pesceconference.in

[2]: https://dbrech.irit.fr/pls/apex/f?p=9999:24

[3]: https://pubpeer.com/publications/2A7F7A96A9E42D48FCAA5DA6548...

I think this can result from cases where paraphrase isn't useful but still culturally expected. For example, you often need to introduce the central point of another paper before you can comment on it, and the ideal way to do that would be to insert the whole abstract verbatim, but people look down on that. So, you end up writing a paraphrase as faithful to the original as possible while still being superficially different. If you're good at English, you can paraphrase in a way that flows well and displays your understanding. ESL writers might be at risk of screwing the wording up, not noticing, and getting ridiculed for poor research.

Inserting the abstract verbatim is very far from ideal. That would usually cause you to repeat a lot of information from your own paper's introduction and then fail to hone in on the precise reason you're citing the paper.

Summarizing another academic work in the context of your own work is a bedrock activity in academic thinking as well as writing. Nobody should rely on a tool to do that.

I admit a scenario in which you have to communicate everything that the abstract communicates is kinda contrieved, though it could happen if the article is the main target of your discussion. But in general, pointless paraphrase of other sources is widespread.

>Summarizing another academic work in the context of your own work is a bedrock activity in academic thinking as well as writing.

Summary has its place, but paraphrase is another thing. Quality of paraphrase is defined by surface-level features like vocabulary, grammar, and re-arrangement of sentences, and second-language English education often fails to advance beyond word-by-word substitution. Even when the exercise of paraphrase demonstrates some kind of understanding of the original text on the part of the author, it doesn't necessarily aid the reader.

I'm not trying to deny that tortured phrases may also indicate paper farming and plagiarism btw, just saying that there are other factors.

I think the term here is plaigarism. Changing random words and shuffling clauses around so the auto checker doesn't flag it while keeping the sentence by sentence structure of someone else's work.

Relatedly, https://en.wikipedia.org/wiki/Article_spinning has been quite common for decades now.

These terms sound like they were mistranslated from another language

A web search for tortured phrases computer science papers suggests that this problem was first widely noticed and discussed in 2021. The reason seems to have been the use of machine translation from languages like Chinese and, maybe, writing or paraphrasing by GPT-1 or -2 level LLMs.

With the current widely available models, it is much easier to produce fraudulent papers that cannot be caught just by searching for unnatural expressions.

that actually makes sense. thanks.

> The reason seems to have been the use of machine translation from languages like Chinese

No, it's more deliberate than that. Per the original paper [1], the tortured phrases observed in these papers can be reproduced by running text through an online tool called "Spinbot" which processes English text by replacing words with (frequently incorrect) synonyms. Many of the replacements made are difficult to explain any other way -- replacing "ant" with "underground creepy crawly" is clearly not the work of a translator, for instance.

[1]: https://arxiv.org/pdf/2107.06751.pdf

Thanks for the clarification!

I used to tutor MSc-level students and they would use this tool all the time for their essays; copy-paste a paragraph from Wikipedia, run through the tool, copy-paste. Just absolute bottom laziness. (The students all got their degrees; welcome to the age of the-student-is-the-customer)

Is this supposed to just be a 2 page PDF, am I missing something? Kind of ironic to call out bad research with such a sparse paper.

Using “grouping calculation” for “clustering algorithm” is absolutely hilarious to me.

This feels like something an LLM would be good at producing and also catching.

Indeed, humans seem to have produced a good amount of data for this task :^)

Nah, this has more symbolic NLP vibes. LLMs mess up in subtler, more uncanny ways.

I wonder if one could use raw data like this to build a sort of spam filter for papers. Perhaps there are common metadata (institution?, date?, time?, name?, topic?...) among BS papers that could be part of a classifier score?

I know that a Proper fix to enshitification is to get the human beings to stop doing it, but I can't help but wonder if there's something I can do on the "receive" side to have a better experience.

Some of these are really nice:

- Tired of having to use "random forest"? Why not "irregular woods" to confuse every single one of your colleagues. "Random woodland" also sometimes comes up apparently.

- When writing a neural network, some layers are usually hidden. But why not mix it up and use "concealed layer" instead?

- Artificial Intelligenge an overused term? Why not "man-made intelligence"?

- Denial of Service attacks are so 2022. The modern thing is "disavowal of administration assaults".

- Tired of using Fast Fourier Transforms? Want to be seen as doing Quantum Fourier Transforms? Why not rename the former to "Quick Fourier Transform" to easily get the abbreviation you want!

- My personal favorite: switching out "big data" for "huge information".

plus-sized tea

> big data

fat geordi

>> big data

> fat geordi

massive worf

Enormous crusher

Plump picard, rotund riker, dumpy data. Might as well get some alliteration going on while we're at it.

"Random woodland" - I would propose an "Erratic woodland".

"Concealed layer" - interesting as Google automatically transforms it into "hidden layer" in search results but when I insist it to use the former then it would turn up some papers. But I like it and would actually use it to enrich the writing.

"Man-made intelligence" - is so lovely and true, I should use it too.

"Huge information" - you can wish but it's only data most of the time.

Reply on news.ycombinator.com