Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

online.wsj.com

Post reply on HN
Print view View original

41–50 of 78 posts

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#41

Post by jodiscr »

_delirium wrote: post #30
dangayle wrote: post #15

I would rather have a bot written stub than nothing at all when I search in Wikipedia. Like it is mentioned in the story, some subjects are under represented while others are super saturated.

If Wikipedia were the only website I'd agree, but I don't find it useful for Wikipedia to have stubs that are simple copies of other freely available sources (especially more authoritative ones), without some kind of synthesis or value-add. In the case of species, for example, I think there is little value in a Wikipedia stub that is just a reformatted copy of the ITIS entry ( http://www.itis.gov/ ). If that's what I…

> Otherwise it feels like some of that autogenerated SEO-style spam, which Google should penalize.

I assumed Google already recognized these near-duplicates and are penalizing them

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#42

Post by vacri »

baddox wrote: post #33
vacri wrote: post #31

Earlier quoted context omitted.

These are (almost) one-sentence articles. They're not appropriate for an encyclopaedia. If anything, they should be in a 'list' article format. If you want these kind of one-sentence descriptors, they would be better served in a specialist publication.

If one-sentence articles are considered a problem, that seems like a reasonable choice for Wikipedia to make, but it would apply to human-written and bot-written articles.

It does apply, regardless of the source of the article. Wikipedia doesn't like stub articles.

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#44

Post by tedks »

gwern wrote: post #29
dkhar wrote: post #3

I'm not very in-tune with Wikipedia's culture (which, I've read, is very nuanced and rigid[1]), but I really don't see why this is a bad thing, given the information in the articles is accurate (and the article gives the impression that glitches are rare). If nobody else was going to create an article about some species of butterfly, I don't see why adding that information would be harmful to Wikipedia. Does it make…

At least part of the problem is that he's generating what one might call 'info trash': he's taking highly structured information from databases, and turning it into natural-language prose, a data source of less value since it's less structured. These prose versions are now going to steadily fall out of sync with the original databases, be much more prominent in Wikipedia and Google, diverge from each other, be harder…

>It would have been better to work on merging databases or exporting them into a structured site, something like Freebase.

Why is this not appropriate for Wikidata?

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#45

Post by frik »

vacri wrote: post #42
baddox wrote: post #33

Earlier quoted context omitted.

If one-sentence articles are considered a problem, that seems like a reasonable choice for Wikipedia to make, but it would apply to human-written and bot-written articles.

It does apply, regardless of the source of the article. Wikipedia doesn't like stub articles.

Most articles started as stubs. At the moment many language versions of Wikipedia have different opinions about stubs.

e.g. german wikipedia don't allow stubs anymore and the admins delete & reverts more pages every day than new ones are created. It's maybe a cultural problem as such admins identify themself with 'their articles' and don't allow any changes.

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#46

Post by urizen »

Crito wrote: post #17
nmrm wrote: post #5

Earlier quoted context omitted.

Reading the article, I think some of it is more of a "why?" than a "why not?", since the articles are factually correct but are mostly just lists of facts and not really something you couldn't find in any number of other resources. For people who do serious article writing, I imagine this might be considered as a "cheapening" their work. For instance, I imagine some editors also resent the notion that encyclopedia ed…

> "and not really something you couldn't find in any number of other resources." The same should be able to be said about everything on Wikipedia, since Wikipedia is not supposed to have original research and should have a source for everything. The "Why?" would then be "Because it is better to search for [obscure butterfly] and find a short list of fact than to search for that butterfly and find nothing at all."

Problem is that for any obscure organism you'll need to know what it is before you can search for it on WP. There is no ability to do that on WP, if the articles on the species has an image the chances are good that it will be a related species. Then there is the problem of some one coming along and adding mangled 'facts' to the article or 'facts' derived from 19th century works.

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#47

Post by dangayle »

Crito wrote: post #18
dangayle wrote: post #15

I would rather have a bot written stub than nothing at all when I search in Wikipedia. Like it is mentioned in the story, some subjects are under represented while others are super saturated.

Exactly. If nothing else, those stub articles serve to give a (not complete, yet more accurate) index-able list of topics not yet adequately covered. Like a /* TODO */

That's exactly the point of the stub articles isn't it? They specifically mention that you should improve it by adding relevant info if possible.

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#48

Post by zobzu »

dkhar wrote: post #3

I'm not very in-tune with Wikipedia's culture (which, I've read, is very nuanced and rigid[1]), but I really don't see why this is a bad thing, given the information in the articles is accurate (and the article gives the impression that glitches are rare). If nobody else was going to create an article about some species of butterfly, I don't see why adding that information would be harmful to Wikipedia. Does it make…

I'm completely pro-bots but it does seem this one writes way too short articles with very little content. Bots need to do a similar-enough job to humans while being fast and reliable.

Not spam the wiki with names and basically no info.

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#49

Post by nmrm »

Crito wrote: post #17
nmrm wrote: post #5

Earlier quoted context omitted.

Reading the article, I think some of it is more of a "why?" than a "why not?", since the articles are factually correct but are mostly just lists of facts and not really something you couldn't find in any number of other resources. For people who do serious article writing, I imagine this might be considered as a "cheapening" their work. For instance, I imagine some editors also resent the notion that encyclopedia ed…

> "and not really something you couldn't find in any number of other resources." The same should be able to be said about everything on Wikipedia, since Wikipedia is not supposed to have original research and should have a source for everything. The "Why?" would then be "Because it is better to search for [obscure butterfly] and find a short list of fact than to search for that butterfly and find nothing at all."

> The same should be able to be said about everything on Wikipedia

I don't think so. Well-written encyclopedic entries are in far shorter supply than bare lists of facts.

I guess the fundamental difference of opinion is between those who feel Wikipedia is an encyclopedia, and those who feel it's a dumping ground for human knowledge. Note that I'm not taking sides, just trying to explain the root causes for the difference of opinion.

Also, on a purely technical note, I very much doubt that you couldn't find the information in bot-generated articles anywhere else using a search engine. If that were the case, where are the bots getting the data?

Re: In Sweden, Sverker Johansson and His 'Bot' Have Created 2.7M Wikipedia Articles

#50

Post by alkonaut »

Some of these are borderline useless, but I can definitely see the point in bots pulling structured data into templates IF the data is rich enough to gain from being in human-readable format OR if the topic is noteworthy enough to warrant a stub, that is, it is likely to become a hand written article in the future.
Post reply on HN