Wikipedia:Village pump (idea lab)

From Wikipedia, the free encyclopedia
  (Redirected from Wikipedia:VPD)
Jump to: navigation, search
  Policy   Technical   Proposals   Idea lab   Miscellaneous  
The idea lab section of the village pump is a place where new ideas or suggestions on general Wikipedia issues can be incubated, for later submission for consensus discussion at Village pump (proposals). Try to be creative and positive when commenting on ideas.
Before creating a new section, please note:

Before commenting, note:

  • This page is not for consensus polling. Stalwart "Oppose" and "Support" comments generally have no place here. Instead, discuss ideas and suggest variations on them.
  • Wondering whether someone already had this idea? Search the archives below, and look through Wikipedia:Perennial proposals.
« Older discussions, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22


Centralized discussion
Proposals: policy other Discussions Ideas

For a listing of ongoing discussions, see the dashboard.


Intelligent Wikipedia that talks to you interactively[edit]

@Rich Farmbrough, Anomie, EpochFail, Headbomb, Beeblebrox, Redrose64, Magioladitis, and Od Mishehu:

Either in voice or text. Or both.

It could start out by saying:

"How can I help you?"

or...

"What would you like to learn about today?"

or...

"What would you like to know?"

You answer its inquery, or ask it a question. It then uses algorithms to access its vast collection of data (Wikipedia, as a corpus, and indices/tables constructed for this project), and provides an answer.

This concept was proposed as a WikiProject to develop something like this way back in 2010, and with the technology advancements that have occurred since then, this project idea may deserve another look:

Dear Wikipedia,

I am Giuseppe (Joseph) Zaccaria (Zachariah), an Italian citizen resident in Iran (my mom is Iranian) born in 19 October 1993, I study mathematical physics in the third year of high school in an Italian school in Tehran. I speak 3 languages fluently (English, Italian, Persian) and have also studied for 3 years Latin and 1 year Spanish. My dream has always been to become a scientist in NASA (although the odds for that are overwhelmingly low), since I love space and anything related to that. I really hope to be able to come to the US, possibly to study.

Now, I use Wikipedia constantly, especially for school projects and studies, and I must say BRAVO!! i think all the students of the world are more than glad to have a site in which you can find almost anything. Nevertheless, each time i have had a question (the answer of which might have been just a few words), Wikipedia gave me a huge array of articles that it thought could be related to what I was looking for, and i must say that that is always a bit disappointing. So, i thought why not add another part in Wikipedia (you know, like wikibooks...) called something like Wikibrain, or WikiAnswers..., where you write a question, and the program automatically searches the articles it thinks might be related to the question, and then finds the answer, without giving you the option to go and look for the answer in the articles yourself. I mean a new program capable of relating words given from the user (i mean the words that are in the question) to other words already existing in the domain (pretty much what it already is doing), and then finding the words in the article (found in the domain) which are most likely related to the primary words that formed the question (given from the user). So, when you ask "distance between Paris and Rome?" the program would automatically find the answer in its articles and say "500 kilometers" (of course this is just an example, I'm sure that if you work on it, it will resolve more complex questions). I am sure that something like this would be a huge step in improving Wikipedia, so I really hope that my idea has interested you and that i might help you in some way. I am really looking forward to receive an answer from you,

Yours sincerely,

Joseph Zachariah

The technology exists for this, right?

What would it take to develop something like this? Would it be any harder than developing WP:AWB, for example? How was AWB developed?

What other features could this thing feasibly have?

What should it be called? WP Genie? Alfred? Jarvis? WikiJarvis? The Transhumanist 20:23, 27 March 2017 (UTC)

You may want to consider just why Wikipedia:Knowledge Engine is a red link—User:GorillaWarfare's summary here is a very good place to start, if you're not familiar with the back-story—for why this is not a rabbit-hole you want to go down. (And no, the technology doesn't exist for this; multinational corporations have spent—literally—billions developing Siri, Google Now and Cortana, and they struggle to answer questions more complicated than "What is the capital of France?".)  ‑ Iridescent 20:29, 27 March 2017 (UTC)
I understand why the idea might be cool, but even if the technology existed, it would be too expensive to implement for a website that, every year or two, has to beg for donations just to remain ad-free (I don't mean "beg" to be an insult). Psychotic Spartan 123 20:36, 27 March 2017 (UTC)
Technology is being developed to do these kinds of things, but as Iridescent said above, it requires immense resources and isn't something we could just sort of "plug in" to Wikipedia. As for your question about whether it would be any harder than developing WP:AWB: yes, considerably. The two are really not comparable.
This is not a project that could be developed by a handful of bot developers; it would require buy-in from the Wikimedia Foundation and the devotion of significant resources. Again, as Iridescent pointed out, this would be a similar move to when the Knowledge Engine project was created, and that's not something I expect the WMF is keen on repeating. GorillaWarfare (talk) 20:42, 27 March 2017 (UTC)
I know the idea lab exists to entertain ideas and not to vote on anything, but it hurts my head thinking about the resources in time and money it would take to do this. Wikipedia's dedicated editors can barely keep up with vandalism, struggle to create quality articles, and I think even hypothesizing on something of this natural is asking a lot. If you think about it, Wikipedia is not a browser, it is a destination. Services like Ceres and Cortana already do this, and under the (basic, usually crappy) answer, a list of sites are given. Among those sites, Wikipedia is usually on the top of the list with an article to answer all the questions on that subject you could possibly imagine (because it, of course, would be a quality article and not a stub). The last thing we need is another project, and another backlog, to add to the list of projects with backlogs that we cannot even keep up with as it is. The idea is cool, but the work it would take far exceeds the return. Just look at all the other backlogs we have and imagine adding another one. Psychotic Spartan 123 21:26, 27 March 2017 (UTC)
Please save the bucket of cold water for the proposal, if this brainstorming session ever inspires one. :) Right now, we're just kicking around some ideas. The most interesting aspect of this exploration (IMHO) is on how much and how could this be automated? It would give interested persons something to apply WikiBrain on. How could that tool be applied to something like this? The Transhumanist 21:36, 27 March 2017 (UTC)

What would it entail? A question generation algorithm, and a database of stock answers? And algorithms to generate the answers in the database? Optionally, editors could provide answers in the database? This is the idea lab -- a place to explore ideas. And so, I'm interested in exploring this idea further. What would such an idea consist of? What would it be built out of? What would its structure be? Any and all thoughts are welcome. The Transhumanist 21:01, 27 March 2017 (UTC) P.S.: Let's brainstorm some more. How about question redirects? (See section below).

One place you can start to see what it would entail is Watson (computer) § Description. isaacl (talk) 05:55, 28 March 2017 (UTC)
What would be needed? Teetering on the edge of The Singularity. In other words, computers that design, build, and program computers that design, build, and program computers that design, build, and program computers...Neonorange (talk) 06:21, 28 March 2017 (UTC)
More generally, it's a leading-edge, high-technology proposal, and Wikipedia is a trailing-edge, low-tech operation. Yes, we'll get this kind of capability, many years after it has been heavily used by Microsoft, Apple, Google etc. Jim.henderson (talk) 19:28, 28 March 2017 (UTC)
I have a feeling, that this is a very farfetched idea and might be implemented to the Wiki years after this proposal. Cheers, FriyMan talk 18:03, 30 March 2017 (UTC)
Strong Oppose. Is this a joke? KMF (talk) 02:11, 8 April 2017 (UTC)

Question redirects[edit]

What about a supplemental set of redirects, which are questions, and which point to the answer in the encyclopedia? The question redirects could be prefixed with "Q:" or something. Such a system could be worked on by editors and bots. Thoughts, more ideas?The Transhumanist 21:06, 27 March 2017 (UTC)

A bot could be created to capture questions asked for which we don't yet have a Q:redirect for, and place it on a backlog list of such question redirects to consider. The Transhumanist 21:10, 27 March 2017 (UTC)
  • I agree, if we end up with all the "agents" belonging to multinationals, humankind's status will be at risk.
  • I further agree that a lot of the technology is available. These are problems I have been thinking about since about 1972/3 and while they are by no means trivial, the apparatus is there.
  • I don't have significant time to devote to this now - the years I was intending to spend on similar Wiki problems were unfortunately disrupted.
All the best: Rich Farmbrough, 21:19, 27 March 2017 (UTC).
(Though the "stock questions and stock answers" machine has already been made by a company in Cambridge.) All the best: Rich Farmbrough, 21:21, 27 March 2017 (UTC).


I tried out some questions in the search box...
"Who is John Wayne?" returned John Wayne at the top of the search results. Why shouldn't this be reprogrammed to go straight to the article instead, and save the reader a point and click?
What is the meaning of life? actually points somewhere. :) Unfortunately, it is just a reasking of the question, rather than an answer. But at least it gets you to the right page.
Who killed Abraham Lincoln? returns Abraham Lincoln: Vampire Hunter at the top of the list.
How far is it to Washington D.C.? gets Save Outdoor Sculpture! as the top result.
The search box currently isn't very good at answering questions, but it answered two of the questions above correctly! Maybe this warrants more testing. The other two may indicate another way to generate semi-random links. :) The Transhumanist 22:05, 27 March 2017 (UTC)
Has anyone working on this talked to the researchers at OCLC about their QuestionPoint[1] product? It does a lot of this already and connects to a real librarian via chat or email (library decides format) when there isn't a stock answer. TeriEmbrey (talk) 14:50, 28 March 2017 (UTC)
Of the four questions you give, one is not answerable by a simple redirect, one already redirects somewhere useful, and two give search results that provide the answer: while Abraham Lincoln: Vampire Hunter is the top search result for "who killed abraham lincoln", Abraham Lincoln is the second result and assassination of Abraham Lincoln the fourth.
More generally, I would like to know more about how this proposal could work. Obviously it would be impractical to create redirects for every conceivable question that is answered by any of our articles. Does wikipedia collect statistics on what the most common search terms are? If it did, could we parse those to find questions? Probably not accurately without a lot of work. How else could we do this? Leave it up to editors' judgements? Is there anything stopping editors creating these kinds of redirects already if they think they are useful? Clearly some of these kind of redirects already exist: see What is the meaning of life?
Alternatively, we could create a set of automatic redirects for a limited subset of possible questions: say, any search in the form "Who is foo?" which doesn't itself exist as an article could automatically redirect for the article for foo. But I can think of only two ways of doing this: 1) use a bot to automatically create these redirects, which would leave us with millions of redirects, many likely to be totally unused, all of which would require editorial maintenance, or 2) patch the search software, which would require persuading a developer that this would be sufficiently useful to be worth development time – and I am not sure that it is... Caeciliusinhorto (talk) 18:55, 28 March 2017 (UTC)
"Is there anything stopping editors creating these kinds of redirects already if they think they are useful?" Yes. The type is not included at WP:POFR, and they get targeted for deletion via WP:RFD, or speedy deleted. It would take a proposal, which is why we're banging the idea around here first. The Transhumanist 22:57, 28 March 2017 (UTC)
The sole work on this that I know of is this very discussion. Thanks for the QuestionPoint link. I had never heard of that. The Transhumanist 23:00, 28 March 2017 (UTC)
"Does wikipedia collect statistics on what the most common search terms are?" I don't know. I've posted your question over at WP:VPT, so we will soon find out. The Transhumanist 23:08, 28 March 2017 (UTC)
See:
https://phabricator.wikimedia.org/T115085
https://lists.wikimedia.org/pipermail/wikimedia-l/2016-July/084745.html
The Transhumanist 00:57, 29 March 2017 (UTC)
"How else could we do this?" Perhaps with WikiBrain? See its homepage on Github The Transhumanist 23:12, 28 March 2017 (UTC)
What is the meaning of life? only exists because it is the topic of an article section. Likewise, Who am I? is the title of many works, and has a disambiguation page. I was thinking that questions would be prefaced with "Q:", in order to avoid clashing with article titles. The Transhumanist 23:17, 28 March 2017 (UTC)
@Caeciliusinhorto: Concerning automatic "redirects", that would most likely be a search engine function. The Transhumanist 23:20, 28 March 2017 (UTC)

Creating a set of redirects is a kludge. It would be better to improve the search engine, which is clearly possible: if I type the first three questions into a very popular search engine, an appropriate Wikipedia page is either the first answer or among the top three results. The last question doesn't seem to be suited to an encyclopedia, but the search engine answers it quite effectively. – Jonesey95 (talk) 23:34, 28 March 2017 (UTC)

But how is the search engine doing it? With a table of redirects? Regex parsing? Something else? The Transhumanist 00:20, 29 March 2017 (UTC)
The search engine uses the redirects and keyword searches of title and content. This is not always fantastic, especially when the search includes a mis-spelling: the engine will prioritise articles with redirects from the mis-spelling, rather than synonymise the word. All the best: Rich Farmbrough, 19:29, 29 March 2017 (UTC).

What's the search engine and extensions used for Wikipedia? The Transhumanist 00:59, 29 March 2017 (UTC)

Wikipedia uses mw:Extension:CirrusSearch. Special:Version has a list of installed extensions at the English Wikipedia. PrimeHunter (talk) 02:00, 29 March 2017 (UTC)
While search engine intelligence is great, redirects are a Chinese room type of AI. Clearly there are context issues which could not be covered well by such a mechanism "Who did John Smith marry?" would do better to redirect the disambiguation page than to any section on any article about a John Smith. All the best: Rich Farmbrough, 19:27, 29 March 2017 (UTC).

Theoretically, Wikidata is just great for this. There is one tool ask (which probably isn't finished), which can answer to some questions. And with SPARQL queries you can query for many things, like distance between Kanzas City and Washington. --Edgars2007 (talk/contribs) 12:57, 30 March 2017 (UTC)

If my Firefox can pass a search argument to Google Search, I don't see why Wikipedia couldn't do the same with site:en.wikipedia.org appended. ―Mandruss  19:26, 30 March 2017 (UTC)

References

Would like to see wikiproject hotels more active[edit]

I feel that articles in hotels are lacking in quality and quantity. Many of them are written like advertisements and wikiproject hotels seems abandoned. I would like to see more quality articles about hotels. I think the way to do that would be to make wikiproject hotels more active, I do not know how to propose this/where to propose this. Admins feel free to move this to where it should be. — Preceding unsigned comment added by Yonikasz (talkcontribs) 02:53, 3 April 2017 (UTC)

I think the most effective way to make a dormant wikiproject more active is to do it yourself. If the former coordinator (or whatever they called themselves) is still active, you might want to drop a note on their talk page. Otherwise, don't wait for permission - start improving articles. Anyone can edit still has some meaning here, after all. Eggishorn (talk) (contrib) 16:24, 6 April 2017 (UTC)
Yonikasz, you can try to WP:REVIVE the WikiProject. You might also be interested in wikivoyage:, which is a travel guide. WhatamIdoing (talk) 19:51, 11 April 2017 (UTC)

Unify Templates in wikidata[edit]

is it possible to Unify Templates and add it to wikidata for all languages. only Variables can be translated. so it will make it easy for newbies, save time and storage space for Wikipedia hard drives.Waso99 (talk) 06:42, 11 April 2017 (UTC)

@Waso99: See meta:Community Tech/Central repository for gadgets, templates and Lua modules. ~barakokula31 (talk) 12:37, 11 April 2017 (UTC)
@Barakokula31: that's great. Thanks Waso99 (talk) 13:58, 11 April 2017 (UTC)

Factual Accuracy indicator in the Revisions[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Wikipedia articles undergo a lot of revisions everyday, and a casual reader going through an article has to re-check to be doubly sure about the veracity of everything he is reading. I propose the introduction of an indicator (ideally a symbol) beside the last revision which has been checked for factual accuracy, in the View History section.

  • Initially we can begin with Good/Featured Articles (which are subject to factual accuracy checks while nomination), and place a symbol besides the revision that granted the GA/FA status indicating that the article has been checked for factual accuracy till that point.
  • Subsequent edits can also be checked for factual accuracy by other trusted users (we may give the privilege to extended confirmed users to check the diffs since the GA/FA and ensure that the factual accuracy is still maintained). These users can then move the symbol to the newer revision that they are authorising.
  • In case the proposal turns out to be successful, we can extend the accuracy indicator to all the articles.

This can help improve the readability and trustworthiness of the encyclopedia. Here is an example - List of Delhi Daredevils cricketers is an outdated Featured List. The list is erroneous and incomplete and has not been actively edited post its successful nomination in 2012. It would therefore be ideal that a symbol be added besides [1] (the FL nomination revision) indicating that the article is factually correct as of this revision. Subsequent editors can continue editing the article, and trusted editors can move the symbol to a subsequent revision, post a fact check of the diff between the revision on which the symbol is currently placed to the current unchecked revision. Jupitus Smart 13:09, 11 April 2017 (UTC)

@Jupitus Smart: Wikipedia:Flagged revisions? --NeilN talk to me 21:07, 20 April 2017 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.