Help talk:Searching/Archive 2

Revise search page

Since this "Searching" page is a prominent link off the proposed main page redesign, the quality of this article needs major improvement. I have looked at other languages and find the de:Hilfe:Suche page on the German Wikipedia useful in its simple, concise layout.

From that, I am proposing here that:

1. We cut down on the wordiness, and be concise.
   1. We could always split off more detailed information into sections such as Wikipedia:Advanced search, Wikipedia:External search engines, etc.
2. I've added a TOC infobox to the page, which includes the all the primary links used in the proposed main page, as well as some key policies. At some point, it might be useful to have this infobox (improvements welcome) on all these key pages listed there, to make it easy to navigate this important information.

Any further ideas or suggestions? -Aude (talk | contribs) 21:04, 14 March 2006 (UTC)

I'm thinking it might be easier/better to just link to Wikipedia:Look it up from the main page, and then refine this page into an "advanced search help" page. (The "searching" link was added to the MainPage redesign at the very last minute, i forget by whom) --Quiddity 22:30, 19 April 2006 (UTC)

Dynamic spelling correction?

I have been looking for a place to suggest this to someone, but I really don't know the proper place, so I thought I'd mention it here. I know that (apparently) wikipedia search engine seems to look first for an article or page of the same title, and take you there; if that fails, it will search for that/those term(s) in other articles and display articles that contain the word(s), with a small link at the top if you want to start a new article.

I'm curious if anyone has ever thought of, or considered the implimentation of a suggestive spelling error system for searches - particularly names of places or people suffer from this issue. If you goto dictionary.com and look up a word that is misspelled or not in the database, the result is a list of terms that you might have misspelled (say, for example 'enthousiasm' results in a list starting with 'enthusiasm, enthusiasms, enthusiast...'). Similarly, yahoo.com, if you search for a term with low results, will attempt to suggest a correction (Did you mean: enthusiasm ?). I often find myself forced to pull up a yahoo window, search for the name I'm looking for, get the suggested correct spelling, and then bring it back to wikipedia so I can find who or what I'm looking for. Would it be possible to impliment a similar suggestive system to the top of a failed wikipedia search? And/or, how would I go about suggesting this properly to the proper channels? Thanks TheHYPO 06:03, 26 April 2006 (UTC)

• I was thinking about this as well, and was about to post a new discussion before I stumbled upon this one. I agree with what TheHYPO said, as I do tend to mispell some of my searches--those that I'm unclear about (thus my reason to search it on Wikipedia). Google also uses this feature as you can see here: a search on "snkae". I hope that this issue can be resolved quite soon. Thanks.

Please see Wikipedia:Village pump (perennial proposals)#Better search feature. This has been a requested feature for quite some time, but the current stance seems to be that it can't easily be done without adversely affecting performance. -- Rick Block (talk) 15:50, 13 August 2006 (UTC)

At the moment, the search engine uses an index that isn't updated at all.

What's up with that? This statement should really be clarified. Why is the search engine not updated? How are we supposed to make sure that we don't create the same articles over and over (with slightly different names) if we can't find them?--345Kai 09:26, 7 May 2006 (UTC)

I cut that statement, which is badly outdated as "at the moment" refers to three years ago [1]. Please point out anything else that isn't clear, seems outdated, and feel free to just go in and fix things if you like. -Aude (talk | contribs) 00:33, 25 May 2006 (UTC)

YES.....finally some1 that is as bad at spelling as i am. i usally end up typing a word into google which corrects it for me and then searching for it in wiki. we should send and email or sumthing to make it so that there it is a spell checker. i think that there might even be a way to make google check the spelling for you. this is the ONLY problem that i have with wiki right now and if it is fix this will be the best site ever. —The preceding unsigned comment was added by 75.4.168.17 (talk • contribs) 00:22, 25 May 2006 (UTC).

It looks to me like the index is again not being updated. As a "for instance", the article for Olin E. Teague was created 30 June, 2006 and as of 26 July, 2006, it's still not findable via search. A week stale is annoying, but given the mutation rate understandable and not so critical. A month stale hurts... --studerby 16:02, 26 July 2006 (UTC)

Still true, as of 31 August, 2006. studerby 02:38, 1 September 2006 (UTC)

Cannot Search Old Revisions

I wanted to run a search that doesn't just look at the current version of a page. I want to run a search that also looks at older versions of a page. I've been researching some vandalism, and would find such a seach capability helpful. -- fsk

This edit section could be moved to the bottom or deleted. User:fsk never examined the history or diff mechanisms.
Sadly this answer may never reach hirm. · Ken 15:48, I can't find wat i'm looking for this website is CRAP

Is it just me or the search isn't working properly? I tried to search "nip it in the bud" and it gave results where the words didn't appear in order. I was about to start an article on the idiom, but then decided to look for a list, and voila. List of idioms in the English language#N. Even if I had seached without the quote marks (which I did previously), shouldn't it give me the more relevant results first? Like one containing the full searched expression as asked? VdSV9•♫ 11:30, 10 June 2006 (UTC)

The WP search engine is terribly simple. It's a known problem. The few developers who might know how to improve it are constantly busy with more critical bugfixes. Last i saw, they don't envision getting around to improving it soon. (google is your friend ;) -Quiddity 17:14, 10 June 2006 (UTC)

On a personal note, I hope they would be able to fix this as soon as possible. --Siva1979^{Talk to me} 18:24, 7 July 2006 (UTC)

I was searching for "willig", and it returned results with "willige" and other variations. Is there a syntax which I should be using to indicate "this exact text with no wildcards"? Thanks! --Keeves 15:42, 14 November 2006 (UTC)

Stop words

Lucene still doesn't search for numbers. I've added this to the article. --Eleassar ^{my talk} 11:03, 17 June 2006 (UTC)

searching for contributions

Is it possible to search for contributions of others ? —The preceding unsigned comment was added by 82.232.235.239 (talk • contribs) .

Search searches the entire content of Wikipedia (as of the last time the index was rebuilt), which includes the contributions of everyone. If you're looking to see someone else's contributions like you can see your own by clicking "my contributions", when you're on their user or talk page there's a "user contributions" link in the toolbox or you can enter (or create an internal link to) special:contributions/username. -- Rick Block (talk) 13:21, 4 July 2006 (UTC)

Search not finding article

I tried to search for Ecclesiastical heraldry (to find out how long the article is), but the article with that title is not in the list results. Should not articles matching the exact search phrase be first or very early in the list? Gimmetrow 16:44, 28 July 2006 (UTC)

The article was created on June 23. The search index has apparently not been rebuilt since then. When it is, the title will appear in the search result (probably first). -- Rick Block (talk) 18:41, 28 July 2006 (UTC)

OK that makes sense, but I expected the search index rebuilt more often than that, weekly or at least monthly. Gimmetrow 18:53, 28 July 2006 (UTC)

Vandalism?

Has this article been vandalised?

Yes, this article has been heavily vandalised, please fix it. why doesn't wikipedia have a very simple button somewhere to report when an article has been vandalised? —The preceding unsigned comment was added by 71.240.163.61 (talk • contribs) .

Link to a specific Google search

I added a sub-section: How to provide a link to a specific Google Search. This page was linked from Wikipedia:Tip of the day/August 7, 2006 but didn't actually have information on using [[google:XYZ]].

A minor issue with the software, re the approach of using a plus sign, +, to replace spaces in the search field... it works fine in Wikipedia, but when I tried to do it at meta.wikimedia.org and at Appropedia.org (also MediaWiki) it didn't work. Word or phrase searches worked fine, but as soon as I put in a plus sign, it didn't create a link, but just showed the [[google:XYZ+ABC]] in nowiki format. It still has the same problem that it turns spaces into underscores (which hopefully can be fixed at some point...) --Singkong2005 ^talk 00:12, 16 August 2006 (UTC)

I've removed the suggestion that such a link might be acceptable in the external links section of an article. It's not acceptable - search results change all the time, and a search that returns more than one result forces the reader to (in theory) look through all the results to find something useful. John Broughton | ^(♫♫) 15:18, 19 January 2007 (UTC)

Varè, Daniele

Daniele Varè (1880-1958) was an Italian diplomat and author. During the 1920's he was the Italian ambassador to China. His best known books were the trilogy known as "The Novels of Yen-Ching". These were set in Peking during the 1920's, and carry the intriguing titles of "The Maker of Heavenly Trousers", "The Gate of Happy Sparrows" and "The Temple of Costly Experience". He also wrote several volumes of memoirs, including "The Laughing Diplomat" and "The Two Imposters" and a biography of the Chinese dowager empress, Tzu-Hsi.

apostrophe/single quote

Unless I’m mistaken, the reason for this behavior is so apostrophes will be recognized as part of a word. Wouldn’t it make more sense to mention that, then say single quotes are identical to apostrophes, rather than vice versa, and that’s why single quotes should be avoided? As it is, it just reads as a stupid bug (no offense to WP/WM developers). —Frungi 01:42, 24 August 2006 (UTC)

So when will the search database be updated?

Having had many frustrating experiences using the search function to find out whether articles exist (and it not returning results for articles which I know exist), I came here to find why it wasn't working properly. I understand from answers to similar complaints above that what is at fault is that the database that the search function uses is rather old and infrequently updated, and so more recently created articles don't appear. So does anyone know the answer to these questions:

• When was the last update to the search database made?
• When will the next update to the search database be made?
• At what point does the gap between updates become so long that this is considered an urgent problem, which if left unaddressed undermines the effectiveness of the search tool entirely? (OK, I admit this last one is a bit loaded - but I did say it was frustrating.)

To put this in context, currently the search button doesn't recognise the article on James Alipius Goold, which was created on the 22nd of May, representing a lag of at least three months and counting (in which time something like 300,000 articles have been created). Thylacoleo 07:27, 31 August 2006 (UTC)

There's an existing bug request to add the ability to find out the last time the search index was rebuilt, see bugzilla:6090. If you have a bugzilla account, voting for this bug is a way to increase its priority. I've suggested the search page include a "results as of <last index build time>" indication as well. I believe the answers to your questions are: only the developers know, no one exactly knows since it's irregularly updated when some developer (I'm not sure specifically who) manually initiates it, search is considered to be a secondary feature given that reasonable alternatives exist (e.g. google). -- Rick Block (talk) 13:48, 31 August 2006 (UTC)

Thanks for the answer, although I'm afraid I don't have a bugzilla account, not being particularly proficient in technical IT matters. I suppose my main problem is that I don't like to leave Wikipedia to use Google when editing. I am a bit baffled that there is an attitude that the Wikipedia Search function can be allowed to atrophy to such an extent because there exists an external site which can do it instead. Especially as the Search function features so prominently in the side bar on every single Wikipedia page - it seems reasonable that a casual reader would presume that it is up to date and useful. Is there a suitable place where I can go to clamour for a more serious approach to the Wikipedia Search function? Wikipedia:Village Pump (technical) perhaps? (Hmm, it seems they already make brief mention of it in their topbox, noting "The search index is often out of date, sometimes taking weeks before it's updated. Because of that, recent changes are not immediately reflected on the search." It might be worth a try raising it again there, anyway.) Thylacoleo 01:23, 1 September 2006 (UTC)

I think the problem is well understood and appreciated. I believe the issue is that the development staff is mostly volunteer (there are something like two paid developers) for a site that is in the top 20 (by traffic) web sites in the world. I agree that there's really no excuse for the search function to have such low priority, but the reality is problems that aren't critical have a hard time popping up enough on the priority list to get any attention. Can you keep another Window open? What I do is keep a Window open with "site:en.wikipedia.org" filled in the search box in google (this restricts google's search to this site). If it were up to me (it's not), the "search" function here would do this by default. -- Rick Block (talk) 02:52, 1 September 2006 (UTC)

Merge with Wikipedia:Look it up

I strongly believe that we need a cross-between a simple version of this page, and an expanded version of Wikipedia:Look it up, as the main Search Help page, that gets linked to from the Main Page and similar. The exhaustive list of options and details should be on a supplementary/sub-page. --Quiddity·(talk) 01:17, 21 August 2006 (UTC)

While I, as of yet, have no strong feelings on the merge proposal, I strongly object to having meta information - the merge notice - on the top of a page that is such high profile. We should direct that notice to the talk page - editors watching will still see it and the discussion can go on without distracting (and potentially confusing) new users. --Trödel 01:24, 21 August 2006 (UTC)

That was fast - thx Quiddity!!! --Trödel 01:51, 21 August 2006 (UTC)

Wow, I can't leave this place for a week without people trying to change things. What do you mean by "an expanded version"? Black Carrot 02:43, 2 September 2006 (UTC)

I'll elaborate:

• Wikipedia:Look it up is an overly short, friendly paragraph on how to do a basic search, and what you might commonly find as a result (dabs and redirs). It is linked to from Wikipedia:Questions, Wikipedia:Help desk, Wikipedia:Reference desk, Wikipedia:How to create a Wikipedia article, and 2 Help:Contents subpages.
• Wikipedia:Searching is a very long page, with many useful (and just as many arcane) details. It is linked to from the top of the Wikipedia:Main Page, Help:Contents, and many other places.

I'm suggesting that this is a bit of a mess, and could be fixed.

Wikipedia:Searching also has a number of problems, which could be dealt with as part of a general overhaul. eg the historical tangent in "Avoid short and common words". Or the stub tag. Or the terrible link to fulltext search as a guide to boolean searches. Or the guides to all the old versions of Opera. etc.

One possible solution, is to make this page into a fairly short but useful overview with the most relevant details (and redirect Wikipedia:Look it up to here); and create 1 secondary page with all the complex/minor/arcane details (eg lumrix search and bookmarklets and browser specific help and tomeraider). Much like the duality of Introduction to general relativity and General relativity, or that of Wikipedia:Introduction and Wikipedia:Tutorial.

This is a full overhaul proposal (as was originally suggested during the Main Page redesign in April). I'm just offering one solution, and calling for further suggestions. Thoughts? --Quiddity 04:01, 9 September 2006 (UTC)

I have no objections to such a discussion. And I appreciate your summary of my work, it captures the spirit quite well. Why do you say Look it up is "overly" short? What else might it include? More specifically, what else might it include that wouldn't clutter it? Remember, my original vision was to create something that 1)most people would bother to read, and 2)would encourage people not to ask dumb questions on the Ref desk. I can't say for sure that it's succeeded in either, but it does seem that fewer questions are answered with "There's a search box in the corner" or "Why don't you try the article?". I worry that lengthening it, or even making it less conversational, would damage that. Also, although I wouldn't stand in the way if you moved it out of Look it up entirely, I came up with that name for a reason as well. People seemed to have difficulty with the concept that this is an encyclopedia. A familiar phrase, the reasoning went, would steer people in the right direction. The content is also quite carefully chosen. These are things that aren't as far as I can tell explained in any other easily accessable place (such as America vs American Flag vs Flag vs Flag of the United States etc) and that make a real difference to whether or not you find what you're looking for. Black Carrot 22:18, 24 September 2006 (UTC)

That's a good point about keeping Look it up as short as possible. I retract the overly designation if I may. And the title would work fine as the intro-to-searching page.

Specifically, I'd suggest the "If you cannot find an appropriate page on Wikipedia" section be moved there, and the "Put your keyword in the searchbox." blue box might be better (shorter/clearer) than the intro paragraph currently there. I'd also suggest the "Skins" section be moved to here, or at least to the bottom, as it's not a central/common problem (afaik?).

I'll have time to show by example (merciless editing!) in a few weeks. --Quiddity 19:09, 25 September 2006 (UTC)

Proposed new section

I propose adding the following section under the "external searches" section:

Search engine settings

Please note that certain parts of Wikipedia will not be indexed by most search engines, due to how Wikipedia configures pages. Specifically:

• URLs embedded in namespaces other than the main namespace have the nofollow attribute set; thus such links will not be followed by search engines. This includes user pages, talk pages, and pages in the Wikipedia namespace.
• Old revisions of pages have the noindex attribute set.
• Deletion discussions, the Wikipedia search facility, and other special pages (like Wikipedia:Recent changes) are disabled by the site's robots.txt file [2].

(end of proposed section). --EngineerScotty 19:03, 26 September 2006 (UTC)

• I agree that we should disclose what's indexed and what's not. It might even be useful to have a separate page about Wikipedia and Search Engines. We can add some comments there about how it's not productive to use Wikipedia for link spam because... This may help discourage people from linkspamming, or at least it will provide rational arguments that others can use to educate the public about Wikipedia. Does there happen to be such a page somewhere already? If so, I haven't seen it. Jehochman (^Talk/_Contrib) 16:44, 11 January 2007 (UTC)

Macron folding

The section "Words with special characters" only mentions diaeresis. From experimentation, it would appear that Wikipedia search does not fold macrons. Is this by design, a known issue, or soon to be fixed? A (very long) discussion is now occurring on WP:MOS-JA. Properly writing Japanese terms in Latin script often requires macrons. However, some have argued that it hinders searching and thus accessibility. Any comments would be welcome. Bendono 07:03, 16 October 2006 (UTC)

watch

Dear Administrators, Because of repeated vandalism by unregistered IP addresses, I suggest that you make this page un-editable for newly registered&unregistered users. —The preceding unsigned comment was added by Happy8 (talk • contribs) .

Semi-protecting this page permanently is a pretty good idea. Any admins agree? Or does it need to be put through the process: Wikipedia:Requests for page protection? --Quiddity 18:51, 4 November 2006 (UTC)

Done. -Quiddity 21:58, 13 November 2006 (UTC)

Search for article name not yielding article

Searches for 1997 Constitution of Thailand, '1997 Constitution of Thailand', and "1997 Constitution of Thailand" do not yield the 1997 Constitution of Thailand article. In fact, the 1997 Constitution of Thailand article does not appear at all in the resulting list of articles for any three of the searches. The article has been around since September. What is causing this problem ? Patiwat 06:54, 4 November 2006 (UTC)

There is no way to tell when the search index was last rebuilt without asking the developers, but it has in the past not been rebuilt over a span of several months. I would assume it has not been rebuilt since this article was created on September 20. -- Rick Block (talk) 13:29, 4 November 2006 (UTC)

Searching for uploaded images

Is this possible? Would be useful to find pictures that might already be on the server rather than upload more similar ones to illustrate a point. 4wd 21:53, 13 November 2006 (UTC)

You can already do that. Just check the "Image" namespace button on the search page. Alternatively, one could search from Special:Imagelist. — User:ACupOfCoffee@ 21:22, 21 November 2006 (UTC)

alban skenderaj

Alban Skënderaj është një këngëtar shqiptar që vjen nga Shqipëria. Albani ka lindur në qytetin e Lushnjës, në 20 prill të vitit 1982. Me origjinë Albani është nga Vlora, por ka lindur në këtë qytet, pasi babai i tij gjatë asaj periudhe punonte në Lushnjë.Në moshën 15-vjeçare është larguar nga Shqipëria për në Itali, ku ka gati nëntë vjet që banon në Pistoja afër Firences.Është tip që i pëlqen natyra dhe beson se çelësi i sukesit të tij është për faktin se, prezantohet tek publiku për atë që është në të vërtetë.Albani tashmë ka bërë famë dhe preferohet nga shumë njerëz. Ai ka marrë në vitin 2005 çmimin kryesor në festivalin "Top Fest" me këngën e tij(me të cilën e filloi karrierën)"Për Ty". Edhe këtë vit në "Top Fest" fitoi çmimim kryesor në bashkëpunim me grupin e njohur kosovar "Kthjellu" me këngën e tyre të suksesshme "Diçka". Deri më tani Alban Skënderaj ka edituar edhe albumin e tij të parë që titullohet "Fllad në shkretëtirë". Gjithashtu këtë verë ka korrur një sukses të madh edhe me këngën e tij hit "Ky ritëm". Është vërtetë çudi se si një djalosh si Albani që më parë nuk e ka njohur askush tashmë të njihet nga të gjithë dhe të ketë shumë adhurues. Shpresojmë që të kemi edhe në të ardhmen të tillë talentë si Alban Skënderaj...

Anyone know what language this is or what it says? My guess is it has nothing to do with searching, but I'm not sure enough to just delete it. -- Rick Block (talk) 17:15, 2 December 2006 (UTC)

Albanian, I think. Something about Alban Skënderaj, an Albanian singer. Prometheus^-X303- 21:05, 2 December 2006 (UTC)

I speak Albanian, and I'll translate it for you by the end of the day. - RyanGibsonStewart (talk) 19:06, 3 December 2006 (UTC)

• Okay, the translation is as follows. Please note that I translated it exactly, and didn't make too many grammatical errors (except for clarifiaction). Translations of songs/titles are in italics. It's not very well written, and needs a lot of cleanup.

Alban Skënderaj is an Albanian singer that comes from Albania. Alban was born in the city of Lushnjë, on April 20, 1982. Alban is originally from Vlorë, but he was born in this city (Lushnjë), since his father worked in Lushnjë during that period. At the age of 15, he left Albania to go to Italy, where he had almost nine years that he lived in Pistoja near Firenze (Florence). He is the type that likes nature and believes that the key to his success is factual, for he is presented to the public for that which is true. Alban now has become famous and is liked by many people. In 2005, he received the main prize at the "Top Fest" festival with his song (with which he began his career) "Për Ty" ("For You"). And in this year at "Top Fest", he won the main prize for collaboration with the well-known Kosovar group, "Kthjellu", with there successful song, "Diçka" ("Something"}. Until now, Alban Skënderaj has edited even his first album, which is titled "Fllad në shkretëtirë" ("Breeze in the Desert"). As well, this spring he has reaped a great success with his hit song, "Ky ritëm" ("This Rhythm"). It is truly astonishing how a guy like Alban, who earlier had not been know by anyone, now is known by all and has this many admirers. We hope to have a talent such as Alban Skënderaj in the future.

Go button functionality

The Go button functionality part of "search" isn't well represented here. There is an article in help: Help:Go_button but it isn't linked. There is a different page linked at the very bottom of the article to Wikipedia:Go_button. The search function provided in "go" is fairly unique and not necessarily intuitive. It deserves to be better documented and that documentation easier to find. Maybe the first page of "search" should have just two entires: Go and Search.

Also, while we are on the search-go topic, the first letter case insensitive is more than puzzling. If I type in "ide" I get the "Ide" page but no hint of the "IDE" page existing. The result is very non-intuitive--except for the fish lovers in the crowd. Rman2000 15:05, 10 December 2006 (UTC)

There's a disambiguation link to IDE in the first line at Ide ;)

As for this page: I believe that this page and Wikipedia:Go button and Wikipedia:Look it up all need to be merged into 2 "search" pages - 1 simple and 1 advanced. But I haven't had the time/energy to create a draft proposal. -Quiddity 19:47, 10 December 2006 (UTC)

First-letter case sensitivity has little to do with the search algorithm (AFAIK) and much more to do with the fact that no article has a lowercase first letter. BigNate37_(T) 14:01, 12 December 2006 (UTC)

Dec 2006 - Search broken still?

Is it me, or is SEARCH just completely blotto? Clicking GO on sikorsky, yields the Sikorsky page, but clicking Search give me NOTHING. Neither do *sikorsky* or even sikor* . I've read the previous notes, and have to wonder if I've broken something in preferences, or..... I have the current namespaces enabled in my preferences:

• Main
• Wikipedia
• Image
• Image Talk
• Template

TIA, David Spalding (☎ ✉ ✍) 17:23, 13 December 2006 (UTC)

I concur.

http://img96.imageshack.us/img96/1885/brokenwikipediasearch1uk8.png (Searched "Requested Articles" and pressed 'Go')

http://img89.imageshack.us/img89/5882/brokenwikipediasearch2lw2.png (Searched "Requested Articles" and pressed 'Search')

It appears that 'Search' is not working at all, and 'Go' has some sort of Internet brain fart when trying to find matching articles. Someone ought to fix this. —The preceding unsigned comment was added by 24.89.253.231 (talk) 00:14, 14 December 2006 (UTC).

Higher matches than exact match!?

The results for searching for "observer":

Observing Relevance: 100.0% - -

Observations Relevance: 98.7% - -

Observables Relevance: 98.7% - -

Observer Relevance: 97.3% - -

How come there are 3 matches that are higher than the actual word itself?

--202.45.98.233 00:07, 23 December 2006 (UTC)

felipe faria

Filipe Faria nasceu em 1982, em Lisboa. Frequentou a Escola Alemã de Lisboa desde o jardim de infância até completar o 12º ano de escolaridade. O contacto e convívio com aquela cultura de origem germânica, tão diferente da nossa, possibilitou a abertura de novos horizontes. Impulsionado pelo forte interesse demonstrado pelo período da Idade Média, e pela descoberta algo fortuita de uma verdadeira relíquia na biblioteca escolar – a Tolkien Bestiary -, cultivou, desde cedo, a paixão pela literatura fantástica. As «Crónicas de Allaryia» assinalam a sua estreia no mundo literário. Uma obra que nasceu de uns esboços de uma aventura, iniciados hà cerca de quatro anos, que lentamente ganharam corpo e forma e evoluíram para um livro de quase 600 páginas. Em 2001 foi o vencedor do Prémio Branquinho da Fonseca, organizado pela Fundação Calouste Gulbenkian e Jornal Expresso. Em 2002 ganhou o Prémio Matilde Rosa Araújo - Revelação na Literatura Infantil e Juvenil. Actualmente encontra-se a frequentar o curso de Línguas e Literaturas Modernas na Faculdade de Ciências Sociais e Humanas. E o resto é uma história ainda por escrever…

Internet Explorer 7

Under browser specific help, perhaps someone could add instructions for setting up the automatic Wikipedia search in the new IE7 search. Drwhit73 05:33, 4 January 2007 (UTC)

Wikipedia's search is aweful!

It deserves something better. And I don't think using Google for searching Wikipedia is a good alternative. But currently, why not just shut down the internal search and rely on Google for now until a working search is ready?

This is not trolling, I'm serious. Wikipedia's search is unusable.--87.122.45.130 20:53, 10 January 2007 (UTC)

Oh, in case you wonder "why disabling" when you don't have to use it? - It saves CPU time.--87.122.45.130 22:50, 10 January 2007 (UTC)

Capitalization in search terms

Can someone explain capitalisation to me? iphone works as a search term and returns Apple's iPhone page. iphone (linksys) does not return Cisco's iPhone (Linksys) page or any other results - for that you have to enter iPhone (linksys). How stupid is that? The search terms should be case-independent. Lloyd Wood 00:12, 11 January 2007 (UTC)

Capitalisation matching kicks in after the first letter. iphone goes to iPhone, because someone created an iphone article (actually Iphone, it's a limitation of the underlying software) and redirected it to iPhone. I've created a similar article and redirect for iphone (linkys), so it will redirect properly now! --Steve _(Slf67) ^talk 04:44, 11 January 2007 (UTC)

How to find sought but not found keywords?

I would like to see a function that list say top200 of typed entries but not found ie like this:

 http://en.wikipedia.org/wiki/Telex
   Telex not found etc..

 1. 'Telex'
 2. ...
 etc..

Any possiblity to implement this..? Can't be that hard.. Electron9 11:34, 11 January 2007 (UTC)

Update on the broken Search function

Is there a page that reflects acknowledgment of the issue? Is there a page that gives status updates? It's tedious to keep checking, I would rather just put an update page on my watchlist. Joie de Vivre 17:11, 11 January 2007 (UTC)

Strange contents page

Why is the contents page for this page constructed manually? Carcharoth 14:53, 29 January 2007 (UTC)

please correct

It says "Search is case sensitive". I thought it was case INsensitive? 81.244.151.163 16:18, 30 January 2007 (UTC)

For Google and most other search engines, search is not case sensitive. For Wikipedia search, it matters much if you search on United States Senate Page versus united states senate page. No correction is needed. -- John Broughton (☎☎) 19:14, 8 February 2007 (UTC)

erasing searches

Why can't one erase their search history from the search bar? —The preceding unsigned comment was added by Richbank (talk • contribs) 04:54, 7 February 2007 (UTC).

Is this a function of your browser? In any case, it looks like entries can be deleted individually (I'm on Firefox; YMMV). -- John Broughton (☎☎) 19:11, 8 February 2007 (UTC)

Utter nonsense about wildcards

The project page says:

"Wildcards

"Be careful about limited wildcards, as they take a toll on the server. See Boolean fulltext search for details on their use."

There is absolutely no reason to be worrying about a "toll on the server".

The biggest problem is that this links to a totally incomprehensible external page, something that provides no clue as to what you might enter in the Wikipedia search box to make it work. Nothing I tried works.

Does it ever work at at all? Whether it does or not, it is irrelevant nonsense that ought to be removed from the talk page unless it can be better explained. Gene Nygaard 14:20, 18 February 2007 (UTC)

More on Numbers

Numbers are sufficiently frequently used in searches that I think their current obscure behaviour in search should be made quite explicit. I would give specific examples - e.g. that in text search "1984" fetches nothing, and "U2" gets the articles: U1, 'U', U571, U7, U-2 etc. On the other hand, in title search ("Go"), the right thing happens. I would hope that eventually the search behaviour will be fixed so this (embarrassing) explanation can be removed. JMPrager 16:47, 19 February 2007 (UTC)

Straight Up Awful

All the forums seemed way too specific for me to comment on this, so I thought the best place would be here. WIKIPEDIA'S SEARCH ENGINE IS AWFUL! I'm sure this has to do with certain issues but I don't understand why the seemingly most important feature is one of the least user friendly. Entries misspelled by one letter turn up bupkis. Entries spelled correctly return a laundry list of the same article but with the title formatted differently (I.E. Denzel Washington denzel washington, etc.). Is there not a way to have more accurate spell correction the way Google does? Also, is there not a way to only show the article that all the different misspellings and capitalizations are redirected to? It seems ludicrous! Any defenders of the current engine only need perform this experiment: search "promiscuity" and see if ANYTHING turns up. Yet when searched in Google Voila! There she is, the wikipedia article second from the top! This does not make any sense. J-Rod 06:02, 26 February 2007 (UTC)

The internal search engine at Wikipedia is acknowledged to be relatively ineffective; that's because the Wikimedia Foundation has chosen, for several reasons, including financial ones, to use as much free software as possible, and there isn't a free search engine out there that offers the functionality of (say) Google. That's why the project page has so many suggestions about external search engines - they're better. -- John Broughton (♫♫) 16:17, 8 March 2007 (UTC)

I don't buy this. Lucene (which Wikipedia apparently uses now) is a very powerful and very fast text search system. It supports fuzzy matching, stemming, snowballing, and all manner of very useful features that would be great to have if they were only taken advantage of and implemented in a sane way so as not to rape disk I/O. It just seems like the front end that has been written to fit it into MediaWiki is braindead. -- mattb @ 2007-03-08T22:49Z

Are numbers treated specially? Where is this documented?

I've just spend half an hour trying to use the search page to find the article about the McCarthy 91 function: I could remember the "91 function" part, but not "McCarthy." Of course, searching for the word "function" alone gives rather a lot of matches--103698 at the moment--but narrowing it down by searching for "91 function" (with or without double quotes) doesn't work: I get even more matches (104037).

I don't understand!

(At the risk of going off on a tangent, it would also be very useful if a search could be restricted to page titles only; perhaps this is already possible, but it wasn't obvious how.)

-Alexander Hanysz 203.122.212.109 11:43, 1 March 2007 (UTC)

If your initial search is unsuccessful, you're much better off doing a google search of the English-only Wikipedia (follow the link on the project page) rather than continuing to spend time on internal search. It is widely acknowledged that the Wikipedia internal search engine is not very good. For example, google has the inurl parameter that allows searching on page titles, while internal search has no equivalent. -- John Broughton (♫♫) 15:58, 8 March 2007 (UTC)

please help me

can anyone tell me about the periods of significant change along the adelaide coast

This type of question should be asked at Wikipedia:Reference desk. -- John Broughton (♫♫) 16:14, 8 March 2007 (UTC)

Is this a bit outdated?

I know Wikipedia uses some Lucene-based search engine these days, yet this page still links to the MySQL FULLTEXT implementation of boolean search operators. The boolean syntax for Lucene's queryParser is different from MySQL FULLTEXT's. Shouldn't we perhaps link to the Lucene doc on queryParser syntax? Obviously a lot of this isn't relevant since it includes functionality that may be disabled or stripped out by Wikipedia's search front end.

On that note, is there any information available on how the search implementation works and what Lucene features are accessible to end users? I tried using Levenshtein fuzzy keyword matching (the tilde character), but the search front end spits out an unhelpful error message when I do, so I assume that the front end is stripping out the character and then assaulting me with an awful default (and perhaps misleading; I never get a different result on subsequent trials with the same query) error message. Incidentally, I get the same unhelpful error when I attempt to use wildcards (which are allowed according to this page). If the search front end isn't helpful enough to explain that this functionality is intentionally disabled, shouldn't we specify that here? This page needs some serious revamping considering that full text search is a rather important function on a site with a mass of text of the magnitude of Wikipedia's. -- mattb @ 2007-03-08T22:42Z

P.S. - Lucene's searching/indexing classes are NOT case-sensitive, so the tidbit regarding case-sensitivity on this page doesn't make any sense unless there is some weird quirk in the Lucene front end that would cause this. Some informal testing seems to indicate that the current search engine is not, in fact, case-sensitive (even the example case given matches correctly with all manner of capitalization variation). -- mattb @ 2007-03-08T22:46Z

The bit about case-sensitivity is likely referring to Go's case sensitivity (which has fairly arcane rules that generally mask even its case sensitivity). I believe all MediaWiki code is open source, so should be available online somewhere. If you are a Lucene expert, I'm sure your help would be welcomed by the developers. user:Brion VIBBER is Wikimedia's chief technical officer (aka "head developer dude"). user:Tim Starling is the main database guy. Another approach might be make suggestions to the wikitech-l mailing list. -- Rick Block (talk) 02:36, 9 March 2007 (UTC)

metadata - where's it come from?

A google search for wikipedia+jericho yields:

Jericho - Wikipedia, the free encyclopedia

Encyclopedic entry includes details about the archaeology of this ancient city.

en.wikipedia.org/wiki/Jericho - 58k - Cached - Similar pages

Jericho (TV series) - Wikipedia, the free encyclopedia

Article provides a synopsis, episode listing with screenshot with original airdate, characters and trivia.

en.wikipedia.org/wiki/Jericho_(TV_series) - 87k - Cached - Similar pages

But the descriptor strings, ie "article provides a synopsis", don't appear within the article itself and don't appear to be present in the page source fed to Firefox. Are they manipulated by the Wikipedia, or are those descriptors set by Google themselves? I note that searches for less common articles, like David Duchovney just return strings like "Return to Me was directed by Bonnie Hunt and starred David Duchovny as Bob and ... Wikipedia® is a registered trademark of the Wikimedia Foundation, Inc., ...", so it certainly looks like the metadata is being manually written at some point. MrZaius^talk 02:25, 9 March 2007 (UTC)

Google is getting this annotation from http://dmoz.org/Arts/Television/Programs/Dramas/Jericho/. See http://www.samspublishing.com/articles/article.asp?p=608629&rl=1 . -- Rick Block (talk) 20:00, 9 March 2007 (UTC)

google-type autocorrection

is it possible? has it been discussed? i currently use google adding 'wikipedia' after words to have it autocorrect my clumsy typing. --Leladax 18:26, 14 March 2007 (UTC)

Search not working

Currently, search is not working. It consistently returns no results and says there was "a problem" with my search, even though I did not do anything incorrectly. Moreover, even before today, searching would intermittently fail with the same symptoms. Searching should, obviously, behave in a perfectly consistent and predictable fashion; why didn't it? And why is it now consistent, but completely useless?! —The preceding unsigned comment was added by 74.104.131.76 (talk) 19:48, 14 March 2007 (UTC).

Have you anything more helpful to add besides trumpeting my IP address all over the 'net for some reason? If you're going to take the time to post a response, how about actually addressing the content of what I actually wrote in some manner?

It's now been well over 24 hours and the only response posted so far has been the one telling everyone my IP address, as if they couldn't find it in the page history if they really wanted to. None has actually said anything about the search functionality, and as of noon Eastern time it is still completely unusable. Anything typed into the box and then "Search" clicked results in "There was a problem with your search". Which isn't very specific. And it's strange that there's "a problem" with EVERY search query. Also that it says that it's "temporary". Apparently you guys include "lasts for days" in your definition of "temporary".

So what is this "problem"? How is it a "problem" with "my search", when my search is simply type and click, like everyone else's? It sounds like it's either a problem with the "my" rather than the "search", meaning you have a "problem" giving a specific person search results for some reason, or it's not a problem with "my search" at all but rather a problem with "your computers". If the latter, I suggest someone get off their can and fix it already.

The person that added your IP is a bot. It's standard courtesy on Wikipedia talk pages to sign your comments, and if you dislike your IP address being shown (which, as you know, can be obtained from the page history anyway), you should register a named account. What are you trying to search for, exactly? The error you're seeing is unfortunately the catch-all message and could just as well mean that there's a syntax error with your query as it could indicate actual server troubles. If you're trying to use any wildcards ( * ), boolean operators ( & | ), fuzzy matching ( ~ ), or basically any advanced Lucene search feature, you will see the error message you describe. -- mattb @ 2007-03-15T16:49Z

Adding my IP just seems pointless because as I noted it is available on the history page. Registering and signing stuff is not something I prefer to do online, either; way too many Web sites throw registration hoops at you, often just to read stuff let alone contribute. So rather than give any special treatment I register at none. Lastly, the search behavior I described occurs for any query, including simple English words, and has done consistently for days now. Surely you've noticed? —The preceding unsigned comment was added by 74.104.131.76 (talk) 00:00, 16 March 2007 (UTC).

If no one signed pages, it would be quite difficult to follow discussions. If you don't want your IP "trumpeted", you know what to do. Leebo ^T/C 19:08, 16 March 2007 (UTC)

I've checked, and "signing" posts seems to mean attaching the IP address to it anyway, so I don't see the point if someone else is going to do it for me anyway. Might as well be the person who wants it done that makes the effort to have it done eh? In any event, this is a digression from the original purpose of this discussion, which was the misbehavior of search. I note that it has apparently now been fixed. That this took several days after the problem was announced here is dismal, however.

As for registration, I'm finding as I look around here a rather disturbing trend. Not only are more and more pages "semi-protected" more and more often, but I've found some pages that don't have a large "This page is semi-protected; etc. etc." with links for requesting unprotection or discussing the reasons but instead have a little lock icon in the top right corner and won't edit. So now some pages' semi-protection is not up for debate or requests for unprotection? Or at least such requests are being discouraged? This isn't part of a gradual move toward forcing everyone to register before they can contribute at all, so they can then be squeezed for every drop of salable personal information by gradually adding more and more questionnaires and other hoops for people to jump through? No doubt, if that is the direction wikimedia is going to meet its future funding needs, the current registration process is quite painless with little information required; the plan would be to make it easy to register and increasingly difficult and/or annoying to avoid registering, and then tighten the screws later, and force already-registered people to fill out new questionnaires or be dumped from the system...I know, I've seen the like before at other sites. I hope you realize that if you do do this, you will quite likely destroy wikipedia, by ruining the free-for-all anyone-may-contribute spirit on which it was founded. A structure with its foundations eroded will rarely remain standing for long. I mention all this because it's gone past even the subtle, tiny lock icon: I just found a spelling error on a page and when I went to correct it, discovered that it was locked. I double checked: no warning icons of any kind on the page itself and the thing at the top even said "edit" rather than "view source", but it wouldn't actually edit and it said it was locked on the edit page -- but ONLY on the edit page. As far as I'm concerned, that was the final straw, and I won't bother clicking any "edit" links here in the future after submitting this note until and unless I have good reason to think this creeping protectionism trend has been reversed! I don't want to see anything locked ever again except with a big fat banner with prominent links to request unprotection and the like and a solid reason such as a recent spree of vandalism. Regardless of the ultimate goals of the people running the show, the trend towards making protection less obvious seems designed to let them get away with doing it more, with weaker justifications, without incurring the cost of being flooded with requests for unprotection and other comments. Sorry -- no can do. That MUST be the price of (semi-)protecting pages or there will be a gradual trend towards locking down the site and turning it from what it used to be into an elite-dominated, narrow-POV collection of a few peoples' blogs in effect. At bare minimum, putting NO indication WHATSOEVER on a page that editing is actually locked must be VERBOTEN, but that is precisely what I encountered only minutes ago elsewhere on this site! But somehow I don't expect anyone in charge to listen or pay any attention to a single user's voice -- and an unregistered one's, at that. I predict you'll be seeing a lot more spelling errors going uncorrected in the future!

Various forms of protection are necessary evils to stem vandalism. What you encountered sounds more like a database lock than a protected page, since protected pages are always clearly marked as such. -- mattb @ 2007-03-19T00:01Z

I don't care what kind of a lock it was, any lock is one lock too many unless there's a vandalism spree currently in progress. And protected pages are not always clearly marked as such. I can confirm that the one for the 9/11 attacks is only subtly marked as such, for starters. And further regarding vandalism, I just saw one of the elementbox templates vandalized -- and fixed in under 8 minutes. Vandalism where it's quickly seen is equally quickly fixed. At least, if it's easy for anyone to make the fix. Overly protective admins, if you'll pardon the pun, might just change that if current trends continue. Reasonably, the time to fix a vandalism will be more or less proportional to the article's frequency of access and to the proportion of people who can actually make the fix. The former also scales up the frequency with which someone is affected by the vandalism, so the number of times a vandalism is viewed is inversely proportional to the proportion of people who can fix it (and to the proportion of viewers who bother to edit to fix things they notice are wrong). Protect stuff a lot, and while the frequency of vandalism incidents may go down, the impact of each such incident actually rises, and at some point the fewer, longer-lasting vandalisms are actually worse than frequent, brief ones.

My own guess here is that in fact any protection of pages other than the main page and other very frequently accessed pages, and semiprotection of pages other than templates and current victims of spree vandalism/edit wars, is counterproductive. The reason being people can't be bothered to jump through yet another bunch of registration forms and will by and large ignore a vandalism to such a page that they'd otherwise have fixed. On the other hand, a would-be vandal with an ax to grind will be bothered to register a throwaway account and use it to trash semi-protected pages. Knowing it will take longer until their handiwork is fixed, they may in fact purposely gravitate towards semi-protected pages, since that way their attack will last until a registered user happens upon it and not just until one or two ordinary users do so. So the current system creates perverse incentives -- it will cause a flurry of registrations by vandals, and focus their attacks on the pages you're trying to protect! In fact, semi-protection is probably largely useless. Blocking known repeat offenders and fully protecting key pages is more important. Semi-protection just discourages normal edits and fixing of vandalism, while only putting a trivial speedbump into the path of the determined attackers. Any kind of discrimination against unregistered users counts likewise -- the effects on a determined attacker are trivial, while discouraging normal people from contributing. That makes such discrimination counterproductive to Wikipedia's aims, as it will discourage contributors far more than it will attackers.

Thus I'd recommend dispensing with semi-protection and with any other registration-required barriers, and instead focus on blocking actual vandals. Use full protection briefly during sprees, and on key pages. Including all widely deployed templates. Templates are legitimately changed only rarely and are a particular target of vandals because a) it's harder for a naive user to locate and fix the vandalism and b) it can affect many pages simultaneously, and even pages that are nominally (semi-)protected. Also, a little research shows that registration actually makes it harder to trace users and easier for them to generate sock-puppet accounts; an unregistered user shows as their IP, and unless they have a dynamic address on a netblock with other Wikipedia editors, actually is more accountable and blockable than a registered user, or at least no less so. Requiring registration for anything creates at most some sort of illusion of greater security, while actually accomplishing nothing useful, unless there's something I've not read or been told. Perhaps something in whatever registration forms exist. Even then only if it's validated in some way, in which case I certainly will never register. Anything that needs validation is a privacy hole. Email addresses could be spammed, and can be easily replaced anyway; anything else, such as CC#s, bank accounts, or physical street addresses is potentially far more dangerous if misused, although securer. Needless to say I won't give out anything over the 'net but an email address, and that extremely grudgingly! Many other people feel likewise.

As a side note, it may be worth mentioning that I am in fact fairly knowledgeable in the field of computer/IT security, and just earlier today and again yesterday recognized something as an IT attack/probe that was mistaken by others for another, more ordinary type of attack. Of course you have no way of confirming my credentials, so take this with a grain of salt if you so choose, but do examine the math regarding the proportions of users and such above; that's on pretty firm ground, close to rigorous, albeit statistical in nature. And check out blogs such as freedomtotinker.com while you're at it; and many linked from there. These include articles where actual IT security experts weigh in on the (lack of) value of "speedbumps" to undesirable behavior, particularly when legitimate users are impacted more than illegitimate ones, or the latter have more motivation to overcome the speedbumps. And on other things besides. In fact, whoever runs the show here (or at least is in charge of the security side of things) should definitely bone up a bit at those sites. —The preceding unsigned comment was added by 74.104.131.76 (talk) 02:59, 19 March 2007 (UTC).

Lovely. This really isn't the place to discuss your discontent with the protection policy. If you really want to get a discourse going on this matter, you should bring it up at Wikipedia talk:Protection policy since it has nothing at all to do with searching. I do think you have some misunderstandings about the protection policy, however, so I'd suggest reading up on that before forming your conclusions. -- mattb @ 2007-03-19T03:59Z

I'm simply continuing a discussion where it came up, and where I check for responses. If you think there are "some misunderstandings" please be more specific, rather than making vague public accusations of stupidity on my part, which is what you appear to be implying. If you mean protection serves some other function than vandalism prevention, I was responding to an earlier person claiming it was for vandalism prevention. Besides, the obvious function besides vandalism prevention would be to coerce people to register to harvest personal information of some kind, which I would not consider to be a legitimate goal of Wikipedia, and therefore would not support. On the other hand, if you think the protection policy as currently implemented is actually effective for vandalism prevention, see above -- it will deter those who'd fix the vandalism quicker more than it will deter vandals, unless getting registered is expensive (in time, money, or privacy, not necessarily in money). As I said above, far more effective would be to use full protection briefly during spree vandalism, and use full protection on widely used templates. Right now a lot of templates, like the elementbox ones mentioned earlier, are vulnerable for no real good reason, while articles that would benefit from wider editability, even by being more rapidly fixed when vandalized, are semi-protected seemingly on a quasi-permanent basis. And forcing vandals to register first is unlikely to be much more than a minor speedbump, particularly as unregistered users are at least as identifiable, blockable, and the like as registered ones and therefore if anything registered ones should be the "untrusted" ones.

But then, IT security is clearly not your strong point, is it? Too bad it's also not the strong point of whoever's actually managing security for Wikimedia...and neither is keeping proper changelogs and the ability to roll back or forward transactions. The search functionality is consistently failing again as of late last night. If they used proper change management, this could never happen. The change that broke search would have been rolled back at the click of a button, instead of taking several whole days to undo. It would certainly not have been promptly repeated, and then remained broken hours later instead of the fixed state that existed yesterday morning being restored within five minutes with a couple of mouse clicks. As far as I can tell, system administration here is being performed by a circus clown leading a team of trained chimpanzees, since even my younger brother would have avoided some of the elementary mistakes I've observed lately. Both in change management/testing AND in security and threat modeling. —The preceding unsigned comment was added by 74.104.131.76 (talk) 16:20, 19 March 2007 (UTC).

And if you are still skeptical of my claims that semi-protection is worse than useless at controlling vandalism, you obviously never bothered to read the comment immediately below this thread! I'm not talking out of my hat when I say it will deter fixing vandalism more than it will deter vandalism; besides all of the (sound) theory I've given, there's empirical evidence sitting right here in this page that nobody else seems to have bothered to read yet. Tsk, tsk.

As of 7PM (UTC) it seems search is once again working. This outage was much shorter -- about 12 hours instead of 3 days. It is still a mystery how their change management can be so terrible that after accidentally gumming up the works it takes them any longer than 5 minutes to hit "undo" or, at worst, 20 to dig out a backup tape and restore some file. That they screw up a configuration edit or similar modification seriously enough so as to cripple basic functionality at a frequency of two to three times a week also does not inspire confidence. And I have this funny feeling that they aren't even discovering the ill effects of their tweaks for quite a while after making them, instead of immediately by testing basic functions promptly after making any change... maybe it's time they outsource the "search" half of the "go/search" functionality to proven experts, namely Google. Lots of other sites do, these days, and Googling "foo site:en.wikipedia.org" is pretty useful anyway, so... (At least the "go" half of that functionality seems to work 100% of the time, even when they've b0rked "search" up something fierce.)

People are less likely to devote the time and energy to a detailed response if you come across as caustic and patronizing. As I stated earlier, this is not the place to discuss this and your concerns will get no exposure whatsoever here. I'm in no mood to continue this discussion, so I again urge you take it to another talk page if you really want to see more interesting dialog on the matter. Good day. -- mattb @ 2007-03-19T19:25Z

Okay, then, oh Wise and Mighty Matt, where do you suggest in Your Infinite Wisdom that recurring problems with the search engine be discussed? Just post the URL and I'll be happy to go there! (By the way, it's down again. What did they do, rig a "works normally/claims there's 'a problem' with every search" mode toggle to a hardware pushbutton and then stick it under a mattress in a place where kids live? It may also be worth noting that search is slow, and even slower when there's "a problem" with the search. But high latency is the least of your problems, obviously; it's just curious that it doesn't affect article pages, talk pages, or even "go" results, only "search" results, and is worse at the same time search returns bogusly empty result lists.

I think he means your off-topic protection and vandalism rants. And you're still being patronizing and caustic. Leebo ^T/C 23:24, 19 March 2007 (UTC)

I was, in fact, referring to the protection/vandalism rants which comprise the bulk of your text. However, rants about recurring problems with the search engine are also inappropriate here. None of the Wikimedia/Mediawiki developers watch this page as far as I'm aware. There are Wikimedia tech mailing lists where you could take your complaints, but given your charm and self-important attitude I think you'd be ignored about twice as fast as you are here. However, I do encourage you to take up your issue in one of the appropriate outlets: WP:PROT for the protection policy, WP:VAND for vandalism, and WP:ML to find the tech mailing list on which your rants may be quickly ignored. -- mattb @ 2007-03-20T00:11Z

P.S. - Search engine latency is "curiously" somewhat disconnected from the rest of Wikipedia latency because the search back-end is based on Lucene, which uses its own index files rather than having anything to do with the replicated MySQL database MediaWiki uses. Slow downs in Lucene do not necessarily beget slow downs in the database, and I would guess that the front end that has been developed for Lucene on Wikipedia doesn't run on the DB servers, even further reducing the correlation between search and "rest of the site" speed. Take this for whatever it's worth though, I'm obviously not a self-professed IT security professional, so it's quite possible I have no idea how this site works. -- mattb @ 2007-03-20T00:29Z

My comments about protection and vandalism are no more off topic than the comment immediately below this thread, which you may note was not authored by me. Or at least would be able to note, except that all the evidence indicates that you are mysteriously unable to even perceive that particular comment.

I'm just kinda mystified by the registration paranoia and conspiracy theories, especially when Wikipedia requires no information to register. But that's just me. Leebo ^T/C 01:29, 20 March 2007 (UTC)

I'm just kinda mystified by the pressure brought to bear on anonymous contributors to register, given the evidence that it doesn't make them any more (or less) accountable. It only makes sense if there's some angle for the administration -- marketing/marketable info being the usual suspect. If, as you claim, it doesn't get them anything at all, either greater accountability of that user or money in one manner or another, then why expend so much effort pushing registration at people, as you are doing? BTW as of around 9 PM (UTC) the search remains broken. This third outage has now exceeded a 24 hour duration. Durations so far: c.72 hours, c.9 hours, c.24 and counting hours. Percentage of time it's working, averaged over the past five days: 17. In other words, it's not working 4/5 of the time lately. This is inexcusable.

Well, as of 12:30 PM (UTC) Thursday, it's working again. I don't know how long that'll keep though. Judging by its recent behavior, perhaps until the wee hours of Friday if we're lucky.

We cannot do a single thing about it. If it makes you feel better to complain here, do whatever floats your boat, but you're accomplishing nothing but bumping pages up our watch lists. -- mattb @ 2007-03-20T21:44Z

The benefits are all to be had by the user. Nobody's pressuring you. We do want you to take on a cooperative attitude if you continue to edit though (sign your posts). Leebo ^T/C 22:29, 20 March 2007 (UTC)

Vandalism

Interestnig, [3] vandalism cannot be undone, because of: Note: This page has been semi-protected so that only established users can edit it. So this means, that an established user (Cruci-fix) wrote " FUCK FUCK FUCK FUCK FUCM< fkljakls" ??? Apparently more established than me...

greetings, Ziga 20:38, 15 March 2007 (UTC)

crappy search engine

wikipedia has the crappiest search engine ever. —The preceding unsigned comment was added by 136.150.200.99 (talk) 00:14, 30 March 2007 (UTC).

Known issue; that's why there is a "External search engines" section on the page. -- John Broughton (♫♫) 01:33, 31 March 2007 (UTC)

delete some stopword on a local Wiki

Is it possible to change the stopwords on a local mediawiki software, without using another search engine like lucene? where can I find the list of stopwords. --141.113.85.21 16:20, 26 April 2007 (UTC)

I think Mediawiki uses mysql FULLTEXT indexes for searching by default. You can change the default stopword list by defining the ft_stopword_file system variable. Be forewarned, however. MySQL FULLTEXT doesn't scale very well, and reducing the stopwords list will exacerbate this problem. -- mattb 16:54, 3 May 2007 (UTC)

Easy search

Somone wrote in 2005:

I come from the German wikipedia, and that language has, like many others, diacritical letters, which are often expressed by some
other means, e.g. use 'ss' instead the sharp s 'ß', or 'ae' instead of 'ä'. The situation is similar in Spanish (e.g., á é í ó ú),
...
Names are often written in the spelling of the original language (Perón of Argentina), or simply as Peron.

And I wrote on 20th of December 2005:

== ignore diacritics ==
I am admin at the lingála wiki and I am contributor at the german and alemanic wikis. There is a problem with the search engin.
When I cannot write the diacritics (as a user) or the use of diacritics is not know to all the users (not in german, french
or english, but in a lot of not teached languages (p.ex. kikongo, lingala, ciluba, kiswahili, ....), I cannot find an article. 
Example from the german wiki: If I search lingala, I am linked to the article Lingala. When I use the
lingala spelling of lingala lingála there is one 7%-result (list of
languages of the world). Example from the lingála wikipedia: If you are congolese and you don't no how
to type ɔ and ɔ́ you cannot find the article about your country in your language: Kɔ́ngɔ - even there
is in some older dictionarries the spelling Kongó.
Well there is a possibility to make for each article 4 or 5 redirects with different spellings. In german there is a redirect from
Fluß to Fluss, but one from Strasse to Straße. There is obviously no rule (in Germany and Austria: Fluß, Straße; in Switzerland and
Liechtenstein Fluss, Strasse).
If the wiki search engine could learn that letters with and without diacritics are (more ore less) the same, that ɔ, ss and ɛ are
similar than o, ß and e, it would be great and very helpful.
*o search also ö ó ô ǒ ɔ  ɔ́  ɔ̂   ɔ̌   
*e search also ë é è ê ɛ   ɛ́  ɛ̂   ɛ̌
*a search also  ä á   â   ǎ    
*u search also ü ú   û   ǔ 
*i search also ï   í   î   ǐ
*ss search also ß
*ß search also ss

In 2005 there was no answer, but the problem is still the same. So I post it after 16 month again. If it is technically possible, it would be great for many languages! --Eruedin 16:10, 3 May 2007 (UTC)

This is a search collation problem (which becomes pretty tricky in languages that use non-Latin characters). If memory serves, default MediaWiki uses MySQL FULLTEXT for its search engine, which should collate a lot of languages' "equivalent" characters. In English Wikipedia, the search backend is a rather kludgy Lucene based implementation. The Lucene API does support some collation, but we (referring to myself and some others in an unrelated project) found it generally easier to write our own procedures for collating the appropriate characters.

In other words this is a limitation of the search software itself, and would require some significant rewriting of the search back end to fix. All you can really do without modifying the search engine is use Google, which has very good collation abilities (probably your best bet), or create a lot of redirects. -- mattb 16:49, 3 May 2007 (UTC)

Redirect pages in search results

I recently posted a question to VP/T, concerning the fact that for some searches, both the Go and the Search button will lead you to a search results page which redundantly includes all redirect pages (example: Search, Go). I know that Google can be used to circumvent this issue (and also the lack of fuzzy logic keyword analysis), but for new readers that option may not be obvious. I believe it would be a genuine improvement if the result pages where made to exclude the redundant redirects. —AldeBaer 06:00, 7 May 2007 (UTC)

Search button on home page

The search button on the home page is very unattractive and I believe we should replace it with something more professional, or at least make it an image instead of a simple character ">". Also, the actual text inside the search box should be smaller to correspond with the size of the box. If the text size is not changed then the size of the search box should be increased because their current proportion is mismatched. —The preceding unsigned comment was added by 134.173.169.36 (talk) 10:12, 10 May 2007 (UTC).

automatic category browsing

When a non-existent category is linked to, clicking it brings it up for editing and presents an option to "browse the existing categories before creating this page." However, this link just takes me to the index. I have a request for a way to have an automatic search occur, like with non-existant articles, except just in the category namespace. As it is now, I have to click on a letter on that category index in order to bring up the all articles (categories) search field and then manually type/copy-paste the category name in--annoying. It would be better to have an automatic category search occur on the non-existent category, as with non-existant articles. This would just take the category name and stick it in the same search string: http://en.wikipedia.org/w/index.php?title=Special%3AAllpages&from=test&namespace=14 (in this case, "test" is the string). -Eep² 00:49, 14 May 2007 (UTC)

favicon on dropdown list

Hi, I user the search page and found that the dropdown list haven't any favicon so i made a small page to show the result. A little bug with the selected option : no image, don't know why.  - lyhana8 (Talk) - 22:19, 16 June 2007 (UTC)

The Firefox 'wp' shortcut has been broken

It used to be possible to type in 'wp' in the url bar for Firefox and it would take you to the page %s, which in turn redirects to the main page. Now when typing in 'wp' it redirects to [4], the number at the end of the url appears to be randomly generated. Who can I talk to to get this fixed so that it redirects to the main page gain? --Android Mouse 01:29, 23 June 2007 (UTC)

title-only search

An option to search only in article titles (vs. the entire article) would be nice. ∞Σɛ_Þ² ^(τ|c) 09:50, 28 June 2007 (UTC)

Google can do this. Add "allintitle:" to the search criteria (and "site:en.wikipedia.org" to restrict the search to the English Wikipedia). -- Rick Block (talk) 13:44, 28 June 2007 (UTC)

Thanks...but this is another example of why MediaWiki's search function sucks (and why a better navigation/search function needs to be developed/implemented). :/ ∞Σɛ_Þ² ^(τ|c) 15:19, 28 June 2007 (UTC)

Pluralised redirects coming up first?

For some reason, pluralised redirects seem to be coming up first. eg. I search for "Hapa". The first result is "Hapas" which is just a redirect to "Hapa". The exact match "Hapa" is listed second. Why is this and can it be fixed? I've noticed it on other searches too. Thanks.--Irrevenant [ talk ] 01:34, 10 July 2007 (UTC)

Search URLs work with Opera too

The search URLs like

http://en.wikipedia.org/wiki/Special:Search?search=%s and http://www.google.com/search?q=site:wikipedia.org+%s are mentioned for Firefox.

FYI and possible update I run them in Opera 9.22 (WinXP) and it probably works in many older Opera versions too.

80.251.192.5 23:07, 1 August 2007 (UTC) OF

Biased

There isn't any true judgment.Because that SIRASA channel did it for their own decision. But finally they gave the right decision in final round by real&true SMS vote.So anyone can't accept for their decision.Because a lot of contestants got excessive judgment by that foolish SIRASA SUPER STAR. Italic text —Preceding unsigned comment added by Sanjanamax (talk • contribs) 08:19, August 26, 2007 (UTC)

and/or

Mention if searching for A B finds pages with A and B, or A or B in them, And say how to do the other operation. Jidanni 19:46, 30 August 2007 (UTC)

It seems as though the search terms are all required. I searched somalia tanzania; all of the top four results I checked included both terms. Neither Somalia nor Tanzania appear as search hits. BigNate37_(T) 19:53, 30 August 2007 (UTC)

Searching within a category?

Is there a way to search only pages within a certain category? Let's say I want to search for the words "lake" and "wood" in all articles that use the Category:Africa. How would I do that? Mlewan 08:48, 4 September 2007 (UTC)

I think searching for "category:africa" lake wood should do this, but it apparently doesn't. I don't know of a convenient way to do this. -- Rick Block (talk) 23:05, 23 September 2007 (UTC)

Shouldn't we add a search box?

Search boxes can be added to pages:

 

Should we add one here? And/or an image of the position of the search box on the wikipedia page (it's easy to miss on the left hand side, a picture might help a little). Yours --h2g2bob (talk) 23:49, 27 September 2007 (UTC)

Suthun Music Entertainment

Suthun Music Entertainment is an American Label Started in 2002 By Arthur Burton. The Label Is Specializing In Promoting,Recording,Distributing Urban/Hip Hop Music

Current Artist

Scrooge Lamar Richardson Alphy Nics South Maine Ladi Unkutt Presidential Tripz

[1]-Official Myspace Page [2]-Official Website

Currently Suthun Music Entertainment is Working With 20 Ent/Universal Records —Preceding unsigned comment added by Suthunmusic (talk • contribs) 23:36, 15 November 2007 (UTC)

Proposal: Link to a search of the same word in another language

I often use the Firefox search box to look for a word in Wikipedia. However, this will only search the english part of wikipedia. I would highly appreciate a language list on the search page which would switch to the search in another language, looking for the same word. Other opinions? —Preceding unsigned comment added by 129.187.179.91 (talk) 08:21, 17 January 2008 (UTC)

I fully agree and would highly appreciate this expansion

(213.100.22.98 (talk) 17:43, 16 November 2008 (UTC))

You can add search for other language versions of Wikipedia to your Firefox search box. Do like this: Just enter any page on for instance the German Wikipedia (or whatever language you are interested in), then click the icon to the left of the Firefox search box and you get a drop down list of search plug-ins you have, and below those you find the search plug-in that the site you currently are visiting suggests with the option to add that search plug-in.

However that only helps people with browsers that has that search plug-in feature. (I know that there are some more browsers that has it.) And it isn't a very visible feature even to Firefox users. So it would be very nice if the search page linked to searches in the other languages in some way.

So I tried to add normal interwiki links to the Special:Search page, that is links that are visible in the left side menu called "languages". I added the interwiki links by editing the MediaWiki:Searchresulttext, that is the text that is shown on top of every search result. Unfortunately that didn't work. But that would have meant about 250 links or so, so it would have been pretty messy anyway.

So the next option is to link from the text in MediaWiki:Searchresulttext to some page where we have links to the searches in other languages. And we already have such a page, the www.wikipedia.org page. Well, that one only allows searching the top 20 language editions of Wikipedia, but that is at least a good start.

An even better thing would probably be if we added the same top 20 language editions in the drop down list of search engines on the Special:Search page itself. Since then you wouldn't even have to type the word again. And that seems to be the original request by 129.187.179.91 above. I just realised I can easily add that to the code in MediaWiki:Common.js/search.js. Should we do that?

--David Göthberg (talk) 04:58, 17 November 2008 (UTC)

Wikipedia is terrible. no joke —Preceding unsigned comment added by 209.106.178.253 (talk) 18:51, 23 February 2010 (UTC)

Other external wikipedia search engines

There are some other websites indexing wikipedia's articles. To name a few: seariki.com, wikiseek.com. Should they be included? —Preceding unsigned comment added by 70.57.167.154 (talk) 08:25, 25 November 2007 (UTC)

Do people actually use this?

I always just look for things by typing the article title into the web address. That's the way I have always done it, and I'm wondering whether people do the same.--h i s ^{s p a c e} _{r e s e a r c h} 10:00, 25 November 2007 (UTC)

Advanced search

An advanced search link would be included near the search box, to look for in some concrete name spaces (i.e. wikipedia: image: and so on), without change every time the search preference (i.e. it can be interesting to look for in the wikipedia: meta pages once, but no forever). --Nukeless 10:54, 29 November 2007 (UTC)

I think this is a great idea. This page would have (at a minimum) the same checkboxes as the "Search in namespaces" box found at the bottom of a search results page. --AlastairIrvine (talk) 14:51, 23 January 2008 (UTC)

A very good idea, sadly still not implemented. Is it possible in some other way to do a search for articles which have a certain word in the title, where that word is not the initial one? Шизомби (talk) 17:03, 29 November 2009 (UTC)

Yes, use intitle:xxx in the search field.--Kotniski (talk) 08:06, 30 November 2009 (UTC)

Search Engine is horrible

Can someone explain why the search engine is so horrible? I have had the following happen to me several times: type something into the search box of my browser and find no hits from Wikipedia, but then when I enter the same phrase into Google, the wiki article I was originally searching for is the first article that appears. I think that the default search of Wikipedia should use Google's algorithm's because Wikipedia has by far the worst searching algorithm I have ever seen on the internet. And seriously, Google's search is pretty much the best out there. </rant> --141.212.142.247 (talk) 16:23, 5 December 2007 (UTC)

Wikipedia's search function uses the open source Lucene engine. It's at least a de facto policy of the Wikimedia Foundation to use open source software (Google's search is not). Given that alternatives (like Google) exist, improving Wikipedia's native search is not a high priority for the (mostly volunteer) developers who work on the software that runs the site. BTW - I think Wikipedia's search beats the hell out of Google's encyclopedia content. -- Rick Block (talk) 19:54, 5 December 2007 (UTC)

I would differ greatly. Google provides for the creation of custom search engines. On my main health website, I have dropped Google's site search for a couple of different Google custom search engines. I would call them open source XHTML. They are as about as free as you can get.

I think that I will add a Wikipedia Custom Search feature to my blog. -- John Gohde (talk) 20:49, 22 December 2007 (UTC)bumface gronola

Contents

The Contents at the top of the page is wrong. Libcub (talk) 06:00, 19 February 2008 (UTC)

It's a customized short version, but all the links work. The full ToC is very long. -- Quiddity (talk) 07:20, 19 February 2008 (UTC)

I can see why one might not want all the level 2 and below headings to show, but shouldn't it at least show all the level 1 headings? It's missing Internal search engines (New),Searching with TomeRaider,If you cannot find an appropriate page on Wikipedia, and Notes. Libcub (talk) 18:03, 19 February 2008 (UTC)

Poor upkeep? A half-hearted attempt to keep things simple? I don't know. I've added them, except "Notes" which doesn't really need one, and removed "Internal search engines" (see below). -- Quiddity (talk) 20:09, 19 February 2008 (UTC)

This page needs an overhaul in general. I still believe some sort of merge (archived proposal) with Wikipedia:Look it up is worth thinking about. -- Quiddity (talk) 07:20, 19 February 2008 (UTC)

Merging makes sense to me. Libcub (talk) 18:03, 19 February 2008 (UTC)

Has now been merged. Lee∴V (talk • contribs) 00:38, 20 December 2009 (UTC)

Also, why is "Notes" indented to the left of the other level 1 headings? Libcub (talk) 18:03, 19 February 2008 (UTC)

Fixed. -- Quiddity (talk) 20:09, 19 February 2008 (UTC)

Removed section

I'm removing this section (and copying below), as it doesn't appear to do anything useful currently (development halted since March 2007). -- Quiddity (talk) 20:09, 19 February 2008 (UTC)

==Internal search engines (New)== Recently, new ''internal'' search engines were created. These search engines can search Wikipedia articles, meanwhile they are user-friendly and are integrated into Wikipedia for a good experience. They aren't as good as external search engines though. ===WikEh?=== WikEh? is a '''Wikipedian-friendly''' search engine that searches the articles you want fast. It also searches images that are on Wikipedia. Users can give it a try by clicking on the link below. * [[Wikipedia:WikEh?/Home/|Search Wikipedia with WikEh?]]

Stop Words

Before when nothing was found after a search links would come up to external search engines such as Google to search the site for the same keyword. This has been replaced with a information box about stop words, which according to the article about them are no longer an issue due to changed software. Can this error be reverted? 76.24.87.115 (talk) 18:22, 23 March 2008 (UTC)

Ouch < | http://en.wikipedia.org/wiki/special:search/ >.

Search used to result in an estimate of how many pages it had thought that it had located; this number had frequently been less than accurate: ostensibly it does not communicate w/ itself very well. Now, that number, in the past several days, has disappeared.

Why?

[[ hopiakuta Please do sign your signature on your message. ~~ Thank You. -]] 01:55, 24 March 2008 (UTC)

Numbers have returned,.... How? Why?

[[ hopiakuta Please do sign your signature on your message. ~~ Thank You. -]] 01:30, 25 March 2008 (UTC)

The count is sometimes inaccurate because it is cached and thus delayed by 1 day or more (it reports what was true yesterday). Why it disappeared in Mar '08, I'm not sure - I see it there now. –xeno^talk 22:07, 25 September 2009 (UTC)

//en.wikipedia.org/wiki/Special:Search/ //en.wikipedia.org/wiki/Special:Search/_~_~_~_~ //en.wikipedia.org/wiki/Special:Search/_fullurl

{{ fullurl: special : search/ }}
{{ fullurl: special : search/ %7e %7e %7e %7e }}
{{ fullurl: special : search/ fullurl }}

[[ hopiakutaPlease do sign your communiqué.~~Thank You, DonFphrnqTaub Persina.]] 22:00, 25 September 2009 (UTC)

? –xeno^talk 22:07, 25 September 2009 (UTC)

That character, I do infer, may be intended to imply, "why?" I am obsessed by all & any searchstring, wiki, google, yahoo!, imdb, et-al; I do not know anything about composing code, except that I can copy & paste, & test what changes if I rearrange characters.

&, I do like to cryptic searchstring, as well. &, in google & yahoo! I do like to preformat quotation: " %22 ", as per my preferred searchformat. Everything that I know about searchstring is entirely by experiment, & copy & paste.

My experimentation has resulted in:

• < http://google.com/search?q=%22%22+%22%22+%22%22+%22%22 >;
• < http://search.yahoo.com/search?fr=chr-greentree_ie&ei=utf-8&type=374563&p=%22%22+%22%22+%22%22+%22%22 >, which does seem rather long, thinking that there had been a much shorter one :
• < http://search.yahoo.com/search;_ylt=A0geu6lFRr1KP9AAUINXNyoA?p=%22%22+%22%22+%22%22+%22%22 >;
• < http://akas.imdb.com/find?s=all&q=equal+justice >;
• < http://akas.imdb.com/find?s=all&q= >.

I have just come across wikipedia : administrators'_noticeboard/IncidentArchive299; you should know that on a page that large, when I would hit something like, f/ instance, the rollerkey, the mouse-rollerkey, it occasionally takes several seconds f/ the machine to adjust to that request. Sometimes, the machine would not only freeze, but, crash.

Further, why should I not copy & paste an error message that would destroy an edit that I may have labored on f/ several days?

&, I do repeat, until that hearing, I am making plans, in order to become a much more "disruptive vandal";

it does seem that you do not believe me: so that means that it does need to be something very convincing, very disruptive, very vandal.

&, maybe someone could figure an archive feature that would actually work on my page; &, how, if I would break it somehow, how I might repair it.

There is a stray-right-bracket on the frontpage; I am determined to figure why: I do want to be the one to do it.

Further, why, please, are there two buttons in the wikicode?:

<inputbox> type=search width=42 buttonlabel=Go searchbuttonlabel=Search break=no </inputbox> {{dablink|For a place to make test edits, see [[Wikipedia:Sandbox]].}}

One of those buttons would seem to be extraneous; but, however, you can see how I have renamed them,.....

Both buttons do seem to offer me exactly the same result: either button, same result.

buttonlabel=Go
searchbuttonlabel=Search break=no

Whereas, Jeunes_Agape had gotten my editing started, I do wonder whether they even know what wiki is. I would like to know how to get a message through to them.

[[ hopiakutaPlease do sign your communiqué.~~Thank You, DonFphrnqTaub Persina.]] 01:00, 26 September 2009 (UTC)

Context

Somebody removed the section that said "By default, when a search is performed with no user logged in, context is not displayed in the list of search results. When a user is logged in, context may be displayed, and this parameter is modifiable in "My Preferences". " Why? 74.69.82.49 (talk) 20:24, 29 March 2008 (UTC)

The change in question. I'm not at all sure, but possibly due to a few of the items in these lists of "Fixed bugs" and "Other changes": Wikipedia:Wikipedia Signpost/2008-03-24/Technology report#Fixed bugs. Lots of tweaks to the search lately. That help? -- Quiddity (talk) 21:40, 29 March 2008 (UTC)

I made the change after seeing [5]. When I log out, I get context in searches, also after closing and reopening the browser. Is it different for you? Maybe there is something in my browser cache influencing results but I don't want to clear the cache. PrimeHunter (talk) 02:28, 30 March 2008 (UTC)

Button name change

Hmm. That was odd. Though it's cleared up, I could have sworn the "Go" and "Search" buttons had been briefly replaced by "I'm Feeling Lucky"—a la Google.com—and "Wacky Search," respectively. Ace Class Shadow; My talk. 22:25, 1 April 2008 (UTC)

That was one of many April Fools' Day pranks. PrimeHunter (talk) 23:59, 1 April 2008 (UTC)