The center for all Wikitravel images!

User talk:LOCODEbot

From Wikitravel Shared
Jump to: navigation, search

Hi Botman,

I had to turn you off for a moment because you may be creating categories for places that don't exist. I'm going to talk to your poppy and let him figure out what to do. -- Andrew Haggard (Sapphire) 14:45, 3 December 2006 (EST)

Certainly at the very least, the coverage is weird. There is no Gorham (New Mexico) that I am aware of, and my knowledge of the state is pretty darn comprehensive. There are some other entries for New Mexico that the bot has created for "places" that are really just wide spots in the road. Meanwhile some considerably more significant places -- Silver City, Truth or Consequences, Raton, (particularly) Roswell, to name a few -- are not there. How is this thing deciding what to include and what not to? Seems like some algorithm tuning is needed. -- Bill-on-the-Hill 15:28, 3 December 2006 (EST)
The bot is only up to "M" in the alphabet; thus, places starting with R, S and T have not yet been added. I just checked the queue, and all the cities you've listed will be added, if we turn the bot back on. --Evan 19:38, 3 December 2006 (EST)

NM issues[edit]

Moved from Category talk:New Mexico by Evan

How did these subcategories come about? There's no such place as Gorham, New Mexico; the closest thing to it is an obnoxious politician named Gorham, and the coordinates cited place the "town" in New Hampshire (where there really is a Gorham). Several of the other places cited are incredibly tiny and unlikely ever to lead to WT articles, let alone photos illustrating same. Explanation would be appreciated. -- Bill-on-the-Hill 11:17, 3 December 2006 (EST)

There's a discussion on Wikitravel Shared:Travellers' pub, but I'll re-iterate here. These are seed categories that come from the UN/LOCODE list. The goal is to eliminate the tedious task of creating a comprehensive list of categories, which is kind of a waste of humans' time. As I mentioned before, the LOCODE database is a nice list for potential destinations, as a) it's got hierarchical info and b) it's not copyright-encumbered and c) it's a code for business systems, so the cities listed must have some commercial importance. Typically this means they have to have some sort of transportation terminus like a train station, a bus depot, or an airport.
I understand that some of these locations will probably never have a Wikitravel article about them; however, I don't think we need to have the same concerns about shared: categories as we have about articles in any of the language versions. If there is an empty category in shared:, no harm done; an empty article in en:, though, invites work. IF we get a photo uploaded for one of these teensy towns, there's no damage in categorizing it with the town it's in (rather than making the uploader figure out the right place to put it).
However, if you think there's value in deleting any of these categories, feel free.
There are plenty of problems with the data set; many countries, for example, don't have a lot of destinations listed, and other countries don't have an intermediate level (like province or state) between country and city. I think that where it's important we can move things around much easier than creating them from scratch. That's been the feedback so far from people, which is why this process has gone ahead.
You have a good eye finding the error on Gorham NM; it is shown in the list as being in NM, although the lat/long is clearly for Gorham NH. I'll forward the correction to the UN; I've already sent them a couple of fixes. With a dataset of about 50,000 entries, I think the quality is pretty good.
Andrew has suspended the bot until you've had a chance to read this and agree; if you're OK with it, change User:LOCODEbot/Run to exactly "yes" and it will resume its work. Let me point out that this has gone through 2 rounds of review (text output version and on wikitravel.org/review). It takes days to do the upload, and the longer we're in this halfway stage, the more confusing it's going to be for people. But if you still have serious reservations, let's continue to discuss. --Evan 17:52, 3 December 2006 (EST)
I've turned the bot back on, but as I browse a few other states I know something about, the same issue seems to exist elsewhere: the bot's ability to recognize a "significant" location is rather poor. The assertion that "LOCODE is a nice match for our potential destinations" strikes me as giving it more credit than it deserves. However, manual fixes are always possible. (Incidentally, Evan, my admin bits don't seem to have transferred here from en:. Can that be fixed?) -- Bill-on-the-Hill 20:15, 3 December 2006 (EST)
Aside from Gorham I found Category:Camp Dennison particularly delightful as its really just a historical site. True, it is kind of a 'community', but it's been eaten up by surround communities. -- Andrew Haggard (Sapphire) 20:27, 3 December 2006 (EST)
Perhaps I should have said, "A nicer fit than any other geographical database I've been able to obtain." For example, the US Census Gazetteer gives 23,000 towns and cities in the US, compared to the 10,000 in the LOCODE set.
No other organization is going to have exactly the same criteria for including locations in their database as we will. I tried to get as close a fit as possible here; the rest is going to depend on hand-tuning. I'd much rather have too many categories than not enough categories. --Evan 20:52, 3 December 2006 (EST)
That's fair enough, although I think it's unlikely that more than half of these locations will ever develop much in the line of images, so there will be some clutter. As for other places I know something about:
  • Category:Arizona looks reasonable -- some really tiny places are mentioned, but nothing important (other than national parks and monuments, which just aren't in the data base, apparently) seems to be missing
  • Category:Colorado has some definite omissions that are already mentioned in en:WT guides
  • Category:Illinois has at least one major glitch -- Bloomington (IL), pop. 60,000 or so, is the heart of a Standard Metropolitan Area, can't imagine how it got missed -- and several minor ones
  • Category:Michigan looks quite good indeed.
No obvious explanation for the unevenness of these, but we'll make do. Would there be any value in/need for a "lacunae" page for identifying the odd omissions in this thing? -- Bill-on-the-Hill 21:16, 3 December 2006 (EST)

Suggestions[edit]

With regard to http://wikitravel.org/en/User_talk:Evan/Jan_2005#UN.2FLOCODE

  • could the bot implement some interface for incoming links?
  • I suggest to change in Template:locode "UN/LOCODE: CCTTT" to "UN/LOCODE:CCTTT". The second is more common an will facilitate searching. It is also the same writing as in the wikipedia redirects.
  • see also http://unlocode\.hmap\.info/?loco=CHGVA which gives links to wikipedia if article exists. The owner probably would also insert links to wikitravel.
  • there is a new issue of UN/LOCODE out, now 58 000 entries

Tobias Conradi 06:13, 11 April 2008 (EDT)

I changed Template:Locode, removing the space and linking to UN/LOCODE with little info how to use it for search in wikitravel or google. I also started coding at http://geocodes\.info/UN/LOCODE:DEBER , currently it shows some basic info and some related stuff. With more and more locations covered by the codes, more people might use them. Tobias Conradi 06:44, 11 April 2008 (EDT)

Variants

Actions

In other languages