Coin Compendium testing begins!

[DiggingNorway]

I really like the queru "look up" that automatically searches if a specimen is already registered, saves a lot of hazzle.
But, when it comes to generally moving up and down the types and subtypes I dont understand why the "curtain/drop down" window has to be hidden by default, and to dig deeper one has to click the hand.... I usually click this hand 4 times before I get to the level I want, why not just leave the level open?

[DiggingNorway]

:)

[DiggingNorway]

my suggestion would simply to shift to having the "open" as default, and add a "hide/close" button/hand, so that once one dig deeper the types are open, makes it faster to go to the next level. If the list too big, then one can simple "close/hide" the list.... so just the reverse as now...

[fwang2450]

I tried EN->Arabic and it looked really cool. My guess was 10% didn't get translated at all, the other 10% suspect. EN->Chinese (either form) did take some creative guessing for about 20% of the content. If you have to grade the EN->Simplified, what % did it get right?

I would say very low. In this whole paragraph  "This is where you can put a brief description or full article information about this page. You can delete this line and replace it with your information. Please do not delete anything else", none of the sentences is translated correctly. Even the translation of the last sentence misses "else", which contradicts the previous sentence.

[badon]

The auto-translation system looks promising. My initial estimate of its rate of mistranslation may be 10% to 20% that is similar to my experience with Google Translate. Can anyone offer a better estimate?

What language pair was it? The EN->Simplified Chinese translation is not good. I can only guess at the meaning.

I would say very low. In this whole paragraph  "This is where you can put a brief description or full article information about this page. You can delete this line and replace it with your information. Please do not delete anything else", none of the sentences is translated correctly. Even the translation of the last sentence misses "else", which contradicts the previous sentence.

I tried EN->Arabic and it looked really cool. My guess was 10% didn't get translated at all, the other 10% suspect. EN->Chinese (either form) did take some creative guessing for about 20% of the content. If you have to grade the EN->Simplified, what % did it get right?

Arabic is a beautiful written language. Thai and Cambodian scripts are quite attractive too. I wish I could read them :)

I took some time to work out a few details of how the translation system should work. It won't be as simple as I was originally hoping, but I think I have formulated a strategy that should allow translation of most of the site. Thankfully, all the extra time I spent making my code flexible and easy to modify will pay off now, because the translation system will require massive changes in how the site operates. But, it should be straightforward to do, and I might be able to finish it in a few weeks to maybe a month of focused effort. My estimates are usually overly optimistic though :)

The point is that, from a technical point of view, I think the problems I initially ran into can be solved. It will take more study and lots of testing to confirm that, and I will have to evaluate a few other options too, in case another strategy might be better for making the CC "future-proof". The strategy I'm thinking of using now will tie translations to one system. If something goes wrong, and we have to switch to another translation system, it could be a nightmare. I'm going to be careful before I decide to commit the CC to the translation system that I've chosen. So far, I think we should be OK committing to the current translation system. I just want to explore alternatives and plan ahead a little before making the final decision.

I think the translation system is working well enough I can set up a few people as translators, and you can start breaking stuff :) I'll work on that right now, and let you know when I'm done. For testing purposes, please stick to translating only the main page, so when stuff breaks, it will be easier to find it and fix it.

DiggingNorway, do you think Norwegian translations would be helpful? As I understand it, close to 100% of Norwegian speakers also speak English. But, since the CC will probably eventually cover Norwegian coins too, it might be good to at least test Norwegian translations, if you want.

[DiggingNorway]

Most Norwegians speak and read English... we do have some special letters: æ, ø and å. The only of which is needed for coins are the "ø" as in "øre", which equal "cents"

I wouldent prioritze that if I were you :)

[fwang2450]

I will take a look during the weekend, and try to translate the main page into Simplified Chinese. I will report any issues I run into. But before you open the translation function to a larger group, a couple of things need to be done first. We need a glossary for the language, and we need a group of qualified reviewers for each language. There must be a mechanism to identify un-reviewed pages, too. I can finish up to year 2000 on the Simplified Chinese glossary over the weekend. The rest may take another week or so.
One issue I ran into working on the glossary is missing English terms. There are some coins that are missing from the CC and even from the NGC census. I will just leave TBD in the glossary. Hopefully they can be replaced by English terms later.

[badon]

Unfortunately, one of the things that won't be possible to translate without a redesign of the site are the type names. There are plenty of ways around that issue, though. Once you get a chance to look over the translation system, I think you'll be pretty excited about how capable it is. It has all the reviewing features you have asked for, and more.

The only thing that needs to be done to make the review features effective is restricting review rights to fewer people. Right now, it's open to everyone with an account on the site, because there's only a few people to worry about.

This translation system has a translation memory feature, which might reduce or eliminate the need for a separately maintained glossary. I don't think we'll be able to see the translation memory feature until we have some translations done. In order to make the site useful to people as quickly as possible, translations get published immediately, with no delay while waiting for someone to review them. The reviewed and unreviewed translations are separated in their own lists, so it should be easy to see.

There's also supposed to be a machine translation feature too, but I haven't figured that out yet either. For now, we have the machine translation button on all pages, so I have left that as a low priority for now.

The translation system is designed to work with hundreds or thousands of translators simultaneously, in a highly organized, efficient, and productive way. I think you will like it a lot.

[badon]

This might be interesting to the Chinese translators:

http://china-mint.info/forum/index.php?topic=4585.msg34894#msg34894

I wonder why NGC changed the spelling from "Lao Tse" to "Lao Zi"? Maybe it was to conform to Pinyin better?

[dobedo]

This is odd when I fooled around with some translation on the main page as suggested:

1. Using Google Translate, I translated "Main Page" into Chinese via the translation system. Saved it, and it looksedgood.
2. Then I translated "Quick Index" and saved it again.
3. Go back to the Main Page, the translated "Quick Index" appeared ok, but the previously translated "Main Page" only showed up for a few seconds, then reverted back to the original English "Main Page" as if it couldn't made up its mind.

The revision history recorded both, but the first translation seemed lost, as I was.

[badon]

Your translations were saved successfully. What you saw was Google Translate doing something to the page. If you close the Google Translate bar at the top of the screen, and refresh the page by clicking the reload button in your web browser, it should show everything correctly.

[fwang2450]

This might be interesting to the Chinese translators:

http://china-mint.info/forum/index.php?topic=4585.msg34894#msg34894

I wonder why NGC changed the spelling from "Lao Tse" to "Lao Zi"? Maybe it was to conform to Pinyin better?

"Lao Zi" is the standard Pinyin, while "Lao Tse" was the spelling used before Pinyin was invented, i.e., before PRC. This applies to many names. You mentioned Sun Tse, but the standard Pinyin is Sun Zi. Beijing was Peking before the PRC. In some cases, the old spelling has been left alone, like the Yangtze River, which in Pinyin should be Yangzi River. In terms of PRC coins, the trend looks to be adopting the standard Pinyin spelling.

[badon]

Well, I guess we should follow suit. I noticed that the language code for pinyin is not available in the dropdown list of language translation targets, but I think there's a way I can add it manually. I'm not sure if we'll need it though.

[fwang2450]

Badon, I just tested a few lines and here are my findings, from a professional translator's perspective:
1. The translation tool is not easy to use. The page is not updated instantly to reflect the translation. Debugging is difficult as one has to exit the translation mode, view the page, enter the translation mode, make the change, exit the translation mode to verify.
2. The pages are not internationalized: codes are mixed with resource strings, which is very dangerous because a careless translator may just delete some programming codes and crash the page; programming codes should be filtered out and hidden from the translator. Dates or any other functions called from elsewhere are still in English.
3. No translation memory function. The same or similar strings have to be translated again and again.

Crowdsourcing is normally done through a list provided by the site owner, like what Facebook does. I know that involves a  lot of background work and may not be possible with the CC, but as is, the technical difficulties in translation may discourage users from volunteer translation. I found myself struggling with codes rather than trying to provide the best translation. Is there any way to improve and make the tool more user friendly?

[badon]

It looks like there is a problem in the setup of the translation memories. The problem seems to be that the version of the translation system I installed is riddled with bugs. I'll have to hunt for an upgrade, or see if I can solve the problem on my own. But, the good news is that it appears to be very sophisticated in the variety of translation memories it supports. Do you know of any translation memory services that are open access? I don't think we need them, but if they're available, it might be worth looking at.