Popup Chinese Translator for Firefox!!

[flameproof]

Version 1.01 is out on Worpress.

Traditional characters still don't work.

[necroflux]

Yup, I just emailed Justin. Not sure how he tested but not working here either. Probably a small mistake.

[hunxueer]

thanks everybody who worked on this and the upgrade and made it available to us on the mainland

much appreciated

[flameproof]

This morning Firefox informed me automatically of a new Perakun version online - 1.02

1.02 works for both, simplified and traditional characters.

You can get it now also from the official site:

https://addons.mozilla.org/en-US/firefox/addon/3349

谢谢

[necroflux]

Works beautifully! Thanks!!!!

[ABCinChina]

Wow, this new version is the shiatsu!

I'm loving the new colors.

[creamyhorror]

1) The colors and fonts are the default Rikaichan colors. They're nicer than the previous choices, though

2) I understand that this is a mouseover popup dictionary and I can't expect too much of it, but its dictionary database is quite lacking compared to Edict for Japanese, which Rikaichan uses. For some reason, I often stumble the most ridiculous definitions for words possible: "scilicet"? Other strange Latin or incredibly technical words? It's like someone chose the worst possible options out of the lot and used them as the *only* definition. Add to that how the pronunciations given are sometimes wrong, and no alternative pronunciations are offered (especially for individual characters, for which it outputs the wrong pronunciations quite often - last one I encountered: 折), and the tool becomes a bit hampered by its spotty database (imo).

Are the StarDict format dictionaries just too hard to convert? How about Dict.cn, which has been in my experience both comprehensive and accurate (not to mention it has example sentences too) - is a lookup interface available for that? I tried using an addon called DictCN but it worked terribly.

Sorry that I'm whining without contributing anything It's just that StarDict is much more useful with its set of dictionaries (Dict.cn, 21世纪, 高级汉语, 现代汉语) and I couldn't help lament that fact.

edit: I just installed the latest version (mine was 0.2.2 or something) and from a few minutes of testing it seems to find words noticeably better on a random page I tried. I see it now offers alternative definitions too, by default (didn't know it was capable of doing that in the past). Was anything major changed? I guess the Adso dictionary is a more current version, so maybe that's also helping.

[Luobot]

I hate to be the one who points this out but the new version of Perapera-kun doesn't provide translations that are as good as the older version.

Here are just a few examples:

说 shows shui4, not shuo1

行 shows hang2, not xing2 or xing4

单 shows shan2, not dan1 or shan4

假 shows jia3 not jia4

那 shows na3, not na4

困 has "a bunch/tie together" but not "sleepy" or other definitions that are in the older version and CEDICT

问题 has "problem, issue, topic" but not "question"

正 has zheng1 "Chinese 1st month of year" but not zheng4 "just (right); main; upright; straight; correct; positive; greater than zero; principle) as in the older version and CEDICT

There are other characters that don't provide any information at all (as in the older version) such as:

牀 chuang2

唔 wu2

In addition to the translation issue, the older version of Perakun highlighted all the characters it was translating, which was helpful. The new version only highlights the first character, which is not as helpful.

1 - Are these problems related to the version of Firefox that the user has installed, and I'm the only one seeing this? (I'm hoping to stick with Firefox 2x until it's desupported.)

2 - If these problems are not related to the version of Firefox, then is there any plan to fix them?

3 - Since CEDICT is updated regularly, is just a text file in a stable, fixed format, and is easy to download, a great update for Perakun would be an import feature that allows the user to import into Perkun the latest version of CEDICT that the user downloads. Any chance of it?

[ABCinChina]

Luobot, since I'm quite familiar with the CEDICT format, I believe whoever formatted the CEDICT file for Perakun wanted to save time by erasing duplicates instead of combining them. There are many duplicates in the CEDICT dictionary and I think it takes a bit of time to format it if you want it perfect.

[c_redman]

I note that all those examples, except for 问题 have more than one pronunciation. These words would exist one multiple lines in CEDICT, one for each pinyin. So, yes, he didn't combine multiple records for the same character. From spot checking, it looks like it's always the first occurrence of the word in the dictionary. As for 问题, the "/problem/issue/topic/" definition was in the 2007/03/31 version of CEDICT (from mdgb.net), but was changed to "/question/problem/issue/topic/" in the 2007/05/20 version. So we have an upper limit on the dictionary version that was used.

In my own applications that use CEDICT, I just append the multiple values with semicolons in both the pinyin fields and English fields, whenever they differ. For example, "假 ~ jiǎ; jià ~ fake/false/artificial/to borrow/if/suppose; vacation". Occassionally, this can result in ambiguous definitions, but for my purposes it's close enough and convenient enough that I accept the tradeoff.

Just to note, earlier versions of Chinese Perapera-kun used data from Adsotrans, while this new version's is CEDICT. But I didn't see this difference noted in the Changelog for the plugin.

[flameproof]

One way to have a workaround for this "problem" is to edit the CEDICT manually and list the most common definition first.

[ABCinChina]

The reason why I say CEDICT has "duplicate entries" is because an entry for 為 would look like two separate entries. When importing into whatever program, it would either ignore the subsequent entries or import them both. I think importing them both doesn't make sense for a pop-up translator. (I forgot exactly how the format goes but it looks something like the two entries below for 為 which is only one word)

為 为 [wèi] /because of/for/to/is/

為 为 [wéi] /as (in the capacity of)/to take sth as/to act as/to become/

To C_Redman, I personally prefer to make the entry for 假 to look like...

[jiǎ] /fake/false/artificial/to borrow/if/suppose/[jià]/vacation/

This way, when you point out 假, you get...

[jiǎ]

fake

false

artificial

to borrow

if

suppose

[jià]

vacation ...as the definition in the pop-up translator

[Luobot]

Just to note, earlier versions of Chinese Perapera-kun used data from Adsotrans, while this new version's is CEDICT

Actually, following the earlier conversation in this thread – especially noting post #80 and #85 -- I’m under the impression that Trevelyan was kind and generous enough to hook Perakun up with an automated update from adso. Assuming this is what in fact happened, then maybe something was lost in this process?

The reason I mentioned CEDICT (in post #108 ) is because CEDICT provides a readily available benchmark to check Perakun against and CEDICT’s current translations seem more complete than those provided by the current version of Perakun. I also mentioned CEDICT because it’s mostly reliable, is frequently updated, and is completely open source. Using CEDICT with an import feature gives the users the power to update the most important part of Perakun, its dictionary, as often as the user feels like it without having to worry about external entities.

On the issue of duplicate entries, as long as there are any differences between multiple rows, the entries are not duplicate, and they should be shown. This is especially true of the key data – that is, the traditional characters, the simplified characters, and the pinyin. It may as well also be true of the definitions, for the purpose of Perakun, on the theory that it’s better to have something not needed than to be missing something that is needed. Of course, if there are 100% duplicates, then they should be discarded. As to how they should display, the older version of Perakun is very clear. For example, the older version displays 为 as:

为 [wéi]

>> act as, take … to be, to do, to serve as, to become

为 [wèi]

>> because of, for, to

plus some other similar definitions that may be redundant, but aren’t in the way, especially as they are listed further down, as separate entries. Duplicates were never a real problem with the older version of Perakun. The main problem was being able to update it, as other open source dictionaries, like CEDICT, got updated. The other little annoying problem, which still exists, is that Perakun doesn’t work across tabs in Firefox and needs to be turned on and off in each tab. Otherwise, it was perfect for browsing in Firefox.

[ABCinChina]

Has anybody tried out the new pop-up translator based on the adso dictionary? If so, how does it compare to Perakun?

[c_redman]

Having looked at the jar file from the plugin, it does appear to be data from adso, and not from CEDICT. My mistake.

[trevelyan]

@ABCinChina:

You'll need an account to download the custom Firefox dictionary plugin we've released, but we've been getting good feedback on it from our users. Details on how to get a premium account are available elsewhere on these forums so I won't go into it here. The major differences are that the Popup Chinese plugin:

• has a simpler interface
  
• is read/write (click entries to edit them, highlight new words to add them)
  
• translates numbers
  
• does not display Japanese data for hanzi
  
• works across tabs and across pages until deactivated (but not in iframes, strangely)
  
• saves edits and additions back to the user's custom vocab list on Popup Chinese, ensuring they are incorporated in future releases of the plugin.
  

Whether these are positives or negatives depend on what people need. We're trying to provide support tools and resources to help people move from basic conversational Chinese to actual fluency, and the plugin is designed to help people in this situation: people who work with documents on a professional basis and find themselves needing to edit/add materials in their reference tools. A single translator can share their plugin with others to ensure consistency in translation terminology and get a small team helping them generate content. This can even be used by other dictionary projects if desired.

The need to register at present is because of the custom dictionary generation. We'll be putting out a free version shortly without this feature and will release this once we have enough feedback to know what changes need to be made.

[trevelyan]

saves edits and additions back to the user's custom vocab list on Popup Chinese, ensuring they are incorporated in future releases of the plugin.

Should clarify that this means future releases of THEIR plugin, not necessarily the main release.

[Konrad]

This plugin is great for tagging words that you don't know so you can review them later on. It would be nice if the process could be made a bit more convenient.

Now when you want to add a word to your dictionary you have to perform the following steps:

1) click on the word

2) click on one of the items in the popup (for example the Chinese characters)

3) click on the Edit button

4) close the popup with the [X]

I'd like to suggest adding an 'Add to vocabulary' link to the popup dialog, which would also close the popup, resulting in the following workflow:

1) click on the word you would like to add

2) click on the 'Add to vocabulary' link

[trevelyan]

Konrad,

Hit "A" when the popup is open/visible and the entry will be submitted automatically (no need for editing). Already implemented so it should work with the version you already have installed. If not just upgrade to the latest. Will eventually write all of this up for the downloads page.

--dave

[Konrad]

Dave:

Even better, thanks!