Jump to content

Talk:Comparison of machine translation applications

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by 71.155.243.128 (talk) at 00:26, 15 July 2009 (Slightly biased?: new section). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

WikiProject iconComputing: Software Unassessed
WikiProject iconThis article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
???This article has not yet received a rating on Wikipedia's content assessment scale.
???This article has not yet received a rating on the project's importance scale.
Taskforce icon
This article is supported by WikiProject Software.

I started this comparison article which I think is necessary to Wikipedia.. I am currently going to have to pause my work on it due to other obligations.. If there is anyone here whom thinks the table should look differently please correct it or help me fill it up if you feel like it before I'll get back to it later today or tommorow. --Acidburn24m 22:11, 20 October 2007 (UTC)[reply]

How do you recommend ordering the languages on the right? Alphabetically? By number of speakers? - Francis Tyers · 22:58, 20 October 2007 (UTC)[reply]
Also, I've added a free software line, but it would also be ok to split the page into two tables "Free software" first then "Proprietary software" second. - Francis Tyers · 23:08, 20 October 2007 (UTC)[reply]
that's a good idea! I added it to a table of it's own. Acidburn24m 05:39, 21 October 2007 (UTC)[reply]
Ok, but none of the other systems apart from Apertium are free software. - Francis Tyers · 11:38, 21 October 2007 (UTC)[reply]
my attempt to create a good table

Through the last hour I've been trying to think how to get this whole table in shape.. I want people to clearly see in the table which options to they got for each language instead of searching through all combinations of all the languages through a huge list... the problem is that I am not sure how such a list would look like ... I tried scrabbling something in Excel but unfortunately it isn't good because I realized that it would require a lot of duplicated rows which would also only complicate the list.

if anyone got any bright ideas how to get a nice simple small table which would show all the combinations without making any duplicate rows please provide your ideas. Acidburn24m 05:44, 21 October 2007 (UTC)[reply]

There are literally hundreds of systems, how should we decide which are included on this page? E.g. Gramtrans would be one I'd immediately think of for translation between the scandinavian languages. - Francis Tyers · 11:38, 21 October 2007 (UTC)[reply]
It might be a good idea to merge "English - French" and "French - English" into one row, "English - French" and then mark it as unidirectional or bidirectional ? - Francis Tyers · 11:39, 21 October 2007 (UTC)[reply]
sounds interesting.. could you try to create the table as you vision it? Acidburn24m 15:55, 21 October 2007 (UTC)[reply]
Ouf, I'll try if I have some time, but it would require some significant restructuring. - Francis Tyers · 10:30, 22 October 2007 (UTC)[reply]
Try creating your idea in Excel and post an screen capture image here of how you think it should look like (just like I did).. if you do so, I could work on the significant restructuring part of building the wikipedia tables themselves. Acidburn24m 14:04, 22 October 2007 (UTC)[reply]
I don't use non-free software, but I could knock up something in Gnumeric. - Francis Tyers · 11:55, 28 October 2007 (UTC)[reply]

another idea I had is maybe making different tables for the availble MT web applications versus the actual MT programs? --Acidburn24m 19:20, 21 October 2007 (UTC)[reply]

Aye, considering that the big three "Google", "Babelfish" and "Windows Live" all use pretty much the same software (SYSTRAN). - Francis Tyers · 10:30, 22 October 2007 (UTC)[reply]

I am wondering, which other Software based MT applications are there besides Babylon? Acidburn24m 20:44, 23 October 2007 (UTC)[reply]

New Table

After giving it some thought I redesigned the whole table to be only focused on all the languages which one could transform to english and all the languages one can transform from english. I decided to sort the list according to this list to get the most relevent resolts to the actual languages which are used on the internet. I still need help with the coding of the table itself and filling it up. --Acidburn24m 06:02, 25 October 2007 (UTC)[reply]

Alternative

Machine translation application
Apertium Google Translate Babel Fish Windows Live Translator¹ Babylon SYSTRAN 1-800-translate
English to French No Yes² (⇄) Yes² (⇄) Yes² (⇄) ? Yes² (⇄) ?
English to German No Yes² (⇄) Yes² (⇄) Yes² (⇄) ? Yes² (⇄) ?
Romanian to Spanish Yes (→) No No No No No No

I was thinking something like this, using arrows after the yes/no to depict the directions, e.g. →, ← and ⇄ - Francis Tyers · 12:00, 28 October 2007 (UTC)[reply]

looks nice.. Acidburn24m 18:38, 28 October 2007 (UTC)[reply]
now that I finished fixing the table it is the time to think how to implement your idea into it. please let me know what you think should be changed exactly before I start editing it again. Acidburn24m 17:31, 1 November 2007 (UTC)[reply]

Table fixed and data has been entered

I know there is still much data to enter.. at this point I need to take a break entering the data. please take you time to go over the data I entered and varify if it's correct. I would also appriciate if you can enter new data into the empty boxes as well. Acidburn24m 19:01, 29 October 2007 (UTC)[reply]

Ok.. I entered what I could find.. I can't find any list of the Machine Translation languages which Babylon can translate between.. if you find it please fill it in the table. if you have any other ideas for MT websites to add please do add them to the list. --Acidburn24m 00:44, 30 October 2007 (UTC)[reply]
Francis Tyers, still waiting for you opinion about my latest changes.. Acidburn24m 17:29, 1 November 2007 (UTC)[reply]

Does anybody know of any other MT applications which can translate English to Hebrew other than Babylon and 1-800-translate ?

If you do, please let me know. Acidburn24m 05:08, 16 November 2007 (UTC)[reply]

I believe Moses (and probably OpenLogos as well) can do the job, though you need to provide the parallel corpus and patience while it trains... —Preceding unsigned comment added by 62.226.31.103 (talk) 04:05, 26 February 2009 (UTC)[reply]

TRANSLATE.EU

Please add TRANSLATE.EU into article. —Preceding unsigned comment added by 87.226.25.208 (talk) 20:08, 19 December 2007 (UTC)[reply]

Should language pairs that no listed application offers be listed?

The paragraph "Languages features comparison" currently also contains language pairs where no listed translation application offers a translation, e.g. Thai to English, Finnish to English, English to Slovenian and so on - as this list could be expanded nearly arbitrarily (there are countless "missing" languages), I propose to delete these from the table. Gestumblindi (talk) 02:55, 24 March 2008 (UTC)[reply]

Agree. - Francis Tyers · 08:53, 8 May 2008 (UTC)[reply]
Funny question. I mean how many (living) languages exist? Several hundred?
But seriously, I'm not sure if it makes sense at all, since for example Moses, and basically all statistical machine translation software, is potentially able to translate any language pair. The problem is the availability of a sufficiently large and qualitative parallel corpus for the respective pair. Even the commercial engines are very likely based on stochastic methods and the only difference is they bought translated text collections for only a small set of languages and are therefor restricted, whereas the free engines are restricted by the lack of free parallel text collections at all.Though one only has too lock around to find quite a lot of material which can be aligned. A good starting point might be OPUS.
I guess what I want to say with this is, it should be at least pointed out that often it's not the capability of the engine but rather the availability parallel text collections that restrict the offer of languages. 62.226.48.30 (talk) 00:18, 27 February 2009 (UTC)[reply]

Web application vs Software package

A lot of those translation companies offer software packages at a cost. The license section doesn't say whether it is referring to the web application or the software package. This needs to be corrected --Voidvector (talk) 14:02, 20 June 2008 (UTC)[reply]

Yup, that needs a lot of work. It is misleading - SYSTRAN is supposed to be free, web-based software according to this article. FYI, SYSTRAN costs nearly $1000, and the software can do a lot more than the free online service. Actually, the software comes much closer to machine translation than the online thingamajig. Major editing is needed. —Preceding unsigned comment added by 74.15.124.120 (talk) 01:29, 18 July 2008 (UTC)[reply]

citatation needed for existance of platform

In "General information" there is no reference of the "platform"s. They should be added or this column should be deleted. -- 84.132.60.151 (talk) 12:35, 1 February 2009 (UTC)[reply]

Include only original software

1-800-translate was included and I believe it should be removed since they do not have machine translation software and if they ever had it was probably a reseller of systran.

Someone mentioned translate.eu... It is just a mashup of other tecnologies, such as google translator.

Also it is interesting to notice that Babylon is not avalaible for download. It is simply a plugin but not the full translator, so it is basically online (or needs web access).

Other old softwares like "Power Translator" seems missing, althought I do not know if they are still in development... —Preceding unsigned comment added by 81.184.58.60 (talk) 19:57, 18 March 2009 (UTC)[reply]

Web-based and offline

A distinction should be made between (browser, internet-connected offline programs , eg Lingoes) and offline programs. Programs as Lingoes still need internet connection to function yet do not use browser. —Preceding unsigned comment added by 81.246.131.229 (talk) 12:52, 26 March 2009 (UTC)[reply]

Slightly biased?

It appears to make Google translate look good by selecting the languages it's capable of translating and not many others. Maybe the other translation services don't do anymore, so maybe it's an accurate representation but it looks biased. (I personally think Google is the best but that's besides the point.) Also the red linked ones should be removed (their articles got deleted for being non-notable, advertising, or having a expired prod). 71.155.243.128 (talk) 00:26, 15 July 2009 (UTC)[reply]