I've been thinking about languages again. I talk to a lot of companies, many of them outside the US. Fortunately for me, English is very popular as a second language, which makes the conversations possible. It doesn't always help me with their web sites and blogs, however, and so I find myself making regular use of automated translation services. The piece I'm missing is a reliable way to translate RSS feeds.
Rafe Needleman posted a quick, easy, and—when I tried it—ineffective method of translating feeds using Yahoo Pipes. The titles get translated, but the body stays in the original language. Not much help. I didn't get any farther with Google Translate, although I'm still experimenting with other combinations of translation and RSS services. If you've found a combination that works, I'd like to hear about it.
What we need is a feed translation service, which takes in a feed, translates it, and creates a new, translated feed. With the acquisition of Feedburner, Google has the pieces. Any chance they'll do it?
If the whole idea of machine translation goes against everything you know about language, I know. I'd rather be able to read all those languages, too, but there will always be languages I can't read, and I can't justify proper translations. I can do a minimally acceptable job reading the French blogs, and I can get the general idea with other Romance languages. There will always be more languages that I can't read, and for those, machine translations are a great service, even with their flaws.
Science project challenge
Speaking of languages, I haven't heard from anyone who's tried my translingual influencer analysis science project. If your company has multilingual capabilities and does influence analysis, this could be a powerful demonstration. Can you identify relevant, influential sources who pick up a topic in one language and write about it in another?
Tags: social media analysis influence analysis language translation rss Google
Interesting challenge.
We do influential analysis, and we do have an algorithm to identify the language/locality.
No translation service, though.
I guess that's the thing with these mashups -- they're unreliable. I'm still reading the German Web 2.0 blog in English thanks to the tip I posted. Not surprised it's not perfect, tho.
-Rafe
Rafe, that's a good summary. It's probably not a good sign that the Babelfish module is grouped with the deprecated modules, either.
Bart, the science project and feed translation are separate topics, of course. I considered separate posts, but I decided to go for the "oh, by the way" feel, instead. If you decide to try the experiment, please let me know what you find. This is all conjecture until someone runs the experiment.
We've actually been working on a way to make that possible for people. One of our first ideas was to create a centralized social bookmarking website that makes content available automatically in different languages - and that's where Kontrib.com came about. We're looking to build out additional services down the road for people to use to leverage machine translation technologies. Do you think there's a big need for this in the RSS community?
So sort of Digg + translation. Interesting. I like the idea of making the site more accessible to a broader audience, but I wonder about how many vote-for-content sites people will try.
I don't know about "big need," but I'll use anything that works. I'm sure I'm missing some good stuff.
Here's an interesting post on a related topic, but going in the other direction—a blogger who writes in a foreign language: English.