Localizing Ubiquity: an open letter to linguists
Localizing Ubiquity: an open letter to linguists from mitcho on Vimeo.
Below is a transcript of this video. Please distribute this video far and wide to anyone who may be interested. ^^
Dear linguist,
My name is mitcho and I am currently involved with an exciting open source community project at Mozilla Labs called Ubiquity. The project’s goal is to empower the user to accomplish more on the web, and to do this, Ubiquity connects disparate services from across the web with language. Let me show you what I mean.
Example 1
Suppose you’re emailing a friend to tell them where to meet later. It would be great if you could insert a map in your email, but you would regularly have to go to a maps application and insert a link.
But with Ubiquity installed, you can just pull up Ubiquity with a keystroke, enter map Shinjuku, pick the right result and rearrange it, and click “insert map.” It’s that simple.
Example 2
Let’s say you’ve stumbled upon an article in French and you want to send a bit of it to a friend. Again, you could just send a link, but with Ubiquity we can do much better than that. We’ll first select the quote we want and translate it to English. There we go. Now I’ll select the bit I actually want to send, pull up Ubiquity and execute email this to Aza. Because Ubiquity interacts with my address book, it knows who I mean by “Aza” and composes my email for me.
As you can see, the Ubiquity interface makes it easy to accomplish tasks with minimal interruption to a user’s workflow. New verbs are easy to write and there’s already an active community writing new verbs and sharing them with users.
The localization of Ubiquity
Right now Ubiquity offers a basic English parser and an experimental Japanese one, but Firefox itself currently ships with 55 distinct localizations. We hope to localize Ubiquity into a number of these languages.
In order to accomplish this, we need support from viewers like you.
For each language—whether Basque, Polish, or Esperanto—we hope to build a parser to take the command and pick out its verb and the arguments and then identify the semantic roles of each argument. Ubiquity will then apply the verb to those arguments and execute.
We’ll need both data and expertise into the languages we hope to localize. If you’re a native speaker or researcher, what is the structure of commands in your language? How do you code for different kinds of arguments? What do your pronouns look like? We’d love to see blog posts, discussions, or quick mockups on these and other topics.
For those of you with NLP experience, we’d love to get more folks involved in writing these parsers. And in the near future, we’d appreciate anyone who wants to simply try out Ubiquity in their language and give us feedback.
The idea of using natural language to direct computers is far from new. But in this limited context of individual commands—single imperative clauses without complexities such as negation or quantification—I believe there is great potential for Ubiquity to support other languages and bring this interface—this new way of interacting with the web—to more people in more communities.
ubiquity.mozilla.com
bringing linguistics to a browser near you
wiki.mozilla.org/Labs/Ubiquity/i18n
help us localize the future
Related posts:
- Localizing Commands for Ubiquity 0.5
- Localizing Ubiquity: commands and nountypes
- Ubiquity i18n: questions to ask
- Ubiquity in Portuguese
- Ubiquity Localization: What’s New, What’s Next
Related posts brought to you by Yet Another Related Posts Plugin.
Tags: demo, i18n, linguistics, localization, Mozilla Planet, ubiquity, video
If you enjoyed this post, make sure you subscribe to my RSS feed (optionally with tweets from my Twitter)!
March 3rd, 2009 at 11:13 am
Dear Ubiquity developers: our research group has developed two corpora (one of Spanish and other of Catalan) annotated at syntactic (constituents & functions; a dependency version is also available) and semantic (Named Entities, wordnet synsets, Semantic Roles) level. There are also available two verbal lexicons of about 2.000 entries. You cat get them at: http://clic.ub.edu/ancora
March 23rd, 2009 at 11:01 am
How do you plan to deal with agglutinative languages such as Hungarian? Hungarian has no dominant word order (all six permutations of SVO are possible for both assertive and interrogative sentences), so to identify the parts of a sentence, one has to do morphological analysis on the words, which can get extremely difficult and messy.
September 8th, 2009 at 8:50 am
to identify the parts of a sentence, one has to do morphological analysis on the words, which can get extremely difficult and messy. Assignments | GCSE Coursework | Dissertations
September 8th, 2009 at 8:50 am
to identify the parts of a sentence, one has to do morphological analysis on the words, which can get extremely difficult and messy. Assignments | GCSE Coursework | Dissertations
October 3rd, 2009 at 2:21 pm
so to identify the parts of a sentence, one has to do morphological analysis on the words, which can get extremely difficult and messy. tiffany jewelry
October 31st, 2009 at 8:29 pm
Very useful source, thanks.
November 6th, 2009 at 2:01 pm
Yay speaking of Hungarian, now that's difficult indeed. I speak it but boy the grammar is difficult as hell…
November 12th, 2009 at 6:31 am
The post is no doubt an eye opening one, there are many questions in every one's mind that how do you code for different kinds of arguments? What do your pronouns look like? We’d love to see blog posts, discussions, or quick mockups on these and other topics.
December 22nd, 2009 at 12:03 am
I think that is a great idea! I'll book mark and check back.
Jenny
January 7th, 2010 at 2:31 am
They some times are
January 7th, 2010 at 1:38 pm
Been a reader of your site for some time now, keep up the great work
January 7th, 2010 at 9:45 pm
Yeah I tried that in school and could not grasp that language, very hard.
January 8th, 2010 at 4:28 pm
Great article
January 9th, 2010 at 2:57 pm
Hungarian has no dominant word order (all six permutations of SVO are possible for both assertive and interrogative sentences), so to identify the parts of a sentence, one has to do morphological analysis on the words, which can get extremely difficult and messy.