On 31 March 2011 01:35, Roman Zegarski <[email protected]> wrote:
> Hello,

Cześć Roman.

> My name is Roman Zegarski (on IRC simply: RomanZegarski ) and I am
> final year student on the Gdańsk University of Technology in Poland
> (Informatics, specialization - Distributed Applications and Internet
> Services).
> I spend some time with topics related to computational linguistics. In
> the past year I worked on student project which target was to build
> virtual student assistant (precisely chatterbot, generating  base of
> knowledge from university moodle serwer. It still need some work, but
> most application functionality is working fine). Currently I am
> working on developement and implementation of words disambiguation
> algorithm using WordNet.
> When I saw Apertium on GSoC, I thought that I can be useful for this project.
> I'm interested in working on "Dictionary induction from wikis".
> I would like to use DBPedia to retrive data form Wikis, also I could
> improve DBPedia extraction framework to retrive more information from
> wiktionary (i.e. using multiple languages).

Wikipedia extraction in the DBPedia framework can more or less be
considered complete; individual infobox mappings are lacking for most
languages, but that's really a human task. Much of it could be
automated, but I don't know if that could be achieved in a three month
project.

The main reason I'm interested in using the DBPedia framework in
particular is because most of the extraction is performed based on
Wikimedia's own template system, and is already set up to load the
template mappings from a wiki (http://mappings.dbpedia.org/). Also,
new ontological properties can be defined on the wiki too - which is
important for extensibility.

There are several other reasons - the DBPedia extraction framework is
robust, the DBPedia web service has a listener on the changes
mechanism in Wikipedia (i.e., it's updated as soon as changes are
made), it was designed to be easily extensible, and it's flexible
about input (it can take XML exports as input, or it can use a
database with the wikipedia tables loaded).

I'm also particularly interested in extraction from Wiktionary, or,
rather, the various language editions of Wiktionary. As is, they are
an often valuable lexical resource, but would be much more useful as a
lexical database -- much of the information contained on each
wiktionary is not language specific, per se, and being able to access
that content in a (relatively) language neutral way would be extremely
valuable.

The big difference is, Wikipedias typically have a similar (if not the
same) layout, whereas Wiktionaries tend to differ wildly. Page layout,
the information presented, how templates are used, etc., are all done
differently. Something like the DBPedia mappings wiki, but for
Wiktionary/ies, would be excellent.

> About my experience: I have done some part time work in C++ and C# on
> commercial projects, and I am experienced in Java from university
> (both projects mentioned eariler are writen in Java).

There would be two parts to the DBPedia extraction project: extracting
the information, and converting that information into a format that
can be used in Apertium. The first part would be primarily in Scala
(though it shouldn't be too hard to pick up), the second would be best
done in Java, using apertium-dixtools as a base.

> Also, could I concentrate work on using DBPedia only? I think lots of
> information can be recived and after improvements of extraction
> framework, using for example OmegaWiki wouldn't bring much more
> posibilities.

Are you sure? :) "440032 expressions for 46042 concepts in 282 languages".

-- 
<Leftmost> jimregan, that's because deep inside you, you are evil.
<Leftmost> Also not-so-deep inside you.

------------------------------------------------------------------------------
Create and publish websites with WebMatrix
Use the most popular FREE web apps or write code yourself; 
WebMatrix provides all the features you need to develop and 
publish your website. http://p.sf.net/sfu/ms-webmatrix-sf
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to