[Corpora-List] Quotable Statistics on Unstructured Data on the WWW

[Gilles Sérasset]

Hi Daniel,

On 6 déc. 2013, at 13:12, Daniel Gerber <dgerber at informatik.uni-leipzig.de> wrote:

> As what do you see text then? Yes, I typically refer to text as being unstructured, tables and so on as semi structured und databases as structured. 

Seeing a structure as a set of lists or table is a very database centric point of view.

I do agree with Adam, as I think a text is very structured. It is a medium that brings me a lot of data I can grasp from it. Maybe this structure (and the data it conveys) is not easily accessible to a computer, but this only proves that computers are dumb devices...

Structures are not only tables...

I'd rather say that a database is an explicitly structured data, while a text may be seen as implicitely structured, but I also think that databases may not represent most data that is conveyed by text.

> I’m sorry that you feel greatly offended by my understanding. But your reply does not answer my question nor does it help me to understand a different point of view any better.

Well, as far as your question is concerned, if you want to compare the size of explicitely vs impliciteley structured data, then you need a common unit of weight. What would you take ? Size in Kb? Well this does not count the data, but rather the size of the signal that conveys the data...

If you have a corpus of say 10000 docs written in English. And a corpus of the very same 10000 docs SPOKEN in English, then the data may be considered the same, so you have as many written data as you have spoken data. The fact that the spoken data will weight 1 Gb while the written takes 10M is just irrelevant. 

Regards,

Gilles,

--
Gilles Sérasset
GETALP-LIG                         BP 53 - F-38041 Grenoble Cedex 9
Phone: +33 4 76 51 43 80                   Fax:   +33 4 76 63 56 86


_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora