[Corpora-List] EmoText - Software for opinion mining and lexical affect sensing

[John F. Sowa]

Justin,

I wasn't making a "restrictive" statement.

JFS
>> I doubt that any statistical technique could do much with that data.

JW
> With the exception of dead metaphors, isn't metaphorical language
> almost by definition highly creative [productive]? In that light,
> your analyses seems rather restrictive...

There are many very useful applications for which full language
understanding is not required.  As an example, I'd cite the Summly
iPhone app for text summarization.  See the following interview:

   http://www.bbc.co.uk/news/technology-16306742

Following is an excerpt from that article:

> The app uses an algorithm to recognise what category of information
> a webpage contains by using "ontological detection" to identify
> its nature which in turn determines which set of instructions should
> be used to provide a consolidated summary of its text.
>
> Or to Summlyfy this in Mr D'Aloisio's own words: "It can detect
> different genres or topics of webpages and apply a specific set
> of metrics to them."

 From their web site, http://www.summly.com/en/technology.html :

> Our core technology is language independent and produces great
> results out of the box for any language.  In addition, specific
> optimisations have been made for English, French, Spanish, Italian,
> Dutch, German, Swedish, Mandarin, Russian, Japanese and Finnish.

This quotation suggests (a) that their algorithms are very useful
and (b) that they don't attempt to do language understanding.

But I still stand by my more general statement:

> Fundamental principle: If you want a computer to understand language,
> you need to design a system that can understand language.

I strongly support any method of processing texts that can make
a profit.  The field of NLP needs such applications.  But for the
long term, I believe that the fields of AI and NLP should focus
on their original goal of language understanding.

John

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