[Corpora-List] Sentiment Analysis "without" Text-Corpora

[Axel Oepkemeier]

Dear Members of the CORPORA list,

I would like to ask the community one short question:

Have you ever seen some (research-)documents about sentiment analysis 
which are using commercial hosts and their databases (e.g. 
Thomson/Dialog) as a source for textual data and NOT using web content 
gathered in a Corpus respectively?

Why?
I have done some very basic sentiment analysis without actually "having" 
a Text-Corpus with the method of formulating appropriate queries on 
Dialog-Databases like:

"SEARCH (concept/idea/product/etc.) (Proximity-Operator) 
(positive/negative occupied words & phrases)" (simplified).

My intention is to create sentiment time series, without the claim of 
measuring too precisely. A more aggregate level (subjectivity at the 
level of individual documents or articles and above by summing up the 
numbers of hits) is intended.

I have waded through a fairly huge number of papers and web documents - 
and found nothing.

Maybe someone can give me a hint or point me to a direction for my 
further search.

Best regards,
Axel

-- 
Axel Oepkemeier

Axel.Oepkemeier at stud.fh-koeln.de
Student of Information Economics
Fachhochschule  Koeln / University of Applied Sciences Cologne
Fakultaet fuer Informations- und Kommunikationswissenschaften
Institute for Information Science