Dear friend

Greetings.

The so far advancement in the field of sentiment analysis is the development of various types of machine learning algorithms.

There are various textual based systems available.

but still there are several issues in this regard because of the textual difference among different languages.

I know there are several papers available in the literature.

I hope this helps.

Best regards

Dr.Indrajit Mandal, Ph.D.

Thank you Indrajit. I did an overview of the sentiment analyses field. I speak about English for the moment. I would add the topic models to your suggestion, but it still doesn't help. My problem is how to include emotion recognition results in the system's knowledge model. So I should have first the emotion classification/regression results with regards of sentiment analyses.

Dear Velina,

if you did not read yet, I would like to suggest you the works of Mike Thelwall related to 'Sentiment Analysis', because I think that they contain useful information for your study field.

Best regards,

Roberto Saia

http://scholar.google.co.uk/citations?user=8jCKL1sAAAAJ&hl=en&oi=sra

Thank you, Roberto, great! Exactly...

I have read only those that concern short texts on the treads.Could you take a quick  look to the dataset/project here, please? If you have time, of course.

Best regards,

Velina

Data OPENED FOR DISCUSSION Multimodal Sentiments Analyses of Financial News

Of course Velina, and if come to mind other useful information, I write to you.

Have a nice day,

Roberto

Thank you, Roberto,  it is already night here... Waiting for your suggestions, I appreciated your advices a lot. Velina

Pritam : Hey akanksha,will you please take my printout ,because i am suffering from fe ver,so i won't be able to go outside.

Akanksha : Hey pritam,i did my prints out yesterday.

Pritam : okay Good  :(

Which type of Sentence is this??

Positive or Negative or Neutral??

....

Sentimental analysis taking More than one word .Okay and Good gives positive response for analyzing sentence.But smiley at the end gives that Pritam is lil upset with akanksha said.So the sentence  as whole falls under negative sentence.So,Making combination of Words and Smileys together and Weighting more for Smileys is still a research topic.Hope it may help you.Tweets from twitter is the dataset you can take it.

Of course you are right, Pooja.

There are several levels studied till now in sentiment analyses. The sentence level is just one of the levels. Discourse level, document level...

General contextual level... World description level. But they are not studied.

You are right to propose to take typographical features as possible source of the conveyed  attitude.

I should attach the project to my question. I do it now. Thank you.

Hi there, just adding a little contribution. I try to work with psychologists and other experts that can contribute to a better understanding of emotions and sentiment analysis. Some promptly say "people lie !".  This is why i came to conclusion that data mining in text samples is not enough, so if you explore audio and video you are probably in a good track. However my  concern is that news, professional news, should report information and not emotions, also people that read news should not put their personal feelings on the message, which should empty news from emotional contents. Wouldn't that be a problem for your research?

I agree, Fernando, you are right. In any case data mining in text only seems to be not enough powerful for the proposed case of study.

However, I wold a bit modify yours: "However my concern is that news, professional news, should report information and not emotions, also people that read news should not put their personal feelings on the message, which should empty news from emotional contents."

Well, it is clear that professional news should report information. But isn't this information somehow inducing emotions? It is subject-dependent, of course. But how much and till where? If there are common emotions inspired to the listeners and an agreement on the sentiment which is involved, there is a sentiment transmission I guess. I don't know. There should be several clusters of listeners.

I see your point, emotions will manifest in the listener/watcher audience, but then you will need to read some watchers, not the news reader, right?, Another point, that could somehow solve that issue is that 'news readers' cannot avoid to feel emotions and thus to express emotions, even trying to hide it from listeners, unless they completely ignore the contents and just read the text; which i think will depend on the nature of the news but in the case of serious issues don't you think that data mining of the text will be enough? I think the problem is in the borders as for the "worst" news would be easy to identify.

Sure, yes, Fernando. Take a look to the attached link.