I am new to IBM Content Analytics product.
Actually I'm using ICA 3.0.
I would like to explore positive and negative expressions (sentiment analysis) in French as it is mentionned in the IBM Content Analytics Information Center/"What's new in version 3.0", that the supported languages are English, Dutch, French, German, Japanese, and Spanish.
By default, I am able to get positive and negative expressions highlighted in English. But, after crawling french documentation, I am not able to detect french positive and negative expressions.
Could you help please?
This topic has been locked.
Pinned topic ICA 3.0: Sentiment Analysis in French?
Answered question This question has been answered.
Unanswered question This question has not been answered yet.
bfoyle 060001WDQ360 Posts
Re: ICA 3.0: Sentiment Analysis in French?2012-11-08T17:48:24ZThis is the accepted answer. This is the accepted answer.Currently Sentiment models are only supported in English and Japanese. This is something that we are looking at for possible inclusion in a future release but in the meantime, you can work with our Services group to utilize ICA Studio to develop your own custom sentiment models and utilize some creative mechanisms to link them back into ICA for analysis and visualization.
If you want to pursue the building of a custom sentiment model, please send me and email and we can arrange to discuss.
Re: ICA 3.0: Sentiment Analysis in French?2012-11-08T18:22:49ZThis is the accepted answer. This is the accepted answer.The Text Analytics Catalog has a Sentiment annotator model that you could use. You would need to find sentiment dictionaries for French, but there are sources available for a fee or free. You would import Positive, Negative, and Negation terms into dictionaries. I don't know if the published model includes dictionaries for "always positive" or "always negative" terms and phrases, but that should be considered. For example, comparators such as happier, angrier, bigger, faster, better, etc. are not negated with a negation term. "I am not happier" doesn't mean that I am unhappy. "I have never been happier" is clearly not negative.
I have a presentation on creating a Sentiment annotator. Let me know if you would like a copy.
Re: ICA 3.0: Sentiment Analysis in French?2012-11-09T09:41:20ZThis is the accepted answer. This is the accepted answer.
- SystemAdmin 110000D4XK
Thank you for your responses.
Indeed, I found in the Text Analytics Catalog that there is a "German User Sentiment". I suppose it is the model you are talking about.
Regarding the presentation on creating Sentiment annotator, I confirm you that I am interested.
Thank you in advance,
Re: ICA 3.0: Sentiment Analysis in French?2012-11-09T11:35:12ZThis is the accepted answer. This is the accepted answer.The attached presentation is a fairly high level guide -- it doesn't go through every interaction, assuming the user understands the basics of using the Studio.
There is a problem with the current version where removing an annotation from a rule only works when covered text is used to create the new annotation. That requires an extra step that will be removed when the problem is fixed.