I have done my PhD in sentiment analysis, and I have not seen any sentiment analysis using Bayesian network. There maybe some, but I did not find any. 

I am wondering, is there any reason that you are looking for Bayesian network? Is there any particular application or domain? 

Regards

Mostafa

@Mostafa Karamibekr,

thank you for your time and your answer.

Actually I’m working on sentiment analysis and opinion mining for modern standard Arabic and I’m trying to find if any works have been done on it using Bayesian networks and probabilistic models

As Mostafa indicated earlier, I have not come across any software that apply Sentiment Analysis using Bayesian network. Maybe if you dig deep in R packages something will come up but I doubt that since I used R for a small project using Sentiment Analysis.  

I was working on sentiment analysis in my PhD and analyzed different classifiers including NaiveBayes. Admittedly, there are some reasons to prefer rapid classification using NB rather than analytical using SVM. However, if you are looking for a perfect classifier, I doubt you would find -- all classifiers (NB or SVM) calculate comparable results. Sometimes a classifier is better, sometimes worse. Even if you filter features using InformationGain.

Anyhow, I used WEKA for classification -- it is open source.

http://www.cs.waikato.ac.nz/ml/weka/

Thesis Opinion Mining and Lexical Affect Sensing

Even i am doing PhD in sentiment analysis and yet i haven't find any one used bayesian network. Yes, Naive Bayes classifier used many but not bayesian.

well, I m diging deeper and deeper every time, trying to find anyone that used bayesian network but still can't find. Also my teamwork in our lab doesn't stop saying that it could not be done with Bayesian network, especially for such a morphological rich language as Arabic.

So I am trying to find any one who is interested in this subject, to work with and cooperate.

For my PhD, I focused more on linguistics approach rather than statistical approaches such as machine learning techniques or others. My focus was on not just polarity classification or subjectivity classification, I was working on recognizing opinions, extract them, create a structure for them, and then analyse them. While SVM and Naive Bayes only classify for example the polarity into positive or negative. 

 To validate my result, I implemented Naive Bayes and SVM with two different features sets, but not Bayesian network which is different from Naive Bayes. 

I hope it helps.

Mostafa

Houssem, from all features I studied in my thesis (lexical, deictic, stylometric, grammatical), the highest classification results are calculated using datasets with lexical features aka Bag-of-Words. Since I don't know arabic, I can't say if you can use these findings in your work. Nevertheless, I would look at. Maybe some findings are useful for your research.

Thank you all for your answers.