Seminar about sentiment analysis and opinion mining.

Recently people's opinions have become an important means of making decisions, not only for individuals but also for government and commercial sectors among others. These opinions have become available and are widely spread due to the evolution of the mechanisms of the Internet and also due to the wide spread of different applications that considered a fertile environment for communication and exchange of information and opinions such as social media Networks (Twitter, Facebook, LinkedIn, etc), reviews (movie reviews, product reviews, etc), blogs, online commercial shops (Amazon, eBay, Google Shopping, etc) and more. This enormous volume of on-line information needs to be structured and organized in a useful way for users to get oriented opinion summary from text related to their search.

Topics

Topic 1 The idea behind this topic is to introduce the different available applications that make use of opinion mining and sentiment analysis. Different techniques used for analyzing reviews are needed to be introduced. Determine the orientation or polarity of a given document or a review into generic positive or negative classes is called a binary classification. Multi class classification is also used to show more flexible classifications such as rating between 1 to 5. Applications of sentiment analysis

1. Product reviews
2. Movie reviews
3. Classification (positive and negative orientation or rating schema)

Topic 2-5This topic deals with the most popular supervised (machine learning) techniques that are used for text classification. Unsupervised techniques on the other side are also used to obtain the polarity of a given document Techniques

a. Supervised
I. Naïve Bayes
II. Support Vector Machines (SVM)
III. Decision Trees

b. Unsupervised
I. Pointwise Mutual Information (PMI)
II. Latent Semantic Analysis (LSA)
III. Bootstrapping
IV. Statistical approaches

Topic 6-8. Sentiment classification takes place using three main levels of classifications; each has its own assumptions and techniques. Students need to show the level of performance for each level. I. Document level classification

II. Sentence level classification
III. Aspect / Feature level classification

Topic 9. In this topic students need to show how semantic techniques are applied to improve the result quality of the opinion mining and sentiment analysis.

I. Semantic Sentiment Analysis

General References

 - Ronen Feldman (2013). Techniques and applications for sentiment analysis. Communications of the ACM , Volume 56 Issue 4.

- Bing Liu (2012). Sentiment Analysis and Opinion Mining. Morgan & Claypool Publishers. - Leung, C., & Chan, S. C. (2009). Sentiment Analysis of Product Reviews.