• semantic vector space models
• sentiment analysis
• Distributional semantics

The aim of the thesis is to propose and discuss some distributional approaches, i.e. approaches based on the assumption that linguistic expressions with similar distributions have similar meanings, to deal with problems related to the computational treatment of sentiment and subjectivity.
In particular, we focus on two tasks: the polarity classification(positive or negative) of words taken in isolation; the polarity classification of more complex linguistic expressions, with a special attention to modifications due to the context.