4. Conclusion

In this article, we have presented the different approaches to text-based opinion analysis. These use automatic language processing, machine learning and data mining techniques to extract opinions from documents. Finally, automatic summarization methods are sometimes used to present the extracted opinions in a comprehensible form that facilitates decision-making.

Automatic opinion processing uses a variety of techniques, the most popular being those based on a representation of the text as a 'bag of words', or those based on predefined lists of subjective words. The latter can be combined with syntactic analysis to take into account linguistic phenomena at sentence level, such as negations, intensifiers or syntactic dependencies between opinion words, subject and/or aspects.

However, the linguistic complexity involved in the expression of...