The development of Web 2.0 has led to the generation of huge amount of available and public information on online platforms. Users are now free to express their sentiments about their experiences towards products, sites, or events. These thoughts and views can be read by millions of people, which can influenced directly their decisions. Therefore, it is very important to analyze online opinions because they can affect the viability of either public or private organizations.
The development of new techniques capable of processing a high volume of text for understanding people’s opinion has become an important task in recent years. Thus, Sentiment Analysis which is the field that analyzes opinions and sentiments in written text has aroused the interest of many research groups. Since early 2000s, many studies has focused on developing algorithms that mine opinions from restaurants, products, movies, etc. However, we find that there is a lack on analyzing reviews for cultural heritage. These organizations need a digitalization process in order to enhance several aspects of their management like visitor’s experiences.
In this thesis, we propose to analyze cultural monuments reviews for extracting valuable insights. We aim at addressing several gaps of sentiment analysis based-methods like sentiment methods inconsistencies, neutrality detection, and opinion summarization by ensembling different machine learning and natural language processing techniques. The results show that our proposed methods are consistent for a domain cultural, which in turn will enhance the decision-making process of these institutions.