Sentiment Analysis in Complex Adaptive Systems

The aim of this work is to present a new algorithm for the evaluation of sentiment in Czech language texts. The algorithm is based on a new dictionary and uses n-gram searching. For the creation of the dictionary, it was important to use language specific phrases and exceptions, which can completely change the final evaluation of a sentiment. The solution also includes automatic search for a new subjects (aspects) of evaluation and also searching for new words determining sentiment. A similar algorithm can also be applied to other languages. The work emphasizes the transformation of the acquired data into valuable information. Our experiment is realized in the experimental adaptive web system in e-learning content domain and in eShop domain. The success and benefits of the algorithm are also discussed in this text.