In recent years, it has become the standard behaviour of consumers to read and write evaluations of other consumers before buying a product or a service. For the companies (such as hotels, authors, producers, etc.), this is a great opportunity to learn more about what is important to their customers and what they do not like. However, this is only possible if they can quickly extract the information from the opinions expressed by the customers – sentiment analysis – which requires automatic data processing in the case of large data volumes. Sentiment analysis depends heavily on words: sentiment words, negations, amplifiers, and words for the product or its aspects. If the sentiment analysis is to achieve more than just classifying a sentence as positive or negative, and if it needs to identify the liked or hated attributes of a product and the scope of negation, it needs linguistic and ontological knowledge.