In this chapter, a methodology is presented for identifying semantic relationships of synonymy, hyponymy, and hyperonymy. Subsequently, three semantic embedding models were generated, designed to capture the richness and complexity of the relationships between terms. The performance of these models is assessed by employing them in the classification of opinions regarding the quality of service in a hotel. This task is accomplished using a convolutional neural network. Furthermore, the results obtained from the classification task are integrated into topic discovery, culminating in a more comprehensive understanding of the quality of service provided by the hotel.