Semantic Similarity is one of the hottest topics and most interesting research challenges that researchers in the A.I. community have to deal with. The truth is that positive results in this context can have a great impact on a wide range of disciplines, both academic and business. This is because a multitude of computational methods require the calculation, even if basic, of the degree of similarity between pieces of textual information.
Today, most of the leading solutions use some kind of method based on the vectorization of words through a deep neural network. However, the results obtained, although good, are difficult to interpret. For that reason, in the near future we will see the proposal of solutions that try to improve the interpretability of the trained models by the human operators who use them.