Understanding human language can seem very simple, people do it every day, but it is a very difficult task for computers [1]. The underlying problem here is that the natural language is full of phenomena that make human comprehension a very difficult task. For this reason, many scientists around the world work in the so-called semantic similarity challenge. The idea of measuring similarity in the area of natural language processing is to be able to assess the interrelation between any two words in a given text.
In recent times, and thanks to the advances made through Deep Learning, new useful resources have been automatically generated to improve the systems. This has made it possible for semantic similarity to become currently one of the hottest research areas inside the Artificial Intelligence community. The idea of training computers to they can be able to recognize the degree of similarity between texts is attracting a lot of attention by both academia and industry.
In recent times, scientists and practitioners have proposed a myriad of new methods and tools to address this challenge. Some of the most outstanding methods are wordnet, word2vec, google normalized distance, semantic similarity controllers based on fuzzy logic, and BERT. Most of these solutions are oriented to achieve high levels of accuracy. However, they pay none or little attention to the interpretability of the solution. This is a pending problem to solve. And this is exactly what this website is intended for.
References:
[1] Jorge Martínez Gil. Semantic similarity aggregators for very short textual expressions: a case study on landmarks and points of interest. J. Intell. Inf. Syst. 53(2): 361-380 (2019)