Universal language model with the intervention of quantum theory

We assume that natural language has a property analogous to quantum mechanics, and the rationale for adopting this analogy comes not only from noticing that the meanings of natural language can be analogous described by in the form of superposition of states, but also from the fact that we notice that the representation of natural language symbols has a duality in which the symbols-meanings correspond to each other. At the same time, natural language processing (NLP) methods based on statistical implementations have been popular for decades and continue to evolve.It makes us wonder whether we can go further than statistical theory and consider introducing the relevant theories of quantum mechanics into the modeling of natural language. Meanwhile, in the last decade or so, NLP has generally adopted the approach of converting natural language symbols into some kind of mathematical representation for processing, and this approach, known as word embedding, has achieved surprisingly good performance in universal language processing tasks.We see that the technical idea of completely converting the symbols that make up natural language into a numerical representation for processing to build a universal language model(ULM) is highly feasible and possible. The experimental progress fed back from these two real-world applications motivates us to explore building natural language models based on quantum theory. Within the past century, quantum theory has become more and more perfect as mankind has studied the physical world in depth.