Natural sciences observe facts and make models. Mathematics more or less arbitrarly generates axioms and investigates all the possible consquences of them. What is the source of semantic knowledge (e.g. one that is described in Saeed Semantics https://www.wiley.com/en-us/Semantics%2C+4th+Edition-p-9781118430163)? How researcher generates the semantic knowledge from the text? Do the researcher is required to use only his/her inner sense of language to classify the possibilities of language use, roles of the words and so on?
The question is somewhat obscure but its aim is to automate the semantics as a science. It is quite is to imagine experimental machine that makse observations and automatically generates and adjust models. It is quite easy to imagine system that generates and proofs/discards mathematical lemmas. In fact - such systems already exist for decades.
But what about semantic science - how we can automate it, how can we automate the discovery of semantic knowledge? There is great necessity for such automation to boost symbolic processing of natural language and symbolic computational lingustics?
I am aware of type logical grammars, categorial grammars, research in line with "Linguistics and Philosophy" https://link.springer.com/journal/10988 but work as expemlified by Saeed Semantis if far, far more richer than these efforts/
One can imagine the use of methods of classification. E.g. researcher extract all the possible uses of nouns from the texts and then researched can make automated clustering of the type of uses of those nouns and in such manner she can discover specified clusters of uses and name thoses clasters as thematic roles of the nouns?