engleski

Formalizing Inferential Evidentiality: From Justification Logic to Machine Learning

Evidentiality is a grammatical category in which the speaker is obligated to state the evidence for his statement, otherwise the statement is ungrammatical. One can use justification logic – that unfolds modalities into justification terms – to formalize inferential rules in such languages. It will be shown how justification logic can be used to formalize indirect evidentiality, that lie on speaker’s background knowledge or inference, unlike direct evidentials that depend on sensory perception. Hence, it will be shown how justification logic axiomatization of such examples can be used in theoretical computer science, where such abstract notions can be processed in machine learning of rules and concepts in natural language.

justification logic, logic of proofs, evidentiality, machine learning

nije evidentirano