METHOD AND SYSTEM FOR PROCESSING USER SPOKEN SPEECH FRAGMENT Russian patent published in 2021 - IPC G10L15/22 G06N20/00 

FIELD: computer technology.

SUBSTANCE: invention relates to the field of computer technology for natural language processing. A method includes receiving from a user an indicator of a user spoken speech fragment; forming a hypothesis based on a text representation based on the user spoken speech fragment; processing, using the first scenario-based trained model and the second scenario-based trained model, the hypothesis based on the text representation in order to form the first scenario-based hypothesis and the second scenario-based hypothesis, respectively; wherein the first scenario-based trained model and the second scenario-based trained model are trained using at least partially different corpus of texts; analyzing, using a machine learning algorithm (MLA), the first scenario-based hypothesis and the second scenario-based hypothesis in order to determine a winning scenario with a higher quantitative confidence indicator; based on the winning scenario, determining, by means of an associated one of the first scenario-based trained model and the second scenario-based trained model, an action that must be performed by an electronic device; performing the action.

EFFECT: increase in the accuracy of determining the action that must be performed by the electronic device in accordance with the user spoken speech fragment.

28 cl, 6 dwg