| CPC G06F 40/253 (2020.01) [G06F 40/284 (2020.01)] | 17 Claims |

|
1. A method performed by a computing device, the method comprising:
detecting an utterance of a user;
extracting, using a speech recognition engine, at least one feature vector associated with the detected utterance of the user;
converting the utterance of the user into an input sentence by converting, based the extracted at least one feature vector and based on trained reference patterns, the utterance of the user into the input sentence, wherein the trained reference patterns are trained based on at least one of a machine learning model or a deep learning model;
receiving a plurality of training sentences each comprising a plurality of tagged corpus;
generating, based on the plurality of training sentences, a plurality of computer-coded grammar rules, wherein the plurality of computer-coded grammar rules comprises at least one computer-coded operator;
dividing, by the computing device, the input sentence into a plurality of morphemes;
performing, based on a plurality of computer code symbols, a computerized compression process of at least two computer-coded grammar rules;
determining, from among the plurality of computer-coded grammar rules and based on the plurality of morphemes, a computer-coded grammar rule for the input sentence, wherein the plurality of computer-coded grammar rules represent structural properties of a sentence defined by rules for a syntactic analysis of the input sentence, wherein the determined computer-coded grammar rule is obtained by compressing the at least two computer-coded grammar rules together, and wherein each of the at least two computer-coded grammar rules comprises at least one common computer-coded grammar element between the at least two computer-coded grammar rules; and
outputting, based on the determined computer-coded grammar rule, a result indicating a meaning of the input sentence.
|