An apparatus for controlling a response utterance being reproduced and predicting a user intention, which is implemented in a computing device including at least one processor and at least one memory for storing instructions executable by the processor, the apparatus comprising:
- a voice signal analysis processor for, when a second voice signal is received from a user while a first response utterance for responding to a first voice signal received from the user is output through a chatbot of a first domain among a plurality of domains, starting analysis on the second voice signal, wherein the voice signal analysis processor is configured to perform the following steps:
- a first sentence identification step, when the second voice signal is received while the first respons e utterance is output through the chatbot, of preprocessing the second voice signal through a pre-stored voice recognition algorithm to identify a first sentence corresponding to the second voice signal;
- a morpheme-based word segment checking step, when the identification of the first sentence completed, of identifying a plurality of morphemes included in the first sentence through a pre-stored natural language processing algorithm to check a plurality of word segments constituting the first sentence based on the identified morphemes;
- a sentence component identification step, when the checking of the word segments is completed by execution of a function of the morpheme-based word segment checking step, of checking a part of De ach of each of he morphemes included in the word segments based on pre-stored part-of-speech classification information to identify a sentence component of each of the word segments;
- and
- a first clause determination step, when the identification of the sentence component for each of the word segments is completed by execution of a function of the sentence component identification step, of checking a combination relation between the identified sentence components determine a clause type for the first sentence;
- an utterance control processor for, when one of preset keywords is identified to be included in a first sentence corresponding to the second voice signal in a state where execution of a function of the voice signal analysis processor is completed, controlling the first response utterance being output through the chatbot to correspond to the identified keyword;
- an intention prediction processor for, when a third voice signal is received from the user in a state where the first response utterance is controlled by execution of a function of the utterance control processor, analyzing the third voice signal through a pre-stored intention analysis algorithm to predict an intention of the user, which is reflected in a second sentence corresponding to the third voice signal, through a result of the analysis, wherein the intention prediction processor is configured to perform the following steps:
- a second sentence identification step, when the third voice signal is received from the user in a state where the first response utterance output through the chatbot is controlled by the execution of the function of the utterance control step, of preprocessing the third voice signal through the pre-stored voice recognition algorithm to identify the second sentence corresponding to the third voice signal;
- and
- a second clause determination step, when the identification of the second sentence is completed, of analyzing a plurality of sentences corresponding to voice signals received from the user before the third voice signal is received through the pre-stored natural language processing algorithm and the identified second sentence to determine a clause type of each of the sentences and the second sentence;
- and
- a response utterance output processor for, when execution of a function of the intention prediction processor is completed, generating a response sentence for responding to the third voice signal in which the predicted intention of the user is reflected to output a response utterance based on the generated response sentence through the chatbot,
- wherein the intention prediction processor is further configured to perform the following steps:
- a keyword named-entity analysis step, when the clause type for each of the plurality of sentences and the second sentence is determined, of identifying full morphemes from both each sentence and the second sentence as keywords using the pre-stored intention analysis algorithm to start a named-entity recognition process for the identified keywords based on a pre-stored morpheme semantic dictionary;
- a semantic information tagg step, as a function of the keyword named-entity analysis step is performed, of chec kina semantic information corresponding to a meaning of each of the identified keywords based on the pre-stored morpheme semantic dictionary, and tagging the checked semantic information to each of the identified keywords;
- and
- an intention checking step, when the semantic information tagging step is completed, of checking the Intention of the user based on the semant information tagged to each of the keywords and the clause type of each of the sentences and the second sentence such that a past intention of the user based on the sentences and a plurality of intentions, which are a candidate group based on a meaning derived from the second sentence, are checked to complete an intention prediction analysis process for the sentences and the second sentence.