System

A system comprising:

• circuitry configured to:
• receive inquiry data comprising a natural-language question, terminal identification information, and language information from a terminal;
• apply morphological analysis, syntactic analysis, and named-entity extraction to the natural-language question to obtain topic-indicating expressions, product identification information, and category candidates;
• acquire related information from a network-accessible storage device based on the topic-indicating expressions and the product identification information;
• construct structured prompt sentences encoding the natural-language question, the topic-indicating expressions, the category candidates, and constraint conditions preserving the related information, and supply the structured prompt sentences to a generative artificial intelligence model to obtain a natural-language answer sentence;
• and
• apply an emotion recognition algorithm to the inquiry data to determine an emotional state and adjust at least one of a response style and a degree of detail of the natural-language answer sentence based on the emotional state.

A method performed by circuitry, the method comprising:

• receiving inquiry data comprising a natural-language question, terminal identification information, and language information from a terminal;
• applying morphological analysis, syntactic analysis, and named-entity extraction to the natural-language question to obtain topic-indicating expressions, product identification information, and category candidates;
• acquiring related information from a network-accessible storage device based on the topic-indicating expressions, the product identification information, and the category candidates;
• constructing structured prompt sentences that explicitly encode the natural-language question, the topic-indicating expressions, the category candidates, and constraint conditions preserving contents of the related information;
• supplying the structured prompt sentences to a generative artificial intelligence model to obtain intent information, category information, and a natural-language answer sentence constrained by the related information;
• applying an emotion recognition algorithm to the inquiry data to determine an emotional state and adjusting at least one of a response style, a degree of detail, and an answer organization of the natural-language answer sentence based on the emotional state;
• and
• transmitting the natural-language answer sentence to the terminal.