System

A system comprising:

• circuitry configured to
• receive, via a communication interface coupled to a packet-switched network, text data from a terminal device,
• apply a natural language processing algorithm to the text data to perform tokenization, part-of-speech tagging, and named entity recognition, and to extract classification information comprising intent information and topic information,
• compute a vector representation of the text data using a transformer-based encoder model, and perform a similarity search against a plurality of stored vector representations in a storage device to identify similar past interaction records,
• construct a prompt sentence for a generative neural network model, the prompt sentence comprising the text data, the classification information, and response data from the identified similar past interaction records as input context,
• input the prompt sentence to the generative neural network model to generate response data,
• execute sentiment analysis processing on the text data using a sentiment classification model to generate emotion information, and adjust at least one of content and style of the response data based on the emotion information,
• apply a predefined rule set to the classification information to generate routing information, and when the routing information indicates that the text data requires handling by an entity having specialized knowledge, search an entity information store to identify candidate entities and rank the candidate entities by a relevance score, and
• transmit the response data, the routing information, and information identifying ranked candidate entities to the terminal device for display via a graphical user interface.

A method comprising:

• receiving, via a communication interface coupled to a packet-switched network, text data from a terminal device,
• applying a natural language processing algorithm to the text data to extract classification information and feature information,
• computing, based on the feature information, a similarity score between a vector representation of the text data and a plurality of vector representations stored in a storage device, and when the similarity score satisfies a predetermined threshold, retrieving associated response data from the storage device,
• constructing a prompt sentence for a generative neural network model based on the classification information and the text data, and inputting the prompt sentence to the generative neural network model to generate response data when the similarity score does not satisfy the predetermined threshold,
• applying a predefined rule set to the classification information to generate routing information identifying a processing path for the text data, and
• transmitting the response data and the routing information to the terminal device for display via a graphical user interface.