System

A system comprising:

• circuitry configured to:
• acquire, via a communication interface coupled to a packet-switched network, article text from a network-accessible information source, construct a multi-task prompt sentence embedding the article text as a content section, and input the multi-task prompt sentence to a transformer-based generative neural network model to generate summary information and theme information as separate output segments;
• retrieve the theme information and generate, using the transformer-based generative neural network model, a first prompt sequence comprising an open-ended question conditioned on the theme information and one or more subsequent prompt sequences constructed from a conversation context data structure storing prior character information entries and system prompt sequences with role annotations;
• receive audio information from a terminal device via the communication interface, extract acoustic feature vectors from the audio information, and convert the audio information into character information by applying a neural network-based speech recognition model using beam search decoding;
• construct an evaluation prompt sentence encoding the theme information, the character information, and multi-dimensional analysis instructions and input the evaluation prompt sentence to the transformer-based generative neural network model to evaluate a language quality metric, a textual correctness metric, and a logical consistency metric;
• parse model output into favorable aspect data and improvement aspect data, cross-validate with rule-based natural language processing checks, and transmit resulting feedback information to the terminal device;
• and
• store character information and evaluation results as session history records comprising per-session metrics and adjust at least one subsequent prompt sequence by selecting a prompt template parameterized by proficiency information derived from the session history records.

A method comprising:

• acquiring, via a communication interface coupled to a packet-switched network, text information from an information source and generating, using a generative neural network model, summary information and theme information from the text information;
• generating, using the generative neural network model and based on the theme information, a first prompt sequence for initiating an interactive session and one or more subsequent prompt sequences for managing progression of the interactive session;
• receiving, via the communication interface, audio information from a terminal device and converting the audio information into character information by applying speech recognition processing to the audio information;
• analyzing the character information together with the theme information using natural language processing to evaluate a language quality metric, a textual correctness metric, and a logical consistency metric of the character information;
• generating feedback information comprising favorable aspect data and improvement aspect data based on the evaluation and transmitting the feedback information to the terminal device via the communication interface;
• and
• storing the character information and the evaluation as session history records and adjusting at least one subsequent prompt sequence based on the session history records.