System

A system comprising:

• circuitry configured to:
• receive, from a terminal device via a packet-switched network, document data and communication data, and extract character information from the document data and communication data by performing normalization and tokenization;
• calculate an importance index for extracted terms by applying a statistical analysis method including term-frequency inverse-document-frequency computation, and extract terms exceeding an importance threshold;
• analyze semantic relationships of the extracted terms by generating semantic embedding vectors using a neural network model, computing similarity between the vectors, grouping semantically similar terms, and selecting standard terms for each group to construct a common language dataset;
• store the common language dataset in a storage device;
• generate a prompt sentence based on the common language dataset and context information derived from the document data, the prompt sentence conforming to a template that specifies a task for a generative neural network model comprising a transformer architecture with self-attention layers;
• transmit the prompt sentence to the generative neural network model, receive response data, and evaluate terminology consistency of the response data against the common language dataset;
• and
• transmit the response data and terminology evaluation results to the terminal device via the packet-switched network.

A method comprising:

• receiving, by circuitry from a terminal device via a packet-switched network, document data and communication data;
• extracting character information from the received document data and communication data, and performing normalization processing and tokenization processing on the character information by using a natural language processing technique;
• applying a statistical analysis method to the tokenized character information to calculate an importance index, and extracting terms or term sequences determined to exceed an importance threshold;
• analyzing semantic relationships of the extracted terms or term sequences by computing similarity measures based on semantic embeddings, integrating synonymous or similar terms to determine standard terms, and constructing a common language dataset including the standard terms, definitions of the standard terms, and related terms;
• storing the common language dataset in a storage device;
• generating, based on the common language dataset and context information derived from the document data or the communication data, a prompt sentence in accordance with a template so as to cause a generative neural network model to execute predetermined processing;
• and
• transmitting the prompt sentence to the generative neural network model via the packet-switched network, receiving response data from the generative neural network model, evaluating terminology consistency of the response data based on the common language dataset, and transmitting the response data to the terminal device.