System

A system comprising:

• circuitry configured to:
• obtain, via a communication interface coupled to a packet-switched network, education-related document information from external information processing apparatuses based on identification information of information-collection targets, and store the document information as unstructured electronic documents in a storage device;
• execute an analysis program on the unstructured electronic documents to remove display-related structure information, extract body text, and generate analysis-target text data by performing character-string normalization and morphological analysis or word segmentation;
• compute feature vectors for the analysis-target text data by computing term-frequency inverse-document-frequency values or dense embeddings using a pre-trained encoder, apply a trained classifier to assign education category labels including a learning-task-creation category, an evaluation-test-creation category, and a guardian-correspondence category and associated relevance scores, and register resulting classified information records including category information, summary information, and search index information in the storage device;
• generate full-text search indexes and vector search indexes for the classified information records in the storage device;
• receive a prompt sentence from a terminal device via the communication interface, compute query feature vectors for the prompt sentence using the same feature extraction pipeline, execute hybrid full-text search and vector similarity search on the storage device to retrieve top-ranked classified information records, and generate context information by extractive summarization of the top-ranked records under a token budget;
• construct model input data including the prompt sentence and the context information, and operate a generative neural network model using the model input data to generate a solution;
• and
• perform content checking and format adjustment on the solution, and transmit presentation data including the solution to the terminal device via the communication interface.

A method comprising:

• receiving, via a communication interface coupled to a packet-switched network, identification information indicating one or more information-collection targets;
• obtaining, via the communication interface, document information from an external information processing apparatus based on the identification information, and storing the document information as unstructured electronic documents in a storage device;
• analyzing the unstructured electronic documents using an analysis program,
• removing structure information unrelated to substantive content, extracting body text, and generating analysis-target text data by performing preprocessing including character-string normalization and morphological analysis;
• applying topic estimation and relevance calculation to the analysis-target text data using vocabulary information or a learning model, and generating classified information records assigned to one or more information categories;
• registering the classified information records, including category information, summary information, and search index information, in the storage device, and generating data structures for full-text search indexes or vector search indexes;
• receiving, via the communication interface, a prompt sentence from a terminal device, executing full-text search and semantic-similarity search on the storage device using the prompt sentence, and generating context information from extracted classified information records;
• and
• generating model input data including the prompt sentence and the context information, operating a generative neural network model using the model input data to generate a solution in response to an individual request, and transmitting the solution to the terminal device via the communication interface.