System

A system comprising:

• circuitry configured to:
• receive, via a communication interface coupled to a packet-switched network, image information of an information recording medium from a terminal device, and perform image preprocessing including noise removal by denoising filter, resolution resampling, density normalization, adaptive binarization, and perspective correction;
• apply an optical character recognition model to the preprocessed image information to extract character information, and perform verification processing including format validation, numerical range checking, and misrecognition correction using a substitution table to generate verified character information;
• generate structured information records from the verified character information comprising date, amount, counterpart identifier, content description, and category code fields;
• generate a prompt sentence assembling the structured information records, classification rule information, and an output schema instruction, operate a generative neural network model using the prompt sentence with constrained output decoding to generate classification data comprising account classification, debit classification, credit classification, and amount, and store the classification data with associated correction history information in a storage device;
• convert the stored classification data into declaration data in a taxation procedure format using transaction aggregation rules and referential consistency checks;
• and
• receive a natural language modification instruction from the terminal device, apply natural language processing to extract a modification intent and target record identifier, and operate the generative neural network model with a modification prompt sentence to generate modified classification data validated against schema constraints.

A method comprising:

• receiving, via a communication interface coupled to a packet-switched network, image information of an information recording medium from a terminal device, and performing image preprocessing on the image information using image processing operations including noise removal, resolution adjustment, density correction, binarization, and geometric correction;
• extracting character information from the preprocessed image information using a character recognition technique, and performing verification processing on the character information including format determination, numerical value determination, date determination, and misrecognition correction;
• generating structured information records from the verified character information, wherein each structured information record comprises fields for at least a date value, an amount value, a counterpart identifier, a content description, and a category code;
• generating a prompt sentence for instructing a generative neural network model to generate classification data based on the structured information records and classification rule information stored in a storage device, operating the generative neural network model using the prompt sentence and the structured information records to acquire classification data and associated correction history information, and storing the classification data in the storage device;
• and
• converting the stored classification data into declaration data in a taxation procedure format using a transaction control procedure that maintains referential consistency, and recording a correspondence among the classification data, the image information, and the prompt sentence as identification information in the storage device.