System

A system comprising:

• circuitry configured to:
• receive, via a communication interface coupled to a packet-switched network, audio information and image information from a terminal device, apply speech recognition processing using a neural network-based automatic speech recognition model with beam search decoding to produce first character information, apply a convolutional neural network model to the image information to produce item identification data with per-label confidence scores above a confidence threshold, and integrate the first character information, the item identification data, and directly input character information with source-type annotations;
• apply named entity recognition and numeric extraction to the integrated text information to extract item data, time constraint data, preference data, and quantity data and normalize the extracted data to produce standardized condition information;
• select a sentence template from a plurality of templates stored in a storage device based on the quantity data and the time constraint data, generate a prompt sentence by substituting standardized field values into template placeholders, and input the prompt sentence to a generative neural network model to produce candidate output information comprising an output name, material item records, and step information records;
• apply pattern matching to parse the candidate output information into a structured data object and generate summarized procedure information by applying token-length-constrained summarization to the step information records;
• search for a visual information reference link using a query constructed from the output name and the item identification data;
• and
• transmit the candidate output information, the summarized procedure information, and the visual information reference link to the terminal device and transmit a request to an external data service comprising the output name and a location parameter to obtain and transmit ranked provider records to the terminal device as order support information.

A method comprising:

• receiving, via a communication interface coupled to a packet-switched network, audio information and image information from a terminal device, applying speech recognition processing to the audio information to produce first character information, applying image recognition processing to the image information to produce item identification data, and integrating the first character information, the item identification data, and directly input character information to produce integrated text information;
• executing natural language processing on the integrated text information to extract and normalize structured condition information;
• generating a prompt sentence by embedding the standardized condition information into a predetermined sentence template and inputting the prompt sentence to a generative neural network model to obtain candidate output information;
• converting the candidate output information into a structured data object and generating summarized procedure information;
• and
• transmitting the candidate output information, the summarized procedure information, and a visual information reference link to the terminal device and transmitting a request to an external data service to obtain provider information and provide order support information to the terminal device.