System

A system comprising:

• circuitry configured to
• receive, via a packet-switched network, a resource identifier and an input character sequence from a terminal device;
• acquire document data from a data source based on the resource identifier and parse the document data to extract a plurality of text data segments into a structured data representation;
• execute a natural language processing pipeline on the plurality of text data segments to compute, for each text data segment, an importance score, an affective state classification, and a topic cluster identifier;
• compute an affective state indicator for the input character sequence and calculate an evaluation value for each text data segment based on a combination of the importance score and an alignment measure between the affective state indicator and the affective state classification;
• generate edited data by selecting text data segments based on the evaluation value, reconstructing a sequence order, grouping the selected text data segments into topic-based sections, and associating emphasis metadata with designated portions;
• construct a first structured input sequence for a generative neural network model to obtain structured analysis result data, and construct a second structured input sequence based on the structured analysis result data and the edited data to obtain inference output data comprising article text;
• and
• transmit, via the packet-switched network, display data derived from the inference output data and the edited data to the terminal device for rendering on a display of the terminal device.

A method performed by a system comprising circuitry, the method comprising:

• receiving, via a packet-switched network, a resource identifier from a terminal device;
• acquiring, based on the resource identifier, document data having a hierarchical markup structure from a data source accessible via the packet-switched network;
• parsing the hierarchical markup structure of the document data to extract a plurality of text data segments and associated metadata into a structured data representation;
• executing a natural language processing pipeline on the plurality of text data segments to compute, for each text data segment, a feature vector comprising at least an importance score, an affective state classification, and a topic cluster identifier;
• generating edited data by selecting a subset of the plurality of text data segments based on the importance score, reconstructing a sequence order of the subset, and associating emphasis metadata with designated portions of the subset;
• constructing a structured input sequence for a generative neural network model based on the edited data and the feature vector, the structured input sequence comprising natural language instruction tokens specifying an output structure and content constraints;
• inputting the structured input sequence to the generative neural network model and obtaining inference output data comprising generated text organized according to the output structure;
• and
• transmitting, via the packet-switched network, display data derived from the inference output data and the edited data to the terminal device for rendering on a display of the terminal device.