classifying, using a logistic regression classification model executing on a processor, a topic into an interaction type in a set of predefined interaction types;

extracting, from a document repository, a set of documents corresponding to the topic;

generating, using a generator sub-model of a generative adversarial model executing on a processor, from input data indicative of a reaction to a previous presentation, a scored sentiment corresponding to the reaction to the previous presentation;

weighting, using a trained attention layer model, the interaction type, the set of documents, and the scored sentiment, the weighting generating a weighted interaction type, a weighted set of documents, and a weighted scored sentiment; and

adjusting, using a natural language generation transformer model executing on the processor according to the weighted interaction type and the weighted scored sentiment, a portion of a document in the weighted set of documents, the adjusting comprising replacing, within the portion, a first sequence of words with a second sequence of words.