Application of Artificial Intelligence-based Content Generation Technology in News Publishing
Pubblicato online: 21 mar 2025
Ricevuto: 10 ott 2024
Accettato: 05 feb 2025
DOI: https://doi.org/10.2478/amns-2025-0642
Parole chiave
© 2025 Hongqiao Li, published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Artificial Intelligence (AI) content generation technology has had a profound impact on many aspects of news publishing. In this paper, a personalized news headline generation model that incorporates user characteristics achieves innovation in news headline generation. Firstly, an additive attention-based variant of Transformer, i.e., Fastformer encoder, is constructed through Transformer, into which preprocessed news text is fed for contextual modeling. A pointer generation network model decoder is used to copy words from the source text to improve the accuracy of the copied information. And the headlines are personalized by injecting user features to make the generated news headlines more in line with the preferences of the news publisher audience. In this way, a personalized news headline generation model that incorporates user features is constructed. After this, comparative experiments were conducted. On the LCSTS dataset, compared with the LSTM+Point model, the model in this paper improves 16.43%, 6.31% and 0.75% on the three metrics of Rouge-1, Rouge-2 and Rouge-L, respectively. When using different decoding strategies, the Rouge-1, Rouge-2, and Rouge-L metrics of the headline generation model based on this paper’s decoding strategy reached 31.28, 12.68, and 28.31.It illustrates the scientific validity of the structure of the model designed in this paper, with headlines that are more readable, richer in semantic information, and likely to appeal to a larger number of readers.