Machine Learning for Enhanced Advertisement Effectiveness: A Computational Model to Analyze Persuasion in Visual Advertisements
Gespeichert in:
| Titel: | Machine Learning for Enhanced Advertisement Effectiveness: A Computational Model to Analyze Persuasion in Visual Advertisements |
|---|---|
| Autoren: | Vemulapalli, Vijay |
| Verlagsinformationen: | Zenodo |
| Publikationsjahr: | 2024 |
| Bestand: | Zenodo |
| Schlagwörter: | Deep learning, Persuasion strategies, Visual advertisements, U-Net model, ConvNextXLarge, Computational advertising, Advertisement effectiveness, Return on Advertising Spend, Convolutional Neural Network |
| Beschreibung: | The competitive landscape of advertising necessitates a nuanced understanding of persuasion strategies to craft impactful advertisements. In this pursuit, we present a novel framework employing deep learning models to analyze picture advertisements and both predict and locate the embedded persuasion strategies. Our approach leverages the U-Net model and the ConvNextXLarge Convolutional Neural Network (CNN) model to extract and analyze visual features indicative of various persuasion strategies. Through rigorous evaluation, we demonstrate that our models exhibit a promising capability to identify and categorize persuasion strategies effectively. The output of this research not only provides a computational method to gauge the persuasion tactics deployed but also paves the way for an assistive tool for companies aiming to enhance their advertisement effectiveness. By integrating our persuasion prediction model, companies can now analyze the potential effectiveness of their advertisements prior to campaign launches. This, in turn, facilitates data-driven decision-making in advertisement design, potentially leading to improved engagement rates and better Return on Advertising Spend (ROAS). Our work presents a significant stride towards marrying machine learning with advertising domain knowledge, opening avenues for further research and real-world applications in the rapidly expanding field of computational advertising. |
| Publikationsart: | article in journal/newspaper |
| Sprache: | English |
| Relation: | https://zenodo.org/records/10792336; oai:zenodo.org:10792336; https://doi.org/10.5281/zenodo.10792336 |
| DOI: | 10.5281/zenodo.10792336 |
| Verfügbarkeit: | https://doi.org/10.5281/zenodo.10792336 https://zenodo.org/records/10792336 |
| Rights: | Creative Commons Attribution 4.0 International ; cc-by-4.0 ; https://creativecommons.org/licenses/by/4.0/legalcode |
| Dokumentencode: | edsbas.2FC85DB9 |
| Datenbank: | BASE |
Nájsť tento článok vo Web of Science | Abstract: | The competitive landscape of advertising necessitates a nuanced understanding of persuasion strategies to craft impactful advertisements. In this pursuit, we present a novel framework employing deep learning models to analyze picture advertisements and both predict and locate the embedded persuasion strategies. Our approach leverages the U-Net model and the ConvNextXLarge Convolutional Neural Network (CNN) model to extract and analyze visual features indicative of various persuasion strategies. Through rigorous evaluation, we demonstrate that our models exhibit a promising capability to identify and categorize persuasion strategies effectively. The output of this research not only provides a computational method to gauge the persuasion tactics deployed but also paves the way for an assistive tool for companies aiming to enhance their advertisement effectiveness. By integrating our persuasion prediction model, companies can now analyze the potential effectiveness of their advertisements prior to campaign launches. This, in turn, facilitates data-driven decision-making in advertisement design, potentially leading to improved engagement rates and better Return on Advertising Spend (ROAS). Our work presents a significant stride towards marrying machine learning with advertising domain knowledge, opening avenues for further research and real-world applications in the rapidly expanding field of computational advertising. |
|---|---|
| DOI: | 10.5281/zenodo.10792336 |