Article: A sustainable community user experience generation framework integrating the generative adversarial network and visual attention model Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.68 pp.70 - 90 Abstract: With societal development, traditional community planning relies on designers' experience and static norms, failing to dynamically address residents' diverse subjective needs. Existing generative adversarial network (GAN) models aid design but suffer from pixel blurring, structural defects and insufficient UX semantic guidance. This study integrates GIS environmental features, embedded coding social features and scale-analysed UX features; via a gated cross-attention mechanism, it enables the generator to focus on key spatial elements. Multi-discriminator collaboration and cycle consistency loss (CCL) ensure generation quality at pixel, structural and semantic levels. SCD-20k experiments show the peak signal-to-noise ratio of Sustainable Community user experience generation (SCUE-Gen) is 34.6 dB. Its experience vector similarity (EVS) is 0.89, outperforming benchmarks like cGAN and Pix2Pix; professional and non-professional satisfaction both exceed 8.8. The framework fits urban planning workflows, offering iterative schemes balancing professional norms and residents' needs, data-driven support for sustainable communities, and interdisciplinary backing for humanistic smart cities. Inderscience Publishers - linking academia, business and industry through research

Title: A sustainable community user experience generation framework integrating the generative adversarial network and visual attention model

Authors: Wuyimei Jin

Addresses: Academy of Fine Arts, Taizhou University, Taizhou, 225300, Jiangsu, China

Abstract: With societal development, traditional community planning relies on designers' experience and static norms, failing to dynamically address residents' diverse subjective needs. Existing generative adversarial network (GAN) models aid design but suffer from pixel blurring, structural defects and insufficient UX semantic guidance. This study integrates GIS environmental features, embedded coding social features and scale-analysed UX features; via a gated cross-attention mechanism, it enables the generator to focus on key spatial elements. Multi-discriminator collaboration and cycle consistency loss (CCL) ensure generation quality at pixel, structural and semantic levels. SCD-20k experiments show the peak signal-to-noise ratio of Sustainable Community user experience generation (SCUE-Gen) is 34.6 dB. Its experience vector similarity (EVS) is 0.89, outperforming benchmarks like cGAN and Pix2Pix; professional and non-professional satisfaction both exceed 8.8. The framework fits urban planning workflows, offering iterative schemes balancing professional norms and residents' needs, data-driven support for sustainable communities, and interdisciplinary backing for humanistic smart cities.

Keywords: fused generative adversarial network; attention model; community user experience; visual attention mechanism; multimodal conditions.

DOI: 10.1504/IJICT.2026.154351

International Journal of Information and Communication Technology, 2026 Vol.27 No.68, pp.70 - 90

Received: 25 Dec 2025
Accepted: 29 Jan 2026
Published online: 23 Jun 2026 *

Title: A sustainable community user experience generation framework integrating the generative adversarial network and visual attention model

Keep up-to-date