Learning GUI Completions with User-defined Constraints

A key objective in the design of graphical user interfaces (GUIs) is to ensure consistency across screens same product. However, designing a compliant layout time-consuming and can distract designers from creative thinking. This paper studies recommendation methods that fulfill such requirements using machine learning. Given desired element type size, suggest placements following real-world GUI processes. Consistency are given implicitly through previous layouts which patterns be learned, comparable existing software We adopt two recently proposed for this task, Graph Neural Network (GNN) Transformer model, compare them with custom approach based on sequence alignment nearest neighbor search (kNN) . The were tested handcrafted datasets explicit patterns, as well large-scale public diverse mobile layouts. Our results show our instance-based learning algorithm outperforms both neural network approaches. Ultimately, work contributes establishing smarter tools professional explainable algorithms increase their efficacy.