An SUI-based approach to explore visual search results cluster-graphs.

Area of Science:

• Computer Science
• Human-Computer Interaction
• Information Retrieval

Background:

• The increasing volume of online visual content complicates information retrieval.
• Existing image search engines often use linear layouts, hindering exploration of related image objects.
• Users struggle to access and explore images effectively through traditional search interfaces.

Purpose of the Study:

• To propose a new Search User Interface (SUI) for enhanced image exploration.
• To enable non-linear reachability and interactive visualization of image search results.
• To address the limitations of conventional image search result presentation.

Main Methods:

• Developed a Search User Interface (SUI) approach.
• Implemented a cluster-graph data model representing images as nodes and multimodal similarities as edges.
• Evaluated the approach using real image datasets, usability tests, and behavioral analysis.

Main Results:

• The proposed SUI facilitates interactive exploration and visualization of image results.
• The cluster-graph model allows navigation through multimodal similarity relationships.
• Usability testing indicated high user satisfaction (76.83%) and usability (83.73%).

Conclusions:

• The novel SUI approach significantly improves the exploration of image search results.
• Interactive visualization and non-linear navigation enhance user experience in image retrieval.
• The findings suggest a promising direction for future image search engine design.