Digital tourism platforms have increased the requirement for intelligent trip planning systems that can produce tailored and context-aware itineraries. Most current techniques depend on user profiles or prior interactions, yet typically neglect the necessity to combine visually attractive semantics with organized subject information, making them impractical and unpersonalized. To address these limitations, this paper introduces VisiTrip-GRAPH, an intelligent computer vision and knowledge graph-assisted novel travel planning framework with personalization capabilities. The proposed VisiTrip-GRAPH uses computer vision algorithms to learn high-level semantic representations of attraction images and user-shared visual content to infer latent travel interests in culture, nature, cuisine, and fun activities. The comprehensive travel knowledge graph represents destinations, points of interest, time, space, travel modes, costs, and relationships between different graph entities. VisiTrip-GRAPH, through computer vision-graph semantic fusion, facilitates adaptive travel itinerary recommendation with semantic relevance, diversity, and practical travel feasibility. The intelligent recommendation engine dynamically optimizes itineraries based on user-specific travel constraints, including time, budget, and travel style. The proposed framework enhances personalization accuracy by 15–22%, itinerary feasibility by 12–20%, and semantic relevance by 18–25%, all while minimizing travel time and offering a wide range of travel options. User satisfaction is consistently high, suggesting that the proposed itineraries better suit customer interests. The findings demonstrate that computer vision-knowledge graph semantic fusion creates comprehensive, diversified, and feasible trip itineraries. Multimodal intelligence systems might underpin the next generation of tailored trip planning software, according to the research.
Wang et al. (Wed,) studied this question.