A Travel Path Recommendation Algorithm Based on Hybrid Particle Swarm and Ant Colony for Social Media Shared Data Mining

Abstract

The effective mining of shared data on social media can help personalized recommendations of tourist attractions and paths for users. This study proposes a tourism path recommendation scheme that combines PSO and ant colony optimization to address the issue of low recommendation accuracy caused by incomplete extraction of effective information in tourism path recommendation algorithms. The tourism path recommendation algorithm obtains a pseudo demand sequence based on the distance between the user's center point, and obtains attribute keywords through the user's evaluation text and text extraction technology. It utilizes the iterative operation of particle swarm ant colony algorithm to obtain the semantic distance and geographic distance of the target user to the current optimal sequence, and updates the preference distance through weighted calculation. For the four benchmark functions, the proposed algorithm had a longer running time under the same number of runs. Under the four benchmark test functions f1, f2, f3, and f4, when the maximum number of runs was reached, the running time of the algorithm was 36.58s, 62.96s, 90.59s, and 64.26s, respectively. The proposed PSO ant colony travel path recommendation algorithm had lower recommendation errors under different running times, and the range of error values for route recommendation was 0.005-0.089. In the training set, the confusion matrix results of the algorithm showed that the accuracy of tourism path recommendation for topics 1 and 5 was 81.25% and 84.26%, respectively, and the recommendation accuracy for the other three topics was also above 75%. The designed algorithm takes into account both emotional and time series dimensions, and has high recommendation accuracy. It has obvious advantages in the actual process of recommending tourist routes.