Enhancing Tourism Industry Attraction Recommendations and Travel Route Personalisation Through UGC Text Mining

Advanced natural language processing (NLP) techniques were applied to 527,843 travel reviews in order to develop a user-generated content-based travel recommendation system. This was compared to traditional methods in a user study of 200 participants, split between the two systems, each evaluating one of them on Recommendation Relevance, Route Efficiency, Attraction Diversity, and Overall Satisfaction based on a 5-point Likert scale. The results showed that on all axes, the performance of the user-generated content (UGC) based system was always better than that of the traditional method. It also performed better regarding Recommendation Relevance, with 4.32 over 3.68, and therefore, users found recommendations more relevant to their interests. Route Efficiency was also superior, at 4.18 compared to 3.52, hence, this system provides more practical and time-saving travel routes. The results make it very evident that the exploitation of UGC is very likely to increase personalisation, diversity, and overall satisfaction with travelling experiences. This is useful information that may be of importance to tourism business executives and scholars, so as to point out how UGC text mining can contribute to the refinement of attractions recommended and route optimisation for trips.