Personalised Running Event Recommendation:  the  system provides a list of more specialised and niece running events to sport tourists. The recommendation is based on the personal information, running event history together with factors of interest for choosing a running event input by users and recorded in User Profile Ontology.  To recommend the suitable event to the user,  Jena API's inference rule applies the recommendation rules to user's instance generated from User Profile Ontology. This results in the matching factor information in the recommended running events from Running Event Ontology to be retrieved in order to recommend the corresponding events to the user.  

Effective and Efficient Ontology-Driven Association Rule Mining Algorithm:  the system was implemented from an ontology-driven association rule mining framework which integrates the ontology-based methodology and assoication rule mining technique. This results in a set of sutiable running events that align with the context and interests of sport tourists  to be recommended from the knowledge-based ontogy through the recommendation based association rule.

Robust and reliableRecommendation Rules: To mitigate small-sample bias, a non-parametric permutation test was conducted with 10,000 iterations to validate our 83 discovered rules. By shuffling features across the 133 entries, we established a null distribution of Lift values and derived corresponding p-values for each rule. Following a Bonferroni correction (set at α = 0.0006 (0.05/83), 70 rules (84.3%) were identified as statistically significant, while any rules failing to meet the p < 0.05 threshold were excluded as noise. Refining the initial 83 rules down to 70 validated ones significantly enhanced the model's performance. By pruning noise, precision increased from 72% to 83%, while recall remained steady at 95%, resulting in an F-measure rise from 78% to 86%. These results confirm that selective rule reduction strengthens predictive reliability without sacrificing coverage.