Preliminary Analysis and Design of a Customized Tourism Recommender System

Abstract

This work presents preliminary analysis and design of the tourism recommender system. The work considers Pokhara city of Nepal as the domain of study as it is the second-largest city and tourism capital of Nepal. The work is built on data from travel websites, published and unpublished reports, literature, and a pilot study conducted in the same city. Tourist decision factors and motivational factors along with seven demographic data are taken into consideration to build an open-end questionnaire for the study. A total sample of 250 respondents is taken as part of the pilot study. The study uses kNN as the classification algorithm to form clusters of data and calculate silhouette to confirm the consistency and validity of clusters. Scatter plots with different combinations of data, UML-based diagrams, and decision tree are used to come up with an initial recommender system design. The work is vital for the design and development of recommendation systems based on a model, mixed, and customized approach. As the requirements directly come from the user end, the system is more specific in terms of recommendations especially for tourists for a specific destination that can be further elaborated to build a generalized model. This work produces an initial system, a knowledge base, different design diagrams, and a fully functional algorithm for the recommender system that can be used by individuals, business entities, and tourism organizations as a tool and knowledge base.