Research On The Method Of Personalized Label Combination Based On Tourism Ontology

With the explosive growth of tourist information and the complex diversity of specific demand, it becomes a problem that how to intelligently offer tourists personalization information. High-tech is enforced to play an important role in the tourism industry.Web 2.0 has changed the behaviors of people and promoted the innovation of method from the collaborative filtering recommendation to tag recommendation. However, most tag data are redundant, fuzzy and deregulated, resulting from simple annotation process. This dramatically degrades performance of tag recommendation.In this dissertation, we describe tourism service resources entirely with ontology-based technology. Then we strive to obtain the tag sets to be recommended with social network analysis technology. Personalized recommendation sets can be further obtained based on the method of deblurring. We also study problems caused by sparse data, cold-start conditions and solve these limitations based on lower-order singular value decomposition. In this way, a combinational algorithm is proposed for personalized tag recommendation system. Finally, we make an evaluation of the algorithm with true data, compare with other tag recommendation methods, and develop an online tourism services system integrated with our algorithm. The dissertation is organized as followed:First, we investigate studies on current situations of tourist information service and recommendation systems at home and abroad. On the basis of the comparative analysis, the demand of recommendation service in the field of tourism can be understood clearly.To describe tourism service resources entirely and normally, we next give an overview of relevant knowledge about domain ontology, and construct the domain ontology of tourism from the angle of categories and hierarchies.Finally, we propose algorithms based on tourism ontology with the social network analysis technology, and test algorithms with some popular measures. Our algorithm is integrated into a developed online tourism recommendation system.