Use this URL to cite or link to this record in EThOS:
Title: Personalization models for travel recommender systems
Author: Alatiyyah, Mohammed Hamad
ISNI:       0000 0004 8509 3727
Awarding Body: Durham University
Current Institution: Durham University
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Thesis embargoed until 19 Mar 2023
Access from Institution:
This thesis proposes three novel personalisation models to improve travelers experiences of using constraint-based Travel Recommender Systems (TRSs). Specifically, the three models are (i) personalisation of a recommended travel plan based on user-dependent constraints; (ii) maximisation of a function to represent users satisfaction levels; and (iii) maximisation of user satisfaction levels derived from a model of conflicts that users frequently experience when travelling as a group. The first model proposed, Item Constraints Data Model (ICDM) is designed to tackle the limitations of existing models, i.e. their inability to generate a recommended travel plan based on a variety of constraint types. Our proposed ICDM aims to overcome this by generating customised travel plans based on specific individual user-constraint considerations. Moreover, our ICDM is able to handle specific constraints defined over particular time periods (e.g., a traveler who wishes to visit outdoor attractions in the after- noons). Another benefit of this approach is that it permits the use of general-purpose optimisation algorithms to generate recommended travel plans. We have conducted an ICDM validation study based on public datasets to compare our ICDMs performance against other models. Our ICDM returned good results on these public data sets when based on a general-purpose optimisation algorithm: Ant Colony Optimisation (ACO). It also performed satisfactorily on a dataset that we assembled from online sources. The second model we propose, the Happiness Model (HM), maximises users satisfaction levels on a tour based on their likes and dislikes. This is relevant because existing models are limited by their propensity to maximise the benefits gained by visiting particular Points Of Interest (POIs) while ignoring other important factors such as connections and waiting times. The main aim of our HM is to simulate a users feelings during their travel experience by maximising the main factors affecting their travel-satisfaction levels. Specifically, the HM optimises travellers journeys not only based on their particular preferences but also considering their effort spent for gaining access to their preferences, which is then reflected their overall happiness level. The validation results demonstrated that the HM model can be used to maximise user satisfaction and represents an abstract model that is able to handle any factor likely to affect user satisfaction. Third, we have addressed the Group Tourist Trip Design Problem (GTTDP), which involves determining a satisfactory trip plan for a group of tourists visiting several POIs. Our proposed, Group Tour Trip Recommender Model (GTTRM) is designed to solve the GTTDP by maximising the group members respective satisfaction levels and reducing the potential for conflict among group members. The novelty of the GTTRM lies in the fact that it solves the GTTDP by deciding on the optimum way to divide up a particular group into a number of sub-groups for specific parts of the trip. Existing models are limited in this respect because they are only able to split up a group at the start of the trip and build recommendations for each group member separately. The results of the GTTRM show that it is effective at maximising individuals satisfaction levels within a group-travel context. In summary, this thesis introduces the three novel models mentioned above to facilitate the building of travel recommendations based on TRSs. Specifically, we show that by considering TRSs as an optimisation problem, we are able to provide highly accurate travel recommendations and overcome the limitations of existing approaches.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available