Personalized Travel Recommendation System Using Average Cumulative Rating Matrix Factorization Technique: Concept and Framework

Abstract

Listen

Translate article

Recommendation systems in travel applications have a purpose to provide custom-made results to travelers while making a travel plan. These recommendation systems should be adaptable if user preferences change dynamically. To get custom-made results, the recommendation systems should be provided with traveler’s interests such as traveler’s specifications, preferences concerning destinations, type of activities they are very much interested to do in their travel plan. However, current recommendation systems are unable to fetch required features from travelers and destination places. Moreover, current systems are lacking to recommend destination places by considering social interest and their experience (i.e. recommendations by considering many traveler’s interests, for example, when two travelers interest matches the places visit by the first traveler can be suggested to the second traveler or vice-versa and travelers experience concerning particular destination place). To address the issues and problems of the current system, we propose and implement a tourist recommendation system which is termed as Average Cumulative Rating (ACR) that supports the extraction of rating and experience which is in the form of text description. The overall score is computed based on rating and traveler experience and feed to the traditional Matrix Factorization (MF) technique for providing custom results for travelers.