Using Heterogeneous Social Media as Auxiliary Information to Improve Hotel Recommendation Performance

Research conducted on social media is currently increasing. Information obtained by users of social media has resulted in the development of many recommendation systems that analyze user preferences in an attempt to locate the most suitable products for recommendation to users. A great number of peo...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2018-01, Vol.6, p.42647-42660
Main Authors: Chang, Jui-Hung, Tsai, Chen-En, Chiang, Jung-Hsien
Format: Article
Language:English
Subjects:
Algorithms
Analytical user posting behavior algorithm
Collaboration
Digital media
Facebook
Filtering
Heterogeneous networks
hotel recommendation system
Hotels
mean reciprocal rank
Prediction algorithms
Recommender systems
Social networks
Twitter
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Research conducted on social media is currently increasing. Information obtained by users of social media has resulted in the development of many recommendation systems that analyze user preferences in an attempt to locate the most suitable products for recommendation to users. A great number of people have used social media platforms, such as Twitter, to develop a hotel recommendation system. Here, we propose a Twitter-based recommendation system via the aid of heterogeneous social media. First, a model is designed to predict user preferences by improving matrix factorization based on user preferences and users' personal data, where basic information about hotels collected from Yelp is used as auxiliary information. On the other hand, an analytical user posting behavior algorithm is created for establishing users' posting behavior vectors based on earlier posts in Twitter and Yelp. This results of the experiments show that the proposed method can improve accuracy by 30% in terms of RECALL compared with the Twitter-based recommendation system without the use of heterogeneous social media. Furthermore, it can improve the accuracy of the mean reciprocal rank by 80% and can increase precision by as much as 100%.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2018.2855690