Cross-domain recommender system using generalized canonical correlation analysis
- PDF / 1,425,376 Bytes
- 27 Pages / 439.37 x 666.142 pts Page_size
- 80 Downloads / 213 Views
Cross-domain recommender system using generalized canonical correlation analysis Seyed Mohammad Hashemi1 · Mohammad Rahmati1 Received: 14 August 2019 / Revised: 21 July 2020 / Accepted: 25 July 2020 / Published online: 14 September 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Recommender systems provide personalized recommendations to the users from a large number of possible options in online stores. Matrix factorization is a well-known and accurate collaborative filtering approach for recommender system, which suffers from cold-start problem for new users and items. When new users join the system, it will take some time before they enter some ratings in the system, until that time, there are not enough ratings to learn the matrix factorization model. Using auxiliary data such as user’s demographic, ratings and reviews in relevant domains, is an effective solution to reduce the new user problem. In this paper, we used the data of users activity from auxiliary domains to build domainindependent users representation that could be used to predict users ratings in the target domains. We proposed an iterative method which applied MAX-VAR generalized canonical correlation analysis (GCCA) on user’s latent factors learned from matrix factorization on each domain. Also, to improve the capability of GCCA to learn latent factors for new users, we propose a generalized canonical correlation analysis by inverse sum of selection matrices (GCCA-ISSM) approach, which provides better recommendations in cold-start scenarios. The proposed approach is extended using content-based features like topic models extracted from user’s reviews. We demonstrate the accuracy and effectiveness of the proposed approaches on cross-domain rating predictions using comprehensive experiments on Amazon and MovieLens datasets. Keywords Collaborative filtering · Cross-domain recommender system · Generalized canonical correlation analysis · Transfer learning
1 Introduction With the development of Internet in recent years, market services such as online marketplace, online book stores and online music stream services play an important role in e-commerce.
B
Mohammad Rahmati [email protected] Seyed Mohammad Hashemi [email protected]
1
Computer Engineering and Information Technology, Amirkabir University of Technology, Tehran, Iran
123
4626
S. M. Hashemi, M. Rahmati
At the beginning, these services were just a database of different products, which provided the ability to search products in different categories. Gradually, and as the number of products increased, users faced more options to choose on these websites. For example, in an online movie store, as the number of movies increases, finding a proper movie among the large number of existing movies becomes difficult and users encounter information overload. In order to solve this problem, new features such as personal search and product recommendations based on the user’s interest are added to these sites, which not only improve user experience and increase user sa
Data Loading...
