Latent Semantic Indexing-Based Hybrid Collaborative Filtering for Recommender Systems
[ X ]
Tarih
2022
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Springer Heidelberg
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
Advances in information technologies increase the number and diversity of digital objects. This increase poses significant problems in reaching the target audience of digital products. Recommender systems (RS) that propose digital objects according to user profiles aim to deal with these problems. In collaborative recommender systems (CRS), recommendations are made considering similar digital objects. In this study, a hybrid model based on latent semantic indexing (LSI) is proposed for the CRS. User-based, item-based, and hybrid models have been developed by using the LSI, which is generally encountered in text analysis, information retrieval, and information access. These improved models were compared with the models based on the most commonly used Pearson correlation coefficient (PCC) in the CRS. Accordingly, it was observed that predictions were better in all models based on LSI. The developed models have lower computational complexity due to the dimension reduction process. Besides, the proposed hybrid model produced more accurate predictions than the user-based and the item-based models.
Açıklama
Anahtar Kelimeler
Recommender systems; Collaborative filtering; Latent semantic indexing; Dimension reduction
Kaynak
Arabian Journal For Science and Engineering
WoS Q Değeri
Q2
Scopus Q Değeri
Q1
Cilt
47
Sayı
8
