A Study on Recommender System Based on Latent Class Model for High Dimensional and Sparse Data  

Oleh:

Sakamoto, Shunsuke ; Mikawa, Kenta ; Goto, Masayuki

Jenis: Article from Proceeding
Dalam koleksi: The 14th Asia Pacific Industrial Engineering and Management Systems Conference (APIEMS), 3-6 December 2013 Cebu, Philippines, page 1-8.
Topik: Collaborative Filtering; Latent Class Model; Latent Dirichlet Allocation; Clustering
Fulltext: 1038.pdf (556.82KB)

Isi artikel

Recently, recommender systems have become an important web-marketing tool because of the diversity of items on electronic commerce sites (EC sites) and the diversification of users’ preferences. One of the typical approaches of recommender systems is collaborative filtering. In this study, we focus on Latent Dirichlet Allocation (LDA), which is one of latent class models. LDA can represent the heterogeneity of the users' preferences andthe characteristics of the items by introducing latent classes. On the other hand, the data density handled in EC sites is generally high dimensional and so sparse because of a very large number of items. Therefore, it is necessary to consider how to handle high dimensional and sparse data when constructing recommender system. However, it is often impossible to learn parameters of LDA when learning data is extremely smaller than the number of parameters. Therefore, the system cannot recommend the suitable items to each user because parameter estimation is not sufficient.Such a problem may occur with other latent class models.To solve this problem, we propose a new way to apply the latent class model to a high dimensional and sparse data. In this study, we introduce a clustering method.A clustering method, e.g. k-means, reduces parameter dimension of the LDA model. That enables to estimate parameters stably in case of the high dimensional and sparse data. We verify the effectiveness of the proposed method for the high dimensional and sparse data.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali