Learning binary codes for collaborative filtering

Abstract

This paper tackles the efficiency problem of making recommendations in the context of large user and item spaces. In particular, we address the problem of learning binary codes for collaborative filtering, which enables us to efficiently make recommendations with time complexity that is independent of the total number of items. We propose to construct binary codes for users and items such that the preference of users over items can be accurately preserved by the Hamming distance between their respective binary codes. By using two loss functions measuring the degree of divergence between the training and predicted ratings, we formulate the problem of learning binary codes as a discrete optimization problem. Although this optimization problem is intractable in general, we develop effective relaxations that can be efficiently solved by existing methods. Moreover, we investigate two methods to obtain the binary codes from the relaxed solutions. Evaluations are conducted on three public-domain data sets and the results suggest that our proposed method outperforms several baseline alternatives.

Collaborative Hashing

“…Inter-View collaborative filtering [10,20]. In this paper we propose a collaborative hashing (CH) scheme for nearest neighbor search with data in matrix form.…”

“…Therefore, the proposed collaborative hashing can serve as a unified framework for applications including: (1) search inside a single view: the most typical example is the visual search using local descriptors [16,17]; (2) search across different views: Recommendation using user-item ratings falls in this direction [2,10,20]. In many successful recommendation algorithms, matrix factorization serves as a core technique for collaborative filtering [10].…”

“…Recently, such bag-ofwords (BoW) model has also been widely used in computer vision and multimedia retrieval, which mainly captures the correlations between local features (even the dictionaries using sparse coding) and visual objects [16,17]. Besides the search applications, the rapidly developed recommendation studies based on collaborative filtering, analyzing relationships between users and interdependencies among items to identify new user-item associations, also impose strong requirements on the nearest neighbor search with a collected rating matrix between users and items [2,10,20].…”

“…It is difficult to directly apply the traditional hashing methods to simultaneously hashing two types of entities and meanwhile preserving their correlations. Recently, [19] and [20] attempted to address the problem by concentrating on the duality between views under different scenarios. [19] viewed both documents and terms as the same type of entities in a bipartite graph, and pursued the binary codes for them following the idea of spectral hashing [18].…”

“…[19] viewed both documents and terms as the same type of entities in a bipartite graph, and pursued the binary codes for them following the idea of spectral hashing [18]. To handle the unobserved/missing ratings in collaborative filtering, [20] directly learned the binary codes that can recover the observed item preferences of all users. Both methods mainly explore the correlations between two views in order to preserve the occurrences or preferences in the training matrix.…”