LorSLIM: Low Rank Sparse Linear Methods for Top-N Recommendations

Cheng, Yao; Yin, Liang; Yu, Yong

doi:10.1109/icdm.2014.112

Cited by 39 publications

(24 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In general, a recommender system models the user preference though the user profiles, such as the user-item pair. The MF is a successful method to model the user preferences with two latent-factors by factorizing the user-item matrix [40][41][42]. A user preference for the item i that presents ̂ is represented by the user-item latent factors ( and ).…”

Section: B Modeling User Preferencementioning

confidence: 99%

A Hybrid Recommender System for Sequential Recommendation: Combining Similarity Models With Markov Chains

2020

View full text Add to dashboard Cite

With the growth of artificial intelligence technology, the importance of recommender systems that recommend personalized content has increased. The general form of the recommender system usually analyzes the users' log information to provide them with contents that they are interested in. However, to enable users to receive more suitable and personalized content, additional information must be considered besides the user's log information. We develop, in the present study, a hybrid recommender system that unifies similarity models-collaborative and content-based-with Markov chains for a sequential recommendation (called U2CMS). U2CMS takes into account both sequential patterns and information about contents to find accurate relationships between items. It uses a higher-order Markov chain to model sequential patterns over several time steps, as well as the textual information of the content to model the recommender system. To show the effectiveness of the U2CMS-with regard to handling sparsity issues, different Nordered Markov Chain, and accurately identifying similarities between items, we carried out several experiments on various Amazon datasets. Our results show that the U2CMS not only has superior performance compared to existing state-of-the-art recommendation systems (including deep-learning based systems), but also it successfully handles sparsity issues better than other approaches. Moreover, U2CMS appears to perform stable when it comes to different N-ordered Markov Chain. Lastly, through visualization, we show the success of our proposed content-based filtering model in identifying similar items. INDEX TERMS Item Similarity Model, Content-based Filtering, Hybrid Recommendation, Sequential Recommendation HONG-JUN JANG received the B.Sc. and Ph.D. degrees in computer science education from Korea University, Seoul, South Korea. His research interests include data mining, machine learning, and data base.

show abstract

Section: B Modeling User Preferencementioning

confidence: 99%

A Hybrid Recommender System for Sequential Recommendation: Combining Similarity Models With Markov Chains

2020

View full text Add to dashboard Cite

show abstract

“…The optimization problem 4 is usually dealt in several works [19,20,5,30] in field of Top-N recommendation problem. Recently, [30] showed that this problem could be simply solved in closed form by method of Lagrange multipliers.…”

Section: Closed Form Solution Of Self-expressive Layermentioning

confidence: 99%

Deep Closed-Form Subspace Clustering

Seo¹,

Koo²,

Jeon³

2019

2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)

View full text Add to dashboard Cite

We propose Deep Closed-Form Subspace Clustering (DCFSC), a new embarrassingly simple model for subspace clustering with learning non-linear mapping. Compared with the previous deep subspace clustering (DSC) techniques, our DCFSC does not have any parameters at all for the self-expressive layer. Instead, DCFSC utilizes the implicit data-driven self-expressive layer derived from closed-form shallow auto-encoder. Moreover, DCFSC also has no complicated optimization scheme, unlike the other subspace clustering methods. With its extreme simplicity, DCFSC has significant memory-related benefits over the existing DSC method, especially on the large dataset. Several experiments showed that our DCFSC model had enough potential to be a new reference model for subspace clustering on large-scale high-dimensional dataset.

show abstract

“…It has been shown that SLIM outperforms other Top-N recommendation methods. A drawback of SLIM is that it can only model relations between items that have been co-purchased/co-rated by at least one user [13]. Therefore, it fails to capture the potential dependencies between items that have not been co-rated by at least one user, while modeling relations between items that are not co-rated is essential for good performance of item-based approaches in sparse datasets.…”

Section: Relevant Researchmentioning

confidence: 99%

“…We compare the performance of the proposed method with seven stateof-the-art Top-N recommendation algorithms, including the item neighborhood-based collaborative filtering method ItemKNN [9], two MF-based methods PureSVD [11] and WRMF [34], SLIM [8] and LorSLIM [13]. We also examine two ranking/retrieval criteria based methods BPRMF and BPRKNN [35], where Bayesian personalized ranking (BPR) criterion is used which measures the difference between the rankings of userpurchased items and the remaining items.…”

Section: Comparison Algorithmsmentioning

confidence: 99%

Top-N Recommendation with Novel Rank Approximation

Kang

Cheng

2016

Proceedings of the 2016 SIAM International Conference on Data Mining

View full text Add to dashboard Cite

The importance of accurate recommender systems has been widely recognized by academia and industry. However, the recommendation quality is still rather low. Recently, a linear sparse and low-rank representation of the user-item matrix has been applied to produce Top-N recommendations. This approach uses the nuclear norm as a convex relaxation for the rank function and has achieved better recommendation accuracy than the state-of-the-art methods. In the past several years, solving rank minimization problems by leveraging nonconvex relaxations has received increasing attention. Some empirical results demonstrate that it can provide a better approximation to original problems than convex relaxation. In this paper, we propose a novel rank approximation to enhance the performance of Top-N recommendation systems, where the approximation error is controllable. Experimental results on real data show that the proposed rank approximation improves the Top-N recommendation accuracy substantially.

show abstract

LorSLIM: Low Rank Sparse Linear Methods for Top-N Recommendations

Cited by 39 publications

References 19 publications

A Hybrid Recommender System for Sequential Recommendation: Combining Similarity Models With Markov Chains

A Hybrid Recommender System for Sequential Recommendation: Combining Similarity Models With Markov Chains

Deep Closed-Form Subspace Clustering

Top-N Recommendation with Novel Rank Approximation

Contact Info

Product

Resources

About