Basic Matrix Factorization Research Articles

Research on product recommendation based on matrix factorization models fusing user reviews

Nowadays, recommendation models based on matrix factorization (MF) suffer from the problem of rating sparsity because user-product rating matrix is usually sparse. To address the problem, it is significant to fuse some contextual data or side information on basic MF models. According to this core idea, this paper proposes a modified recommendation model, MFFR (matrix factorization fusing reviews) which recommend products by considering the fusing information on user reviews and user ratings. First, MFFR constructs user-product preference matrix from user reviews by using Latent Dirichlet Allocation (LDA) topic model. Then MFFR predicts ratings and generates personalized top-n recommendation products by using MF model to learn comprehensive latent factors of user-product rating matrix and user-product preference matrix simultaneously. The experimental results of three published datasets demonstrate that our model MFFR can achieve more accurate predicted ratings and hits more correct products of top-n recommendation than the comparative traditional models. MFFR can effectively raise the quality of recommendation, especially in the high level of rating sparsity.

Attribute interaction aware matrix factorization method for recommendation

Matrix factorization (MF) models are effective and easy to expand and are widely used in industry, such as rating prediction and item recommendation. The basic MF model is relatively simple. In practical applications, side information such as attributes or implicit feedback is often combined to improve accuracy by modifying the model and optimizing the algorithm. In this paper, we propose an attribute interaction-aware matrix factorization (AIMF) method for recommendation tasks. We partition the original rating matrix into different sub-matrices according to the attribute interactions, train each sub-matrix independently, and merge all the latent vectors to generate the final score. Since the generated sub-matrices vary in size, an adaptive regularization coefficient optimization strategy and an adaptive latent vector dimension optimization strategy are proposed for sub-matrix training, and a variety of latent vector merging methods are put forward. The method AIMF has two advantages. When the original rating matrix is particularly large, the training time complexity of the MF-based model becomes higher and the update cost of the model is also higher. In AIMF, because each sub-matrix is usually much smaller than the original rating matrix, the training time complexity is greatly reduced after using parallel computing technology. Secondly, in AIMF, it is not necessary to modify the matrix factorization model to incorporate attributes and their interactive information into the model to improve the performance. The experimental results on the two classic public datasets MovieLens 1M and MovieLens 100k show that AIMF can not only effectively improve the accuracy of recommendation, but also make full use of parallel computing technology to improve training efficiency without modifying the matrix factorization model.

Personalized Landmark Recommendation Based on Geotags from Photo Sharing Sites

Geotagged photos of users on social media sites provide abundant location-based data, which can be exploited for various location-based services, such as travel recommendation. In this paper, we propose a novel approach to a new application, i.e., personalized landmark recommendation based on users’ geotagged photos. We formulate the landmark recommendation task as a collaborative filtering problem, for which we propose a category-regularized matrix factorization approach that integrates both user-landmark preference and category-based landmark similarity. We collected geotagged photos from Flickr and landmark categories from Wikipedia for our experiments. Our experimental results demonstrate that the proposed approach outperforms popularity-based landmark recommendation and a basic matrix factorization approach in recommending personalized landmarks that are less visited by the population as a whole.

Web document summarization by exploiting social context with matrix co-factorization

In the context of social media, users usually post relevant information corresponding to the contents of events mentioned in a Web document. This information posses two important values in that (i) it reflects the content of an event and (ii) it shares hidden topics with sentences in the main document. In this paper, we present a novel model to capture the nature of relationships between document sentences and post information (comments or tweets) in sharing hidden topics for summarization of Web documents by utilizing relevant post information. Unlike previous methods which are usually based on hand-crafted features, our approach ranks document sentences and user posts based on their importance to the topics. The sentence-user-post relation is formulated in a share topic matrix, which presents their mutual reinforcement support. Our proposed matrix co-factorization algorithm computes the score of each document sentence and user post and extracts the top ranked document sentences and comments (or tweets) as a summary. We apply the model to the task of summarization on three datasets in two languages, English and Vietnamese, of social context summarization and also on DUC 2004 (a standard corpus of the traditional summarization task). According to the experimental results, our model significantly outperforms the basic matrix factorization and achieves competitive ROUGE-scores with state-of-the-art methods.

Field-Aware Matrix Factorization for Recommender Systems

Predicting user response is one of the core machine learning tasks in recommender systems (RS). The matrix factorization (MF)-based model has been proved to be a useful tool to improve the performance of recommendation. Many existing matrix factorization-based models mainly rely on adding some side information into basic MF to enable the model to fully express the data. However, most of the side information is measured based on the statistics or empirical formula. Also, the latent features of side information cannot be deeply mined. In this paper, we focus on mining the influence of field information (useful side information) to improve the performance of prediction. Based on the MF framework, we propose a field-aware matrix factorization (FMF) model. In FMF, the interactions between user/item and field can be captured and learned in the latent vector spaces. We propose efficient implementations to train FMF. Then, we comprehensively analyze FMF and compare this model with the state-of-the-art models. The analysis of experiments on two large data sets demonstrates that our method is very useful in RS.

Recommendation using DMF-based fine tuning method

Recommender Systems (RS) have been comprehensively analyzed in the past decade, Matrix Factorization (MF)-based Collaborative Filtering (CF) method has been proved to be an useful model to improve the performance of recommendation. Factors that inferred from item rating patterns shows the vectors which are useful for MF to characterize both items and users. A recommendation can concluded from good correspondence between item and user factors. A basic MF model starts with an object function, which is consisted of the squared error between original training matrix and predicted matrix as well as the regularization term (regularization parameters). To learn the predicted matrix, recommender systems minimize the squared error which has been regularized. However, two important details have been ignored: (1) the predicted matrix will be more and more accuracy as the iterations carried out, then a fix value of regularization parameters may not be the most suitable choice. (2) the final distribution trend of ratings of predicted matrix is not similar with the original training matrix. Therefore, we propose a Dynamic-MF algorithm and fine tuning method which is quite general to overcome the mentioned detail problems. Some other information, such as social relations, etc, can be easily incorporated into this method (model). The experimental analysis on two large datasets demonstrates that our approaches outperform the basic MF-based method.