Learning bidirectional asymmetric similarity for collaborative filtering via matrix factorization

Memory-based collaborative filtering (CF) techniques have been widely implemented for predicting ratings to unseen items by aggregating ratings of similar users or items in recommender systems (RS). Usually, sufficient ratings from similar users or similar items are not available in the rating matrix, due to the data sparsity problem. Further, these techniques suffer from correlation based problems inherent in used similarity measures. Consequently, higher prediction accuracy cannot be achieved. In this paper, we propose the use of fuzzy Naive Bayesian (FNB) classifier for user based CF and item based CF for implicitly computing similarity between users as well as items on the basis of conditional probabilities and develop fuzzy Naive Bayesian classifier to user based CF (FNB-UB-CF) and item based CF (FNB-IB-CF). We further develop a hybrid RS (FNB-UB-IB-CF) by combining the proposed FNB-UB-CF and FNB-IB-CF. Their combinations would be helpful in alleviating the sparsity because both user ratings and item ratings are employed. Experimental results demonstrate that the proposed methods are indeed more robust against data sparsity and give better recommendation quality using a popular MovieLens dataset.

Collaborative and content-based filtering are the majorly used approaches towards implementing recommender systems which can successfully predict the item(s) to be recommended to users based upon their preferences. Each of the methods has its own advantages and disadvantages that are specific to the situation they are used in. The hybrid approach aims to combine both techniques in multiple ways to optimize results and overcome shortcomings. In this paper, we propose a hybrid approach based recommendation engine to improve accuracy. While content based recommendations are made on the similarity of items’ attributes, collaborative recommendations are based on similarity of users. Collaborative filtering is achieved by matrix factorization technique. One of the most effective algorithms for matrix factorization: Singular Value Decomposition (SVD) has been used to perform collaborative filtering and combined with content based model to predict the item ratings per user. The paper aims to achieve the final result with minimum possible errors as per user preferences and recommended items.

The rating matrix of a personalized recommendation system contains a high percentage of unknown rating scores which lowers the quality of the prediction. Besides, during data streaming into memory, some rating scores are misplaced from its appropriate cell in the rating matrix which also decrease the quality of the prediction. The singular value decomposition algorithm predicts the unknown rating scores based on the relation between the implicit feedback of both users and items, but exploiting neither the user similarity nor item similarity which leads to low accuracy predictions. There are several factorization methods used in improving the prediction performance of the collaborative filtering technique such as baseline, matrix factorization, neighbour-base. However, the prediction performance of the collaborative filtering using factorization methods is still low while baseline and neighbours-base have limitations in terms of over fitting. Therefore, this paper proposes Ensemble Divide and Conquer (EDC) approach for solving 2 main problems which are the data sparsity and the rating scores’ deviation (misplace). The EDC approach is founded by the Singular Value Decomposition (SVD) algorithm which extracts the relationship between the latent feedback of users and the latent feedback of the items. Furthermore, this paper addresses the scale of rating scores as a sub problem which effect on the rank approximation among the users’ features. The latent feedback of the users and items are also SVD factors. The results using the EDC approach are more accurate than collaborative filtering and existing methods of matrix factorization namely SVD, baseline, matrix factorization and neighbours-base. This indicates the significance of the latent feedback of both users and items against the different factorization features in improving the prediction accuracy of the collaborative filtering technique.

Evaluation Of Recommendation System On A Movie Dataset

Recommendation systems have permeated society and everyones daily life with the expanding use of applications. It plays an important role in providing recommendations to active users based on either their historical interests or interests shown by similar users. This paper aims to introduce several components of a recommendation system, such as Collaborative Filtering (CF) and Matrix Factorization, and combining the dominant approach and main technique: Singular Value Decomposition (SVD). Additionally, with the advances in technology, various recently established SVD-based models are mentioned in this paper, attempting to give an answer to the question that though studies have shown the coexistence of advantages and disadvantages for each mode, the trade-offs between their predictive accuracy, computational cost, and robustness to data sparsity have still not been fully answered. Furthermore, this paper concentrates on the comparison and analysis of diverse matrix factorization variants, meanwhile evaluating their performances. For instance, basic SVD, BiasSVD, and SVD++. Finding indicates that SVD shows more efficiency in small-scale datasets, BiasSVD is suitable for cases with noticeable biases, and SVD++ gives high-quality results when handling both explicit and implicit feedback data. This study provides practical guidelines for practitioners to select the most desirable model based on their existing database.

Collaborative filtering recommendation algorithm is the most important recommended application methods at present. The selection of users (items) similarity measurement methods in this method may affect the final recommendation effect. Probability matrix factorization algorithm is one of the more and more methods applied in collaborative filtering algorithms. This paper proposes to convert the users (items) consumption network into a bipartite graph and use belief propagation algorithm to obtain fuzzy nearest neighbor set, improving users (items) similarity measure and incorporating it into probability matrix factorization. Experimental tests on the data set showed that the proposed method performance was good.

Making recommendations from top-N user-item subgroups

Matrix Factorization (MF) is one of the most successful Collaborative Filtering (CF) techniques used in recommender systems due to its effectiveness and ability to deal with very large user-item rating matrix. However, when the rating matrix sparseness increases its performance deteriorates. Expanding MF to include side-information of users and items has been shown by many researchers both to improve general recommendation performance and to help alleviate the data-sparsity and cold-start issues in CF. In regard to item feature side-information, most schemes incorporate this information through a two stage process: intermediate results (e.g., on item similarity) are first computed based on item attributes; these are then combined with MF. In this paper, focussing on item side-information, we propose a model that directly incorporates item features into the MF framework in a single step process. The model, which we name FeatureMF, does this by projecting every available attribute datum in each of the item features into the same latent factor space with users and items, thereby in effect enriching item representation in MF. Results are presented of comparative performance experiments of the model against three state-of-the-art item information enriched models, as well as against four reference benchmark models, using two public real-world datasets, Douban and Yelp, with four training:test ratio scenarios each. It is shown to yield the best recommendation performance over all these models across all contexts including data-sparsity situations, in particular, achieving over 0.9% to over 6.5% MAE recommendation performance improvement over the next best model, HERec. FeatureMF is also found to alleviate cold start and to scale well, almost linearly, in regard to computational time, as a function of dataset size.

A new item similarity based on [formula omitted]-divergence for collaborative filtering in sparse data

Collaborative filtering, the most successful recommender system technology to date, helps people make choices based on the opinions of other people. Existing collaborative filtering methods, mainly user-based and item-based methods, predict new ratings by aggregating rating information from either similar users or items. However, a large amount of ratings of similar items or similar users may be unavailable because of the sparse characteristic inherent to the rating data. For this reason, we present a Hybrid Predictive Algorithm with Smoothing (HSPA). HSPA uses item-based methods to provide the basis for data smoothing and builds the predictive model based on both users? aspects and items? aspects in order to ensure robust to data sparsity and predictive accuracy. Moreover, HSPA utilizes the user clusters to achieve high scalability. Experimental results from real datasets show that HSPA effectively contributes to the improvement of prediction on sparse data.

Collaborative filtering recommender systems provide their users with relevant items based on information from other similar users. Popular collaborative filtering approaches such as Pearson correlation coefficient and cosine similarity, compute the similarity between users based on the set of their co-rated items. However, similarities are commonly computed without taking the popularity of the set of two users' co-rated items into consideration, e.g. an item rated by very many users should have less impact on the similarity measure, and analogously an item rated by few should have a larger impact on the similarity score of two users. In this paper, we investigate the effects of common weighting schemes on different types of users, i.e. new users with few ratings (so-called cold start users), post cold start users, and power users. Empirical studies over two datasets have shown in which of these cases weighting schemes are beneficial in terms of recommendation quality.

Towards comprehensive approaches for the rating prediction phase in memory-based collaborative filtering recommender systems

Because of its simplicity and effectiveness, collaborative filtering (CF) became one of the most successful recommendation algorithms. User-based CF is one classic method of CF algorithms. In order to solve the problem that common rating items are often too few to be used to effectively calculate the similarity of two users in user-based CF, we proposed an improved collaborative filtering model with item similarity called ISCF in this paper. In ISCF model, the similarity of items was considered in user-based collaborative filtering, which contributes to alleviate the problem of data sparsity and therefore calculate the similarity of user. Experimental results illustrate that our approach ISCF outperforms the average method and user-based CF. Compared with user-based CF, the average improvement in the percentage of ISCF at MAE and RMSE are 21.9% and 17.7%, respectively. In addition, our approach ISCF can predict more items than user-based CF, and the average improvement in the percentage of ISCF at prediction diversity is 33.86%.

Recommender Systems (RSs) have emerged as powerful tools to provide the users with personalized recommendations and to guide them in their decision making process. Among the various recommendation approaches, Collaborative Filtering (CF) is considered as one of the most popular techniques in RSs. CF techniques are categorized into model-based and memory-based. Model-based approaches consist in learning a model from past ratings to perform predictions while memory-based ones predict ratings by selecting the most similar users (user-based) or the most similar items (item-based). In both types, recommendations are fully based on users’ past ratings. However, aside from users’ ratings, exploiting additional information such as items’ features would enhance the accuracy of the provided predictions. Another crucial challenge in the RSs area would be to handle uncertainty arising throughout the prediction process. That is why, in this paper, we propose an item-based Collaborative Filtering under the belief function theory that not only takes advantages of both model- and memory- based CF approaches but also integrates items’ contents in the recommendation process.

Collaborative filtering (CF) is the state-of-the-art approach to item recommendation. However, it can neither recommend new items with no user feedbacks, nor could it recommend long-tail items easily. Content-based filtering can solve both problems through content analysis. However, content-based filtering alone has a much worse performance than CF. In this paper, we fuse user feedbacks and content analysis into the probabilistic matrix factorization framework. In particular, we propose a recursive dynamic programming approach to computing item similarity matrix from item content. Item latent factors are predicted from the item similarity matrix when no usage data is available. We investigate how performances of recommendation algorithms vary on items with different popularities. Results show that our approach has better performance than the same hybrid model with naive item similarity measures and Matrix Factorization.