Collaborative Filtering Tasks Research Articles

Exploration of Collaboration Filtering Techniques for Product Recommendation

Today, recommendation system has been globally adopted as the most effective and reliable search engine for knowledge extraction in the field of education, economics and scientific research. Collaborative filtering is a proven techniques used in recommender system to make predictions or recommendations of the unknown preferences for users based on the known user preferences. In this paper, collaborative filtering task and their challenges are explored, study the different recommendation techniques and evaluate their performance using different metrics.

Neural variational collaborative filtering with side information for top-K recommendation

Collaborative filtering (CF) is one of the most widely applied models for recommender systems. Despite its success, CF-based methods suffer from rating sparsity and cold-start problem, which leads to poor quality of recommendations. Previous studies have gave great attention to construct hybrid methods, by incorporating side information and user rating. Variational autoencoder (VAE) has been confirmed to be highly effective in CF task, due to its Bayesian nature and non-linearity. However, rating sparsity remains a great challenge to most VAE models, which leads to poor latent user/item representations. In addition, most existing VAE-based methods model either latent user factors or latent item factors, resulting in the incapacity to recommend items to a new user or suggest a new item to existing users. To address these problems, we design a novel deep hybrid framework for top-k recommendation, neural variational collaborative filtering (NVCF), and propose three NVCF-based instantiation. In generative process, the side information of user and item is incorporated to alleviate rating sparsity, for learning better latent user/item representations. In inference process, a Stochastic Gradient Variational Bayes approach is employed to approximate the unmanageable distributions of latent user/item factors. Experiments performed on four public datasets have indicated our methods significantly outperform the state-of-the-art hybrid CF models and VAE-based methods.

Stochastic block models with multiple continuous attributes

The stochastic block model (SBM) is a probabilistic model for community structure in networks. Typically, only the adjacency matrix is used to perform SBM parameter inference. In this paper, we consider circumstances in which nodes have an associated vector of continuous attributes that are also used to learn the node-to-community assignments and corresponding SBM parameters. Our model assumes that the attributes associated with the nodes in a network’s community can be described by a common multivariate Gaussian model. In this augmented, attributed SBM, the objective is to simultaneously learn the SBM connectivity probabilities with the multivariate Gaussian parameters describing each community. While there are recent examples in the literature that combine connectivity and attribute information to inform community detection, our model is the first augmented stochastic block model to handle multiple continuous attributes. This provides the flexibility in biological data to, for example, augment connectivity information with continuous measurements from multiple experimental modalities. Because the lack of labeled network data often makes community detection results difficult to validate, we highlight the usefulness of our model for two network prediction tasks: link prediction and collaborative filtering. As a result of fitting this attributed stochastic block model, one can predict the attribute vector or connectivity patterns for a new node in the event of the complementary source of information (connectivity or attributes, respectively). We also highlight two biological examples where the attributed stochastic block model provides satisfactory performance in the link prediction and collaborative filtering tasks.

Neural Binary Representation Learning for Large-Scale Collaborative Filtering

Integrating hashing into collaborative filtering (CF) has been a promising solution to address the efficiency problem of large-scale recommender systems, i.e., hashing users and items into binary codes and then making recommendations in Hamming space. However, most of the existing hashing methods for CF solely focus on modeling the user-item similarity (a.k.a. preference) but omit the user-user and item-item similarities, which cannot well preserve the original geometry in the vector space. Another tough issue is how to effectively tackle the encoding loss from the continuous vector space to the discrete Hamming space. In this paper, we propose a neural binary representation (NBR) learning approach by combining hashing with a neural network for the large-scale CF tasks. Our NBR problem is formulated as a Hamming similarity loss plus two anchor smoothing terms, which jointly preserve the intrinsic user-item, user-user, and item-item similarities. To compensate the encoding loss introduced by hashing, the anchors for users and items are pre-learned using both user-item interactions and the side-information of users and items through an AutoEncoder model. In particular, to solve the NBR problem, we develop a computationally efficient algorithm, which learns the binary codes in a fast bit-by-bit way and achieves less quantization deviation than the conventional CF hashing schemes. The extensive experiments on three benchmarks validate the superiority of our NBR approach in comparison to the state-of-the-art CF hashing methods.

Collaborative Filtering With User-Item Co-Autoregressive Models

Deep neural networks have shown promise in collaborative filtering (CF). However, existing neural approaches are either user-based or item-based, which cannot leverage all the underlying information explicitly. We propose CF-UIcA, a neural co-autoregressive model for CF tasks, which exploits the structural correlation in the domains of both users and items. The co-autoregression allows extra desired properties to be incorporated for different tasks. Furthermore, we develop an efficient stochastic learning algorithm to handle large scale datasets. We evaluate CF-UIcA on two popular benchmarks: MovieLens 1M and Netflix, and achieve state-of-the-art performance in both rating prediction and top-N recommendation tasks, which demonstrates the effectiveness of CF-UIcA.

A comparative study on effects of some exclusive conditions in fuzzy co-clustering for collaborative filtering

Fuzzy co-clustering is a promising approach for efficiently realizing collaborative filtering, in which personalized recommendation is achieved by summarizing the intrinsic user-item preferences through dual clustering of users and items in cooccurrence information matrices. In cases of applying fuzzy co-clustering, we can select roughly three partition models supported by different partition constraints: user targeting partition, (weak) dual exclusive partition and their hybrid approach. This paper presents a comparative study on the applicability of the three partition models to collaborative filtering tasks through empirical demonstration with two real world data sets. The experimental results reveal that user targeting partition is most suitable for the task while dual exclusive partition can also be used with sparse data sets.

Towards self-learning based hypotheses generation in biomedical text domain.

The overwhelming amount of research articles in the domain of bio-medicine might cause important connections to remain unnoticed. Literature Based Discovery is a sub-field within biomedical text mining that peruses these articles to formulate high confident hypotheses on possible connections between medical concepts. Although many alternate methodologies have been proposed over the last decade, they still suffer from scalability issues. The primary reason, apart from the dense inter-connections between biological concepts, is the absence of information on the factors that lead to the edge-formation. In this work, we formulate this problem as a collaborative filtering task and leverage a relatively new concept of word-vectors to learn and mimic the implicit edge-formation process. Along with single-class classifier, we prune the search-space of redundant and irrelevant hypotheses to increase the efficiency of the system and at the same time maintaining and in some cases even boosting the overall accuracy. We show that our proposed framework is able to prune up to 90% of the hypotheses while still retaining high recall in top-K results. This level of efficiency enables the discovery algorithm to look for higher-order hypotheses, something that was infeasible until now. Furthermore, the generic formulation allows our approach to be agile to perform both open and closed discovery. We also experimentally validate that the core data-structures upon which the system bases its decision has a high concordance with the opinion of the experts.This coupled with the ability to understand the edge formation process provides us with interpretable results without any manual intervention. The relevant JAVA codes are available at: https://github.com/vishrawas/Medline-Code_v2. Supplementary data are available at Bioinformatics online.

Axiomatic Analysis of Language Modelling of Recommender Systems

Language Models constitute an effective framework for text retrieval tasks. Recently, it has been extended to various collaborative filtering tasks. In particular, relevance-based language models can be used for generating highly accurate recommendations using a memory-based approach. On the other hand, the query likelihood model has proven to be a successful strategy for neighbourhood computation. Since relevance-based language models rely on user neighbourhoods for producing recommendations, we propose to use the query likelihood model for computing those neighbourhoods instead of cosine similarity. The combination of both techniques results in a formal probabilistic recommender system which has not been used before in collaborative filtering. A thorough evaluation on three datasets shows that the query likelihood model provides better results than cosine similarity. To understand this improvement, we devise two properties that a good neighbourhood algorithm should satisfy. Our axiomatic analysis shows that the query likelihood model always enforces those constraints while cosine similarity does not.

Matrix completion by least-square, low-rank, and sparse self-representations

Conventional matrix completion methods are generally based on rank minimization. These methods assume that the given matrix is of low-rank and the data points are drawn from a single subspace of low-dimensionality. Therefore they are not effective in completing matrices where the data are drawn from multiple subspaces. In this paper, we establish a novel matrix completion framework that is based on self-representation. Specifically, least-square, low-rank, and sparse self-representations based matrix completion algorithms are provided. The underlying idea is that one data point can be efficiently reconstructed by other data points belonging to a common subspace, where the missing entries are recovered so as to fit the common subspace. The proposed algorithms actually maximize the weighted correlations among the columns of a given matrix. We prove that the proposed algorithms are approximations for rank-minimization of the incomplete matrix. In addition, they are able to complete high-rank or even full-rank matrices when the data are drawn from multiple subspaces. Comparative studies are conducted on synthetic datasets, natural image inpainting tasks, temperature prediction task, and collaborative filtering tasks. The results show that the proposed algorithms often outperform other state-of-the-art algorithms in various tasks.

Making recommendations from top-N user-item subgroups

Group-aware collaborative filtering (CF) has recently become a hot research topic in recommender systems, which typically divides a large CF task on the entire data (i.e. rating matrix) into some smaller CF tasks on subgroups (i.e., sub-matrices). This leads to an effective way to improve current CF systems in accuracy and efficiency. However, existing approaches consider each subgroup separately, ignoring relationships among subgroups. In this paper, motivated by the intuition that there are similar users or items among different subgroups, we propose an improved group-aware CF algorithm which predicts a rating using a weighted sum of similar ratings from multiple subgroups. Our algorithm is based on Matrix Factorization and CodeBook Transfer (CBT), especially that we construct N matrix approximations based on N best sub-matrices, and then integrate the N approximations via a linear combination. We conduct experiments on real-life data to evaluate the performance of our algorithm in comparison with traditional CF algorithms and other state-of-the-art social and group-aware recommendation models. The empirical result and analysis demonstrate that our algorithm achieves a significant increase in recommendation accuracy.

영화 추천 시스템의 초기 사용자 문제를 위한 장르 선호 기반의 클러스터링 기법

Social media has become one of the most popular media in web and mobile application. In 2011, social networks and blogs are still the top destination of online users, according to a study from Nielsen Company. In their studies, nearly 4 in 5active users visit social network and blog. Social Networks and Blogs sites rule Americans’ Internet time, accounting to 23 percent of time spent online. Facebook is the main social network that the U.S internet users spend time more than the other social network services such as Yahoo, Google, AOL Media Network, Twitter, Linked In and so on. In recent trend, most of the companies promote their products in the Facebook by creating the “Facebook Page” that refers to specific product. The “Like” option allows user to subscribed and received updates their interested on from the page. The film makers which produce a lot of films around the world also take part to market and promote their films by exploiting the advantages of using the “Facebook Page”. In addition, a great number of streaming service providers allows users to subscribe their service to watch and enjoy movies and TV program. They can instantly watch movies and TV program over the internet to PCs, Macs and TVs. Netflix alone as the world’s leading subscription service have more than 30 million streaming members in the United States, Latin America, the United Kingdom and the Nordics. As the matter of facts, a million of movies and TV program with different of genres are offered to the subscriber. In contrast, users need spend a lot time to find the right movies which are related to their interest genre. Recent years there are many researchers who have been propose a method to improve prediction the rating or preference that would give the most related items such as books, music or movies to the garget user or the group of users that have the same interest in the particular items. One of the most popular methods to build recommendation system is traditional Collaborative Filtering (CF). The method compute the similarity of the target user and other users, which then are cluster in the same interest on items according which items that users have been rated. The method then predicts other items from the same group of users to recommend to a group of users. Moreover, There are many items that need to study for suggesting to users such as books, music, movies, news, videos and so on. However, in this paper we only focus on movie as item to recommend to users. In addition, there are many challenges for CF task. Firstly, the “sparsity problem”; it occurs when user information preference is not enough. The recommendation accuracies result is lower compared to the neighbor who composed with a large amount of ratings. The second problem is “cold-start problem”; it occurs whenever new users or items are added into the system, which each has norating or a few rating. For instance, no personalized predictions can be made for a new user without any ratings on the record. In this research we propose a clustering method according to the users’ genre interest extracted from social network service (SNS) and user’s movies rating information system to solve the “cold-start problem.” Our proposed method will clusters the target user together with the other users by combining the user genre interest and the rating information. It is important to realize a huge amount of interesting and useful user’s information from Facebook Graph, we can extract information from the “Facebook Page” which “Like” by them. Moreover, we use the Internet Movie Database(IMDb) as the main dataset. The IMDbis online databases that consist of a large amount of information related to movies, TV programs and including actors. This dataset not only used to provide movie information in our Movie Rating Systems, but also as resources to provide movie genre information which extracted from the “Facebook Page”. Formerly, the user must login with their Facebook account to login to the Movie Rating System, at the same time our system will collect the genre interest from the “Facebook Page”. We conduct many experiments with other methods to see how our method performs and we also compare to the other methods. First, we compared our proposed method in the case of the normal recommendation to see how our system improves the recommendation result. Then we experiment method in case of cold-start problem. Our experiment show that our method is outperform than the other methods. In these two cases of our experimentation, we see that our proposed method produces better result in case both cases.

A hierarchical model for ordinal matrix factorization

This paper proposes a hierarchical probabilistic model for ordinal matrix factorization. Unlike previous approaches, we model the ordinal nature of the data and take a principled approach to incorporating priors for the hidden variables. Two algorithms are presented for inference, one based on Gibbs sampling and one based on variational Bayes. Importantly, these algorithms may be implemented in the factorization of very large matrices with missing entries.The model is evaluated on a collaborative filtering task, where users have rated a collection of movies and the system is asked to predict their ratings for other movies. The Netflix data set is used for evaluation, which consists of around 100 million ratings. Using root mean-squared error (RMSE) as an evaluation metric, results show that the suggested model outperforms alternative factorization techniques. Results also show how Gibbs sampling outperforms variational Bayes on this task, despite the large number of ratings and model parameters. Matlab implementations of the proposed algorithms are available from cogsys.imm.dtu.dk/ordinalmatrixfactorization.

Second-order Mining for Active Collaborative Filtering

Abstract Active learning for collaborative filtering tasks draws many attentions from the research community. It can capture the user's interest with greatly reduced labeling burden for the online user. High quality recommendation can thus be made with good user experience. In this paper we address the efficiency challenge of current active learning methods for online and interactive applications by using the second-order mining techniques. According to the global latent semantic model learnt from the feedbacks of historical users to items, we propose an intuitive and efficient query strategy for the item selection for new active user. The time complexity in each query is reduced greatly to constant O (1). Experimental results on the public available data sets show the efficiency and effectiveness of our method.

A Survey of Collaborative Filtering Techniques

As one of the most successful approaches to building recommender systems, collaborative filtering (CF) uses the known preferences of a group of users to make recommendations or predictions of the unknown preferences for other users. In this paper, we first introduce CF tasks and their main challenges, such as data sparsity, scalability, synonymy, gray sheep, shilling attacks, privacy protection, etc., and their possible solutions. We then present three main categories of CF techniques: memory-based, model-based, and hybrid CF algorithms (that combine CF with other recommendation techniques), with examples for representative algorithms of each category, and analysis of their predictive performance and their ability to address the challenges. From basic techniques to the state-of-the-art, we attempt to present a comprehensive survey for CF techniques, which can be served as a roadmap for research and practice in this area.

COLLABORATIVE FILTERING FOR MULTI-CLASS DATA USING BAYESIAN NETWORKS

As one of the most successful recommender systems, collaborative filtering (CF) algorithms are required to deal with high sparsity and high requirement of scalability amongst other challenges. Bayesian networks (BNs), one of the most frequently used classifiers, can be used for CF tasks. Previous works on applying BNs to CF tasks were mainly focused on binary-class data, and used simple or basic Bayesian classifiers.1,2 In this work, we apply advanced BNs models to CF tasks instead of simple ones, and work on real-world multi-class CF data instead of synthetic binary-class data. Empirical results show that with their ability to deal with incomplete data, the extended logistic regression on tree augmented naïve Bayes (TAN-ELR)3 CF model consistently performs better than the traditional Pearson correlation-based CF algorithm for the rating data that have few items or high missing rates. In addition, the ELR-optimized BNs CF models are robust in terms of the ability to make predictions, while the robustness of the Pearson correlation-based CF algorithm degrades as the sparseness of the data increases.