Improve Recommendation Accuracy Research Articles

Recommendation in a Changing World

Users’ preferences, and consequently their ratings and reviews to items, change over time. Likewise, characteristics of items are also time-varying. By dividing data into time periods, temporal Recommender Systems (RSs) improve recommendation accuracy by exploring the temporal dynamics in user rating data. However, temporal RSs have to cope with rating sparsity in each time period. Meanwhile, reviews generated by users contain rich information about their preferences, which can be exploited to address rating sparsity and further improve the performance of temporal RSs. In this article, we develop a temporal rating model with topics that jointly mines the temporal dynamics of both user-item ratings and reviews. Studying temporal drifts in reviews helps us understand item rating evolutions and user interest changes over time. Our model also automatically splits the review text in each time period into interim words and intrinsic words. By linking interim words and intrinsic words to short-term and long-term item features, respectively, we jointly mine the temporal changes in user and item latent features together with the associated review text in a single learning stage. Through experiments on 28 real-world datasets collected from Amazon , we show that the rating prediction accuracy of our model significantly outperforms the existing state-of-art RS models. And our model can automatically identify representative interim words in each time period as well as intrinsic words across all time periods. This can be very useful in understanding the time evolution of users’ preferences and items’ characteristics.

A novel collective matrix factorization model for recommendation with fine‐grained social trust prediction

SummaryRecommender systems are playing an increasing role in improving user satisfaction as they can recommend items which might be highly interested to users. Recent advances have proven that social relations such as trust and distrust relations among users are helpful in improving recommendation accuracy. Traditional social recommendation methods directly utilize unweighted trust and distrust relations into collaborative filtering framework. These methods will lose their power when the trust or distrust relation data is sparse, which significantly hinders the improvement of rating prediction accuracy. To address this problem, we transform the unweighted trust and distrust relations into fine‐grained weighted social trust matrix which is denser and encodes the trust and distrust degree for pair of users. The weighted social trust matrix is then combined with the rating matrix in a collective matrix factorization framework to implement rating prediction task. Experimental results based on Extended Epinions dataset show that the proposed collective matrix factorization model with fine‐grained weighted social trust matrix can achieve better accuracy than conventional social recommendation algorithms such as SoRec and its extensions.

Leveraging tagging and rating for recommendation: RMF meets weighted diffusion on tripartite graphs

Recommender systems (RSs) have been a widely exploited approach to solving the information overload problem. However, the performance is still limited due to the extreme sparsity of the rating data. With the popularity of Web 2.0, the social tagging system provides more external information to improve recommendation accuracy. Although some existing approaches combine the matrix factorization models with the tag co-occurrence and context of tags, they neglect the issue of tag sparsity that would also result in inaccurate recommendations. Consequently, in this paper, we propose a novel hybrid collaborative filtering model named WUDiff_RMF, which improves regularized matrix factorization (RMF) model by integrating Weighted User-Diffusion-based CF algorithm(WUDiff) that obtains the information of similar users from the weighted tripartite user–item–tag graph. This model aims to capture the degree correlation of the user–item–tag tripartite network to enhance the performance of recommendation. Experiments conducted on four real-world datasets demonstrate that our approach significantly performs better than already widely used methods in the accuracy of recommendation. Moreover, results show that WUDiff_RMF can alleviate the data sparsity, especially in the circumstance that users have made few ratings and few tags.

TruCom: Exploiting Domain-Specific Trust Networks for Multicategory Item Recommendation

Recommender systems (RSs) have become important tools for solving the problem of information overload. With the advent and popularity of online social networks, some studies on network-based recommendation have emerged, raising the concern of many researchers. Trust is one kind of important information available in social networks and is often used for performance improvement in social-network-based RSs. However, most trust-aware RSs ignore the fact that people trust different subsets of friends pertaining to different domains, such as music and movies, because people behave differently in diverse domains according to different interests. This paper proposes a novel recommendation method called TruCom. In a multicategory item recommendation domain, TruCom first generates a domain-specific trust network pertaining to each domain and then builds a unified objective function for improving recommendation accuracy by incorporating the hybrid information of direct and indirect trust into a matrix factorization recommendation model. Through relevant benchmark experiments on two real-world data sets, we show that TruCom achieves better performance than other existing recommendation methods, which demonstrates the effectiveness and reliability of TruCom.

Recommend products with consideration of multi-category inter-purchase time and price

This study focuses on diversifying recommendations and improving recommendation accuracy by incorporating multi-category purchase interval and price in a novel way. We suggest a method to model the drift of a user’s interest for different categories based on sequential pattern mining. Additionally, we propose a new approach to model personalized multi-category inter-purchase interval for the user. By employing fuzzy set theory, we also put forward an approach to model the price preferences for each user. We propose a recommender system that incorporates the methods mentioned above. The experimental results based on real purchase records show that the proposed recommendation algorithm has high stability and superior performance in recommending products with different characteristics. The results also demonstrate the effectiveness of incorporating purchase interval and price factor in recommender systems. This study demonstrates the existence of particular inter-purchase intervals among different categories, and indicates that price is an important factor influencing customers’ decision making.

GLOMA: Embedding Global Information in Local Matrix Approximation Models for Collaborative Filtering

Recommender systems have achieved great success in recent years, and matrix approximation (MA) is one of the most popular techniques for collaborative filtering (CF) based recommendation. However, a major issue is that MA methods perform poorly at detecting strong localized associations among closely related users and items. Recently, some MA-based CF methods adopt clustering methods to discover meaningful user-item subgroups and perform ensemble on different clusterings to improve the recommendation accuracy. However, ensemble learning suffers from lower efficiency due to the increased overall computation overhead. In this paper, we propose GLOMA, a new clustering-based matrix approximation method, which can embed global information in local matrix approximation models to improve recommendation accuracy. In GLOMA, a MA model is first trained on the entire data to capture global information. The global MA model is then utilized to guide the training of cluster-based local MA models, such that the local models can detect strong localized associations shared within clusters and at the same time preserve global associations shared among all users/items. Evaluation results using MovieLens and Netflix datasets demonstrate that, by integrating global information in local models, GLOMA can outperform five state-of-the-art MA-based CF methods in recommendation accuracy while achieving descent efficiency.

Exploiting both Vertical and Horizontal Dimensions of Feature Hierarchy for Effective Recommendation

Feature hierarchy (FH) has proven to be effective to improve recommendation accuracy. Prior work mainly focuses on the influence of vertically affiliated features (i.e. child-parent) on user-item interactions. The relationships of horizontally organized features (i.e. siblings and cousins) in the hierarchy, however, has only been little investigated. We show in real-world datasets that feature relationships in horizontal dimension can help explain and further model user-item interactions. To fully exploit FH, we propose a unified recommendation framework that seamlessly incorporates both vertical and horizontal dimensions for effective recommendation. Our model further considers two types of semantically rich feature relationships in horizontal dimension, i.e. complementary and alternative relationships. Extensive validation on four real-world datasets demonstrates the superiority of our approach against the state of the art. An additional benefit of our model is to provide better interpretations of the generated recommendations.

Improving Recommendation Accuracy and Diversity through Cost-Awareness Probabilistic Spreading

Improving Recommendation Accuracy and Diversity through Cost-Awareness Probabilistic Spreading

Exploiting long‐term and short‐term preferences and RFID trajectories in shop recommendation

SummaryShop recommendation in large shopping malls is useful in the mobile internet era. With the maturity of indoor positioning technology, customers' indoor trajectories can be captured by radio frequency identification devices readers, which provides a new way to analyze customers' potential preferences. In this paper, we design three methods for the top‐N shop recommendation problem. The first method is an improved matrix factorization method fusing estimated prior customer preference matrix that is constructed by Session‐based Temporal Graph computing. The second method is a Bayesian personalized ranking method based on the first method. The third method is by tensor decomposition combined with Session‐based Temporal Graph. Besides, we exploit customer history radio frequency identification devices trajectory information to find customers' frequent paths and revise predicted rating values to improve recommendation accuracy. Our methods are effective in modeling customers' temporal dynamics. At the same time, our approach considers repeated recommendation of the same shop by designing rating update rules. The test dataset is formed by JoyCity customer behavior records. JoyCity is a large‐scale modern shopping center in downtown Shanghai, China. The results show that our approaches are effective and outperform previous state‐of‐the‐art approaches. Copyright © 2016 John Wiley & Sons, Ltd.

SalesExplorer: Exploring sales opportunities from white-space customers in the enterprise market

In commercial sales and services, recommender systems have been widely adopted to predict customers’ purchase interests using their prior purchasing behaviors. Cold-start is a known challenge to existing recommendation techniques, e.g., the popular collaborative filtering method is not applicable to predict the interests of “white-space” customers since they have no prior purchasing history in the targeted product categories.This paper presents SalesExplorer, a new recommendation algorithm to address “white-space” customer issue in the commercial sales and services segment. To predict the interests of customers who are new to a product category, we propose a statistical inference method using customers’ existing purchase records from other product categories, a Probabilistic Latent Semantic Analysis (PLSA)-based transfer learning method using customers’ business profile content, and a kernel logistic regression-based model which combines these two recommendations to produce the final results with higher accuracy. Experimental study using real-world enterprise sales data demonstrates that, comparing with a baseline and two state-of-the-art methods, the proposed combinatorial algorithm improves recommendation accuracy by 32.14%, 13.13% and 9.85%, respectively.

STD: An Improved Social Recommendation Model with Temporal Dynamics of Social Relationships

In online social networks, social recommendation incorporates social relationships to solve data sparsity problem. However, existing studies ignore temporal dynamics of social relationships. To improve recommendation accuracy, this paper proposes an improved social recommendation model, which takes temporal dynamics of social relationships into consideration. It infers users’ social influence from their interaction records, and weights social influence differently according to the time distance. It also employs matrix factorization technique to fuse temporal dynamics of social relationships and user preference features together, demonstrating time-dependent change of user preference. An empirical analysis on Epinions dataset demonstrates that our approach performs better on improving predicted accuracy compared with current social recommendation models.

Tourism Destination Recommender System for the Cold Start Problem

With the advent and popularity of e-commerce, an increasing number of consumers prefer to order tourism products online. A recommender system can help these users contend with information overload; however, such a system is affected by the cold start problem. Online tourism destination searching is a more difficult task than others on account of its more restrictive factors. In this paper, we therefore propose a tourism destination recommender system that employs opinion-mining technology to refine user preferences and item opinion reputations. These elements are then fused into a hybrid collaborative filtering method by combining user- and item-based collaborative filtering approaches. Meanwhile, we embed an artificial interactive module in our recommender system to alleviate the cold start problem. Compared with several well-known cold start recommendation approaches, our method provides improved recommendation accuracy and quality. A series of experimental evaluations using a publicly available dataset demonstrate that the proposed recommender system outperforms existing recommender systems in addressing the cold start problem.

Mining unstructured content for recommender systems: an ensemble approach

Recommendation of textual documents requires indexing mechanisms to extract structured metadata for attribute-aware recommender systems. Applying a variety of text mining algorithms has the advantage of capturing different aspects of unstructured content, resulting in richer descriptions. However, it is difficult to integrate them into a unique model so that these descriptions can efficiently improve recommendation accuracy. This article proposes a generic model based on ensemble learning that combines simple text mining methods in a post-processing approach. After executing each text mining technique, each set of metadata of a particular type is applied to the recommender module, which generates attribute-specific rankings. Then, the resulting recommendations are ensembled to generate a final personalized ranking to the user. We evaluated our ensemble technique with two attribute-aware collaborative recommenders (k-Nearest Neighbors and BPR-Mapping) and we demonstrate its generality by means of comparisons among different types of ensembles. We used two datasets from different domains, the first is from the Brazilian Embrapa Agency of Technology Information website, whose documents are written in Portuguese language, and the second is the HetRec MovieLens 2k, published by the GroupLens Research Group, whose movies' storylines are written in English. The experiments show that, particularly to the k-NN recommender, better accuracy can be obtained when multiple metadata types are combined. The proposed approach is extensible and flexible to new indexing and recommendation techniques.

Jointly modeling content, social network and ratings for explainable and cold-start recommendation

Model-based approach to collaborative filtering (CF), such as latent factor models, has improved both accuracy and efficiency of predictions on large and sparse dataset. However, most existing methods still face two major problems: (1) the recommendation results derived from user and item vectors of a set of unobserved factors are lack of explanation; (2) cold start users and items out of user-item rating matrix cannot be handled accurately. In this paper, we propose a hybrid method for addressing the problems by incorporating content-based information (i.e, users׳ tags and items׳ keywords) and social information. The main idea behind our method is to build content association based on three factors-user interest in selected tags, tag–keyword relation and item correlation with extracted keywords, and then recommend the items with high similarity in content to users. Two novel methods-neighbor based approach and 3-factor matrix factorization are proposed for building tag–keyword relation matrix and learning user interest vector for selected tags and item correlation vector for extracted keywords. Besides, we introduce a social regularization term to help shape user interest vector. Analysis shows that our method can generate explainable recommendation results with simple descriptions, and experiments on real dataset demonstrate that our method improves recommendation accuracy of state-of-the-art CF models for previous users and items with few ratings, as well as cold start users and items with no rating.

Index partitioning through a bipartite graph model for faster similarity search in recommendation systems

Scalability of a recommendation system is an important factor for large e-commerce sites containing millions of products visited by millions of users. Similarity search is the core operation in recommendation systems. In this paper, we explain a framework to alleviate performance bottleneck of similarity search for very large-scale recommendation systems. The framework employs inverted index for real-time similarity search and handles dynamic updates. As the inverted index gets larger, retrieving recommendations become computationally expensive. There are various works devoted to solve this problem, such as clustering and preprocessing to compute recommendations offline. Our solution is based on bipartite graph partitioning for minimizing the affinity between entities in different partitions. Number of operations in similarity search is reduced by executing search within the closest partitions to the query. Parts are balanced, so that computational loads of partitions are almost the same, which is significant for reducing the computational cost. Sequential experiments with several different recommendation approaches and large datasets consisting of millions of users and items validate the scalability of the proposed recommendation framework. Accuracy drops only by a small factor due to partitioning, if any. Even slight improvements in recommendation accuracy are observed in our collaborative filtering experiments.

A New Recommender System Using Context Clustering Based on Matrix Factorization Techniques

Recommender system can efficiently alleviate the information overload problem, but it has been trapped in the recommendation accuracy. We proposed a new recommender system which based on matrix factorization techniques. More factors including contextual information, user ratings and item feature are all taken into consideration. Meanwhile the k-modes algorithm is used to reduce the complexity of matrix operations and increase the relevance of the user-item ratings sub-matrix. Compared with several major existing recommendation approaches, extensive experimental evaluation on publicly available dataset demonstrates that our method enjoys improved recommendation accuracy.

Integrating heterogeneous information via flexible regularization framework for recommendation

Recently, there is a surge of social recommendation, which leverages social relations among users to improve recommendation performance. However, in many applications, social relations are absent or very sparse. Meanwhile, the attribute information of users or items may be rich. It is a big challenge to exploit these attribute information for the improvement of recommendation performance. In this paper, we organize objects and relations in recommendation system as a heterogeneous information network, and introduce meta path based similarity measure to evaluate the similarity of users or items. Furthermore, a matrix factorization based dual regularization framework SimMF is proposed to flexibly integrate different types of information through adopting the similarity of users and items as regularization on latent factors of users and items. Extensive experiments not only validate the effectiveness of SimMF but also reveal some interesting findings. We find that attribute information of users and items can significantly improve recommendation accuracy, and their contribution seems more important than that of social relations. The experiments also reveal that different regularization models have obviously different impact on users and items.

AUI&GIV: Recommendation with Asymmetric User Influence and Global Importance Value.

The user-based collaborative filtering (CF) algorithm is one of the most popular approaches for making recommendation. Despite its success, the traditional user-based CF algorithm suffers one serious problem that it only measures the influence between two users based on their symmetric similarities calculated by their consumption histories. It means that, for a pair of users, the influences on each other are the same, which however may not be true. Intuitively, an expert may have an impact on a novice user but a novice user may not affect an expert at all. Besides, each user may possess a global importance factor that affects his/her influence to the remaining users. To this end, in this paper, we propose an asymmetric user influence model to measure the directed influence between two users and adopt the PageRank algorithm to calculate the global importance value of each user. And then the directed influence values and the global importance values are integrated to deduce the final influence values between two users. Finally, we use the final influence values to improve the performance of the traditional user-based CF algorithm. Extensive experiments have been conducted, the results of which have confirmed that both the asymmetric user influence model and global importance value play key roles in improving recommendation accuracy, and hence the proposed method significantly outperforms the existing recommendation algorithms, in particular the user-based CF algorithm on the datasets of high rating density.

Personalized Travel Package With Multi-Point-of-Interest Recommendation Based on Crowdsourced User Footprints

Location-based social networks (LBSNs) provide people with an interface to share their locations and write reviews about interesting places of attraction. The shared locations form the crowdsourced digital footprints, in which each user has many connections to many locations, indicating user preference to locations. In this paper, we propose an approach for personalized travel package recommendation to help users make travel plans. The approach utilizes data collected from LBSNs to model users and locations, and it determines users’ preferred destinations using collaborative filtering approaches. Recommendations are generated by jointly considering user preference and spatiotemporal constraints. A heuristic search-based travel route planning algorithm was designed to generate travel packages. We developed a prototype system, which obtains users’ travel demands from mobile client and generates travel packages containing multiple points of interest and their visiting sequence. Experimental results suggest that the proposed approach shows promise with respect to improving recommendation accuracy and diversity.

A fuzzy model for managing natural noise in recommender systems

E-commerce customers demand quick and easy access to products in large search spaces according to their needs and preferences. To support and facilitate this process, recommender systems (RS) based on user preferences have recently played a key role. However the elicitation of customers preferences is not always precise either correct, because of external factors such as human errors, uncertainty and vagueness proper of human beings and so on. Such a problem in RS is known as natural noise and can bias customers recommendations. Despite different proposals have been presented to deal with natural noise in RS none of them is able to manage properly the inherent uncertainty and vagueness of customers preferences. Hence, this paper is devoted to a new fuzzy method for managing in a flexible and adaptable way such uncertainty of natural noise in order to improve recommendation accuracy. Eventually a case study is performed to show the improvements produced by this fuzzy method regarding previous proposals.