Existing Recommendation Methods Research Articles

Exploiting Aesthetic Features in Visual Contents for Movie Recommendation

As one of the most widely used recommender systems, movie recommendation plays an important role in our life. However, the data sparsity problem severely hinders the effectiveness of personalized movie recommendation, which requires more rich content information to be utilized. Posters and still frames, which directly display the visual contents of movies, have significant influences on movie recommendation. They not only reveal rich knowledge for understanding movies but also useful for understanding user preferences. However, existing recommendation methods rarely consider aesthetic features, which tell how the movie looks and feels, extracted from these pictures for the movie recommendation. To this end, in this paper, we propose an aesthetic-aware unified visual content matrix factorization (called UVMF-AES) to integrate visual feature learning and recommendation into a unified framework. Specifically, we first integrate the convolutional neural network (CNN) features and aesthetic features into probabilistic matrix factorization. Then we establish a unified optimization framework with these features for the movie recommendation. The experimental results on two real-world datasets show that our proposed method UVMF-AES is significantly superior to the state-of-the-art methods on movie recommendation.

FRAIPA version 2: A fast recommendation approach based on self-adaptation and multi-thresholding

Finding relevant and interesting items on e-commerce websites is a major challenge in the information era. Recommender systems are widely used to help users deal with the information overload problem by giving them personalized recommendations. Furthermore, they help businesses make more profits. Collaborative filtering represents one of the most successful recommendation approaches. In general, several existing recommendation methods have demonstrated good performance in terms of prediction quality. However, they may require prohibitive computational times, and they often confront the sparsity problem, which negatively affects the efficiency of the system.In this paper, we propose a fast recommendation algorithm based on self-adaptation and multi-thresholding. It is designed to deal with the mentioned drawbacks, and improve the prediction quality. In addition, it is able to converge automatically. Extensive experiments on two real-world data sets demonstrate that our proposed method can achieve significantly better performance than other state-of-the-art methods. Particularly, it improves the MAE between 1.02% and 12.93%, and the computational time between 25.38% and 54.83%.

Data imputation using a trust network for recommendation via matrix factorization

Existing recommendation methods suffer from the data sparsity problem which means that most of users have rated only a very small number of items, often resulting in low recommendation accuracy. In addition, for cold start users evaluating only few items, rating predictions with the methods also produce low accuracy. To address these problems, we propose a novel data imputation method that effectively substitutes missing ratings with probable values (i.e., imputed values). Our method successfully improves accuracy of recommendation methods from the following three aspects: (1) exploiting a trust network, (2) imputing only a part of missing ratings, and (3) applying them to any recommendation methods. Our method employs a bidirectional connection structure within a distance level for finding reliable users in exploiting a trust network as useful information. In addition, our method imputes only some missing ratings, called fillable ratings, whose imputed values are expected to be accurate with a sufficient level of confidence. Moreover, our imputation method is independent of, thus applicable to, any recommendation methods that may include application-specific ones and the most accurate one in each domain. We conduct experiments on three real-life datasets which arise from Epinions and Ciao. Our experimental results demonstrate that our method has recommendation accuracy better than existing recommendation methods equipped with imputation methods or trust networks, especially for cold start users.

A Hierarchical Approach to Enabling Supplier Choice In On-Demand Platforms

In peer-to-peer logistics platforms, capacity cannot be set. Instead, capacity must be enticed from decentralized suppliers (who provide access to their resources). Current peer-to-peer platforms for crowdsourced delivery and ride-sharing do not operationally prioritize suppliers' underutilized resources when making assignments. This paper is interested in evaluating a new hierarchical approach, recasting the platform's role as one providing personalized menus of requests for suppliers to choose from. Supplier choice can increase participation (capacity) and resource utilization. However, coordinating decentralized resources using multiple, simultaneous recommendations is complicated because platform performance is the result of interdependent suppliers' selection outcomes, in which some requests may not be selected and other requests are selected by more than one supplier. Therefore, the objective of this research is to determine when providing suppliers with choices is beneficial to the platform and to quantify this benefit for different environments. A novel bilevel optimization formulation explicitly models the two stage decision process: first, the platform determines which set of requests to recommend to which suppliers, and second, suppliers having a choice to select which request (if any) they would like to serve. By harnessing the problem's structure, the computationally expensive mixed-integer linear bilevel problem is transformed into an equivalent single-level problem that is computationally superior. A computational study based on ride-sharing quantifies the value of providing suppliers with choices. When a platform's knowledge of suppliers' selections is imperfect, our hierarchical approach outperforms existing recommendation methods, namely, centralized, many-to-many stable matching, and decentralized approaches. We show that a platform's lack of knowledge over suppliers' selections can be compensated by providing choices in environments with either inflexible suppliers or when suppliers' utilities have higher variance than the platform's utilities. We find that providing choices and recommending alternatives to more than one supplier can be beneficial to not only the platform, but also freelance suppliers and demand requests.

Personalized question recommendation for English grammar learning

AbstractLearning English grammar is a very challenging task for many students especially for nonnative English speakers. To learn English well, it is important to understand the concepts of the English grammar with lots of practise on exercise questions. Previous recommendation systems for learning English mainly focused on recommending reading materials and vocabulary. Different from reading material and vocabulary recommendations, grammar question recommendation should recommend questions that have similar grammatical structure and usage to the question of interest. The content similarity calculation methods used in existing recommendation methods cannot represent the similarity between grammar questions effectively. In this paper, we propose a content‐based approach for personalized grammar question recommendation, which recommends similar grammatical structure and usage questions for further practising. Specifically, we propose a novel structure named parse‐key tree to capture the grammatical structure and usage of grammar questions. We then propose 3 measures to compute the similarity between the question query and database questions for grammar question recommendation. Additionally, we incorporated the proposed recommendation method into a Web‐based English grammar learning system and presented its performance evaluation in this paper. The experimental results have shown that the proposed approach outperforms other classical and state‐of‐the‐art methods in recommending relevant grammar questions.

Enhancing collaborative filtering via topic model integrated uniform euclidean distance

Recommendation system can greatly alleviate the “information overload” in the big data era. Existing recommendation methods, however, typically focus on predicting missing rating values via analyzing user-item dualistic relationship, which neglect an important fact that the latent interests of users can influence their rating behaviors. Moreover, traditional recommendation methods easily suffer from the high dimensional problem and cold-start problem. To address these challenges, in this paper, we propose a PBUED (PLSA-Based Uniform Euclidean Distance) scheme, which utilizes topic model and uniform Euclidean distance to recommend the suitable items for users. The solution first employs probabilistic latent semantic analysis (PLSA) to extract users' interests, users with different interests are divided into different subgroups. Then, the uniform Euclidean distance is adopted to compute the users' similarity in the same interest subset; finally, the missing rating values of data are predicted via aggregating similar neighbors' ratings. We evaluate PBUED on two datasets and experimental results show PBUED can lead to better predicting performance and ranking performance than other approaches.

Joint user knowledge and matrix factorization for recommender systems

Currently, most of the existing recommendation methods treat social network users equally, which assume that the effect of recommendation on a user is decided by the user’s own preferences and social influence. However, a user’s own knowledge in a field has not been considered. In other words, to what extent does a user accept recommendations in social networks need to consider the user’s own knowledge or expertise in the field. In this paper, we propose a novel matrix factorization recommendation algorithm based on integrating social network information such as trust relationships, rating information of users and users’ own knowledge. Specifically, since we cannot directly measure a user’s knowledge in the field, we first use a user’s status in a social network to indicate a user’s knowledge in a field, and users’ status is inferred from the distributions of users’ ratings and followers across fields or the structure of domain-specific social network. Then, we model the final rating of decision-making as a linear combination of the user’s own preferences, social influence and user’s own knowledge. Experimental results on real world data sets show that our proposed approach generally outperforms the state-of-the-art recommendation algorithms that do not consider the knowledge level difference between the users.

Temporal Factor-Aware Video Affective Analysis and Recommendation for Cyber-Based Social Media

As an important cyber-enabled application, online video recommendation is seeing significant interest from both industry and academia. To effectively recommend video content becomes a popular research topic. However, it has been found that existing recommendation methods based on video affective analysis ignore the temporal factor, leading to poor performance especially when the order of emotion components does affect the recommendation quality. This motivates us to study the feature of emotion fluctuation, which we call Temporal Factor of Emotion (TFE). In this paper, a novel recommendation method based on the Grey Relational Analysis (GRA) is proposed to tackle this problem. GRA preserves the temporal factor of objects during analysis and is suitable for analyzing systems with unknown correlation (a set of independent videos). In our work, first, specific video features are extracted and mapped to the well-known Lovheim emotion-space, through the SVMs (Support Vector Machine). Then, GRA is applied to compute the quantitative relation among videos by using extracted emotions as factors. Finally, a pick-filter pattern and GRA-based recommendation method under the Fisher model are proposed. To evaluate the performance of our method, an online video recommendation system is developed. Experimental results of both user study and parameter evaluation demonstrate that the GRA-based method can improve accuracy of video affective analysis and performance of video recommendation.

Version-Aware Rating Prediction for Mobile App Recommendation

With the great popularity of mobile devices, the amount of mobile apps has grown at a more dramatic rate than ever expected. A technical challenge is how to recommend suitable apps to mobile users. In this work, we identify and focus on a unique characteristic that exists in mobile app recommendation—that is, an app usually corresponds to multiple release versions. Based on this characteristic, we propose a fine-grain version-aware app recommendation problem. Instead of directly learning the users’ preferences over the apps, we aim to infer the ratings of users on a specific version of an app. However, the user-version rating matrix will be sparser than the corresponding user-app rating matrix, making existing recommendation methods less effective. In view of this, our approach has made two major extensions. First, we leverage the review text that is associated with each rating record; more importantly, we consider two types of version-based correlations. The first type is to capture the temporal correlations between multiple versions within the same app, and the second type of correlation is to capture the aggregation correlations between similar apps. Experimental results on a large dataset demonstrate the superiority of our approach over several competitive methods.

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.

The essential role of time in network-based recommendation

Random walks on bipartite networks have been used extensively to design personalized recommendation methods. While aging has been identified as a key component in the growth of information networks, most research has focused on the networks' structural properties and neglected the often available time information. Time has been largely ignored both by the investigated recommendation methods as well as by the methodology used to evaluate them. We show that this time-unaware approach overestimates the methods' recommendation performance. Motivated by microscopic rules of network growth, we propose a time-aware modification of an existing recommendation method and show that by combining the temporal and structural aspects, it outperforms the existing methods. The performance improvements are particularly striking in systems with fast aging.

A topic community‐based method for friend recommendation in large‐scale online social networks

SummaryOnline social networks (OSNs) have become more and more popular and have attracted a great many users. Friend recommendation, which is one of the important services in OSN, can help users discover their interested friends and alleviate the problem of information overload. However, most of existing recommendation methods only consider either user link or content information and hence are not effective enough to provide high quality recommendations. In this paper, we propose a topic community‐based method via Nonnegative Matrix Factorization (NMF). This method first applies joint NMF model to mine topic communities existing in OSN by combing link and content information. Then it computes user pairwise similarities and makes friends recommendation based on topic communities. Furthermore, this method can be implemented using the MapReduce distributed computing framework. Extensive experiments show that our proposed method not only has better recommendation performance than state‐of‐the‐art methods but also has good scalability to deal with the problem of friend recommendation in large‐sale OSNs. Moreover, the application case demonstrates that it can significantly improve friend recommendation service in the real world OSN. Copyright © 2016 John Wiley & Sons, Ltd.

Recurrent neural network based recommendation for time heterogeneous feedback

In recommender systems, several kinds of user feedback with time stamps are collected and used for recommendation, which is called the time heterogeneous feedback recommendation problem. The existing recommendation methods can handle only one kind of feedback or ignore the time stamps of the feedback. To solve the time heterogeneous feedback recommendation problem, in this paper, we propose a recurrent neural network model to predict the probability that the user will access an item given the time heterogeneous feedback of this user. To our best knowledge, it is the first time to solve the time heterogeneous feedback recommendation problem by deep neural network model. The proposed model is learned by back propagation algorithm and back propagation through time algorithm. The comparison results on four real-life datasets indicate that the proposed method outperforms the compared state-of-the-art approaches.

The effect of heterogeneous dynamics of online users on information filtering

The rapid expansion of the Internet requires effective information filtering techniques to extract the most essential and relevant information for online users. Many recommendation algorithms have been proposed to predict the future items that a given user might be interested in. However, there is an important issue that has always been ignored so far in related works, namely the heterogeneous dynamics of online users. The interest of active users changes more often than that of less active users, which asks for different update frequency of their recommendation lists. In this paper, we develop a framework to study the effect of heterogeneous dynamics of users on the recommendation performance. We find that the personalized application of recommendation algorithms results in remarkable improvement in the recommendation accuracy and diversity. Our findings may help online retailers make better use of the existing recommendation methods.

Two-level matrix factorization for recommender systems

Many existing recommendation methods such as matrix factorization (MF) mainly rely on user---item rating matrix, which sometimes is not informative enough, often suffering from the cold-start problem. To solve this challenge, complementary textual relations between items are incorporated into recommender systems (RS) in this paper. Specifically, we first apply a novel weighted textual matrix factorization (WTMF) approach to compute the semantic similarities between items, then integrate the inferred item semantic relations into MF and propose a two-level matrix factorization (TLMF) model for RS. Experimental results on two open data sets not only demonstrate the superiority of TLMF model over benchmark methods, but also show the effectiveness of TLMF for solving the cold-start problem.

A Fuzzy Preference Tree-Based Recommender System for Personalized Business-to-Business E-Services

The Web creates excellent opportunities for businesses to provide personalized online services to their customers. Recommender systems aim to automatically generate personalized suggestions of products/services to customers (businesses or individuals). Although recommender systems have been well studied, there are still two challenges in the development of a recommender system, particularly in real-world B2B e-services: (1) items or user profiles often present complicated tree structures in business applications, which cannot be handled by normal item similarity measures and (2) online users' preferences are often vague and fuzzy, and cannot be dealt with by existing recommendation methods. To handle both these challenges, this study first proposes a method for modeling fuzzy tree-structured user preferences, in which fuzzy set techniques are used to express user preferences. A recommendation approach to recommending tree-structured items is then developed. The key technique in this study is a comprehensive tree matching method, which can match two tree-structured data and identify their corresponding parts by considering all the information on tree structures, node attributes, and weights. Importantly, the proposed fuzzy preference tree-based recommendation approach is tested and validated using an Australian business dataset and the MovieLens dataset. Experimental results show that the proposed fuzzy tree-structured user preference profile reflects user preferences effectively and the recommendation approach demonstrates excellent performance for tree-structured items, especially in e-business applications. This study also applies the proposed recommendation approach to the development of a Web-based business partner recommender system.

Motif-Plus: incorporation of network motifs into top-n friendship recommendations

Although recently there have been a variety of attempts for friendship recommendations, most existing approaches overemphasise the use of global information implied in historical data while overlooking such essential transitivity rule in social networks as the 'triadic closure' principle. To overcome this limitation, we analyse triangular network motifs in social networks and design a novel principle named 'triangular power' for friendship recommendations on the basis of the 'triadic closure' rule. With this principle, we propose to quantify the willing-power that drives two people become friends by making use of triangular motifs and propose a novel strategy called Motif-Plus that utilise local information of a social network for friendship recommendations. We further incorporate this strategy into existing recommendation methods by aggregating ranks generated by individual approaches and Motif-Plus. With large-scale random sub-sampling validation experiments on three real social network datasets (Facebook, Twitter and Delicious), we compare the performance of five existing methods with Motif-Plus incorporated to their original forms. Results show that methods with Motif-Plus outperform their original forms in not only accuracy and retrieval criteria but also recommendation diversity significantly, suggesting that friendship recommendations will be benefit from the fusion of global and local information of social networks.

Virtual goods recommendations in virtual worlds.

Virtual worlds (VWs) are computer-simulated environments which allow users to create their own virtual character as an avatar. With the rapidly growing user volume in VWs, platform providers launch virtual goods in haste and stampede users to increase sales revenue. However, the rapidity of development incurs virtual unrelated items which will be difficult to remarket. It not only wastes virtual global companies' intelligence resources, but also makes it difficult for users to find suitable virtual goods fit for their virtual home in daily virtual life. In the VWs, users decorate their houses, visit others' homes, create families, host parties, and so forth. Users establish their social life circles through these activities. This research proposes a novel virtual goods recommendation method based on these social interactions. The contact strength and contact influence result from interactions with social neighbors and influence users' buying intention. Our research highlights the importance of social interactions in virtual goods recommendation. The experiment's data were retrieved from an online VW platform, and the results show that the proposed method, considering social interactions and social life circle, has better performance than existing recommendation methods.

How to Recommend by Online Lifestyle Tagging (OLT)

With the rapid development of the Internet, the online shopping market expands constantly. Inspired by fierce competition and complex and diverse consumer demand, personalized recommendation has become an effective marketing tool for e-commerce enterprises. However, the existing recommendation methods based on online consumer behavior or preferences are characterized by poor accuracy and low efficiency. The paper mainly conducts three studies, the study1 proves that seven online lifestyles, which are "Comfortable, Entertainment, Luxury, Tradition & Conservation, Rational, Fashion Sense, and Social Activities", affect Chinese consumers' purchase. However, the different online lifestyles have different effects on purchase, thus the response rates of recommending. The study2 proposes a new personalized recommendation method "online lifestyle tagging (OLT)" based on online lifestyle and user behavior tags to identify online lifestyles. In the study3, the efficiency of OLT is tested and verified using data collected from enterprises, it suggests that OLT has a significantly higher response rate than traditional behavior-based methods. This study demonstrates that OLT improves the accuracy and efficiency of personalized recommendation, and thus contributes to the theory of personalized recommendation and marketing methods based on lifestyle.

Lazy Collaborative Filtering for Data Sets With Missing Values

As one of the biggest challenges in research on recommender systems, the data sparsity issue is mainly caused by the fact that users tend to rate a small proportion of items from the huge number of available items. This issue becomes even more problematic for the neighborhood-based collaborative filtering (CF) methods, as there are even lower numbers of ratings available in the neighborhood of the query item. In this paper, we aim to address the data sparsity issue in the context of neighborhood-based CF. For a given query (user, item), a set of key ratings is first identified by taking the historical information of both the user and the item into account. Then, an auto-adaptive imputation (AutAI) method is proposed to impute the missing values in the set of key ratings. We present a theoretical analysis to show that the proposed imputation method effectively improves the performance of the conventional neighborhood-based CF methods. The experimental results show that our new method of CF with AutAI outperforms six existing recommendation methods in terms of accuracy.