Existing Recommendation Algorithms Research Articles

E-Commerce Recommendation Technology Based on Collaborative Filtering Algorithm and Mobile Cloud Computing

Personalized recommendation technology, as one of the core technologies of an E-commerce platform, has attracted a lot of attention with the rise of E-commerce in the Internet industry. Mobile cloud computing-based E-commerce has also exploded in popularity. You can buy whatever you want without leaving the house. Consumers are becoming increasingly receptive to online shopping as a result of this convenience; the E-commerce model demonstrates great modern business value. With its convenient and quick characteristics, online shopping has become fashionable and trendy; however, the popularity of the Internet and the rapid development of E-commerce has resulted in information overload, making it difficult for users to find the goods they require among a vast amount of product information. As a result, the E-commerce recommendation system was born. However, there is currently very little in-depth research on personalized recommendation technology in the field of o2o E-commerce, and most existing recommendation algorithms need to be improved in terms of accuracy and recommendation efficiency.

NCGAN:A neural adversarial collaborative filtering for recommender system

The major challenge of recommendation system (RS) based on implict feedback is to accurately model users’ preferences from their historical feedback. Nowadays, researchers has tried to apply adversarial technique in RS, which had presented successful results in various domains. To a certain extent, the use of adversarial technique improves the modeling of users’ preferences. Nonetheless, there are still many problems to be solved, such as insufficient representation and low-level interaction. In this paper, we propose a recommendation algorithm NCGAN which combines neural collaborative filtering and generative adversarial network (GAN). We use the neural networks to extract users’ non-linear characteristics. At the same time, we integrate the GAN framework to guide the recommendation model training. Among them, the generator aims to make user recommendations and the discriminator is equivalent to a measurement tool which could measure the distance between the generated distribution and users’ ground distribution. Through comparison with other existing recommendation algorithms, our algorithm show better experimental performance in all indicators.

Service recommendation through graph attention network in heterogeneous information networks

Recommending suitable services to users autonomously has become the key to solve the problem of service information overload. Existing recommendation algorithms have some limitations, and discard the side information of the node, or ignore the information of the intermediate node, or omit the feature information of the neighbouring nodes, or not model the pairwise attentive interactions between users and services. To solve the above-mentioned limitations, this paper proposes a service recommendation approach by leveraging the graph attention network (GAT) and co-attention mechanism in heterogeneous information networks (HINs). Specifically, different types of meta-paths are first constructed, and a feature expression is learned for each node in HINs. Then, the feature information of mashups/services is aggregated by the co-attention mechanism. Finally, the multi-layer perceptron (MLP) is applied to recommend suitable services for users. Experiments on a real-world dataset illustrate that the proposed method outperforms other state-of-the-art comparison methods.

Multi-Objective Optimization-Based Recommendation for Massive Online Learning Resources

As an important research topic in intelligent teaching systems, personalized recommendation services of learning resources can effectively solve the “information overload” problem and provide effective learning. However, the traditional learning resource recommendation technology mainly aims to improve recommendation accuracy and cannot effectively ensure the diversity and novelty of recommendation results. In this paper, the learning resource recommendation task is modelled with a multi-objective optimization problem. This paper proposes the Multi-Objective Evolutionary Algorithm-based online learning Resource Recommendation Model to balance the system’s accuracy, novelty, and diversity. The proposed model includes the following four steps: learning clustering, optimization goal setting individual representation, and genetic operator. According to the experimental results, this algorithm can improve the recommendation performance of online learning resources. Compared with the existing recommendation algorithms, more accurate, diverse, and novel learning resource recommendation results can be obtained with the proposed algorithm.

A Smart Tourism Recommendation Algorithm Based on Cellular Geospatial Clustering and Multivariate Weighted Collaborative Filtering

Tourist attraction and tour route recommendation are the key research highlights in the field of smart tourism. Currently, the existing recommendation algorithms encounter certain problems when making decisions regarding tourist attractions and tour routes. This paper presents a smart tourism recommendation algorithm based on a cellular geospatial clustering and weighted collaborative filtering. The problems are analyzed and concluded, and then the research ideas and methods to solve the problems are introduced. Aimed at solving the problems, the tourist attraction recommendation model is set up based on a cellular geographic space generating model and a weighted collaborative filtering model. According to the matching degree between the tourists’ interest needs and tourist attraction feature attributes, a precise tourist attraction recommendation is obtained. In combination with the geospatial attributes of the tourist destination, the spatial adjacency clustering model based on the cellular space generating algorithm is set up, and then the weighted model is introduced for the collaborative filtering recommendation algorithm, which ensures that the recommendation result precisely matches the tourists’ needs. Providing precise results, the optimal tour route recommendation model based on the precise tourist attraction approach vector algorithm is set up. The approach vector algorithm is used to search the optimal route between two POIs under the condition of multivariate traffic modes to provide the tourists with the best motive benefits. To verify the feasibility and advantages of the algorithm, this paper designs a sample experiment and analyzes the resulting data to obtain the relevant conclusion.

Modeling Dynamic User Preference via Dictionary Learning for Sequential Recommendation

Capturing the dynamics in user preference is crucial to better predict user future behaviors because user preferences often drift over time. Many existing recommendation algorithms -- including both shallow and deep ones -- often model such dynamics independently, i.e., user static and dynamic preferences are not modeled under the same latent space, which makes it difficult to fuse them for recommendation. This paper considers the problem of embedding a user's sequential behavior into the latent space of user preferences, namely translating sequence to preference. To this end, we formulate the sequential recommendation task as a dictionary learning problem, which learns: 1) a shared dictionary matrix, each row of which represents a partial signal of user dynamic preferences shared across users; and 2) a posterior distribution estimator using a deep autoregressive model integrated with Gated Recurrent Unit (GRU), which can select related rows of the dictionary to represent a user's dynamic preferences conditioned on his/her past behaviors. Qualitative studies on the Netflix dataset demonstrate that the proposed method can capture the user preference drifts over time and quantitative studies on multiple real-world datasets demonstrate that the proposed method can achieve higher accuracy compared with state-of-the-art factorization and neural sequential recommendation methods. The code is available at https://github.com/cchao0116/S2PNM-TKDE2021.

A movie recommendation model combining time information and probability matrix factorisation

A deep analysis and discussion of matrix factorisation technologies are given in this paper taking into account the defects of traditional collaborative filtering recommendation algorithms. In addition, we provide an analysis of the effects of feature vector dimensions on the recommendation quality and efficiency of a probability matrix factorisation (PMF) algorithm. A PMF algorithm will lead to inaccurate recommendations if it does not consider possible dynamic changes in a user's interest over time. Accordingly, a TPMF model, a PMF algorithm integrated with time information, is proposed in this article. Its feasibility and effectiveness are empirically verified using movie recommendation datasets, and higher prediction accuracy is confirmed compared to existing recommendation algorithms.

Attention-Enhanced Graph Neural Networks for Session-Based Recommendation

Session-based recommendation, which aims to match user needs with rich resources based on anonymous sessions, nowadays plays a critical role in various online platforms (e.g., media streaming sites, search and e-commerce). Existing recommendation algorithms usually model a session as a sequence or a session graph to model transitions between items. Despite their effectiveness, we would argue that the performance of these methods is still flawed: (1) Using only fixed session item embedding without considering the diversity of users’ interests and target items. (2) For user’s long-term interest, the difficulty of capturing the different priorities for different items accurately. To tackle these defects, we propose a novel model which leverages both the target attentive network and self-attention network to improve the graph-neural-network (GNN)-based recommender. In our model, we first model user’s interaction sequences as session graphs which serves as the input of the GNN, and each node vector involved in session graph can be obtained via the GNN. Next, target attentive network can activates different user interests corresponding to varied target items (i.e., the session embedding learned varies with different target items), which can reveal the relevance between users’ interests and target items. At last, after applying the self-attention mechanism, the different priorities for different items can be captured to improve the precision of the long-term session representation. By using a hybrid of long-term and short-term session representation, we can capture users’ comprehensive interests at multiple levels. Extensive experiments demonstrate the effectiveness of our algorithm on two real-world datasets for session-based recommendation.

Research on User Interest Expression and Recommendation Service based on Three-dimensional Relationship of Users and Items

The existing recommendation algorithms often rely heavily on the original score information in the user rating matrix. However, the user's rating of items does not fully reflect the user's real interest. Therefore, the key to improve the existing recommendation system algorithm effectively is to eliminate the influence of these unfavorable factors and the accuracy of the recommendation algorithm can be improved by correcting the original user rating information reasonably. This paper makes a comprehensive theoretical analysis and method design from three aspects: the quality of the item, the memory function of the user and the influence of the social friends trusted by the user on the user's rating. Based on these methods, this paper finally proposes a collaborative filtering recommendation algorithm (FixCF) based on user rating modification. Using data sets such as Movielens, Epinions and Flixster, the data sets are divided into five representative subsets, and the experimental demonstration is carried out. FixCF and classical collaborative filtering algorithms, existing matrix decomposition-based algorithms and trust network-based inference are compared. The experimental results show that the accuracy and coverage of FixCF have been improved under many experimental conditions.

Cross-Domain Recommendation Based on Sentiment Analysis and Latent Feature Mapping

Cross-domain recommendation is a promising solution in recommendation systems by using relatively rich information from the source domain to improve the recommendation accuracy of the target domain. Most of the existing methods consider the rating information of users in different domains, the label information of users and items and the review information of users on items. However, they do not effectively use the latent sentiment information to find the accurate mapping of latent features in reviews between domains. User reviews usually include user’s subjective views, which can reflect the user’s preferences and sentiment tendencies to various attributes of the items. Therefore, in order to solve the cold-start problem in the recommendation process, this paper proposes a cross-domain recommendation algorithm (CDR-SAFM) based on sentiment analysis and latent feature mapping by combining the sentiment information implicit in user reviews in different domains. Different from previous sentiment research, this paper divides sentiment into three categories based on three-way decision ideas—namely, positive, negative and neutral—by conducting sentiment analysis on user review information. Furthermore, the Latent Dirichlet Allocation (LDA) is used to model the user’s semantic orientation to generate the latent sentiment review features. Moreover, the Multilayer Perceptron (MLP) is used to obtain the cross domain non-linear mapping function to transfer the user’s sentiment review features. Finally, this paper proves the effectiveness of the proposed CDR-SAFM framework by comparing it with existing recommendation algorithms in a cross-domain scenario on the Amazon dataset.

Multi-Purpose Hybrid Recommendation System on Artificial Intelligence to Improve Telemarketing Performance

The purpose of this study is to incorporate telemarketing processes to improve telemarketing performance. For this application, we have attempted to mix the model of machine learning to extract potential customers with personalisation techniques to derive recommended products from actual contact. Most of traditional recommendation systems were mainly in ways such as collaborative filtering, which predicts items with a high likelihood of future purchase, based on existing purchase transactions or preferences for products. But, under these systems, new users or items added to the system do not have sufficient information, and generally cause problems such as a cold start that can not obtain satisfactory recommendation items. Also, indiscriminate telemarketing attempts can backfire as they increase the dissatisfaction and fatigue of customers who do not want to be contacted. To this purpose, this study presented a multi-purpose hybrid recommendation algorithm to achieve two goals: to select customers with high possibility of contact, and to recommend products to selected customers. In addition, we used subscription data from telemarketing agency that handles insurance products to derive realistic applicability of the proposed recommendation system. Our proposed recommendation system would certainly solve the cold start and scarcity problem of existing recommendation algorithm by using contents information such as customer master information and telemarketing history. Also. the model could show excellent performance not only in terms of overall performance but also in terms of the recommendation success rate of the unpopular product.

A novel distributed Social Internet of Things service recommendation scheme based on LSH forest

For the Social Internet of Things (SIoT), the interaction among ever increasing number of smart devices results in an exponential increase of services, which leads to an extreme difficulty for users to find suitable services. To address this issue, most existing recommendation algorithms are based on the data stored on the centralized server and distributed schemes are ignored. Meanwhile, distributed recommendation algorithms face the problems of privacy leakage and efficiency, which decrease the quality of experience (QoE). Therefore, we propose a novel SIoT service recommendation scheme called SIoT- SR, which adopts LSH Forest while combining with collaborative filtering algorithm to predict the Quality of Service (QoS) data of users. The LSH forest implements binary search by sorting and also has the ability to self-correct parameters. It can achieve a good tradeoff among memory, accuracy, efficiency, and privacy. Finally, we validate the effectiveness of the scheme based on the dataset WS-DREAM. The experimental results show that SIoT-SR has high prediction accuracy and efficiency while saving computing resources and are suitable for service recommendation of SIoT with resource-constrained devices.

Holistic User Context-Aware Recommender Algorithm

Existing recommender algorithms lack dynamism, human focus, and serendipitous recommendations. The literature indicates that the context of a user influences user decisions, and when incorporated in recommender systems (RSs), novel and serendipitous recommendations can be realized. This article shows that social, cultural, psychological, and economic contexts of a user influence user traits or decisions. The article demonstrates a novel approach of incorporating holistic user context-aware knowledge in an algorithm to solve the highlighted problems. Web content mining and collaborative filtering approaches were used to develop a holistic user context-aware (HUC) algorithm. The algorithm was evaluated on a social network using online experimental evaluations. The algorithm demonstrated dynamism, novelty, and serendipity with an average of 84% novelty and 85% serendipity.

A Recommendation Algorithm Based on Item Genres Preference and GBRT

The most mature and widely used collaborative filtering algorithm is facing the problem of data sparsity, which is not conducive to the acquisition of user preferences, thus affecting the recommendation effect. Introducing item genres into recommendation algorithm can reduce the impact of data sparsity on recommendation effect. The personalized preferences of users can be extracted more effectively from the preference information of item genres, and the recommendation accuracy can be further improved. On this basis, this paper proposes a recommendation algorithm based on item genres preference and GBRT, which divides similar users by K-means clustering algorithm, extracts user preferences of item genres and auxiliary features, and establishes a rating prediction model combined with GBRT algorithm. The experiments on the common datasets of Movie Lens 100K and Movie Lens 1M show that the proposed algorithm achieves 0.8%-7% optimization on the evaluation index MAE, which indicates that the impact of data sparsity is reduced to a certain extent and the recommendation efficiency is better than the existing recommendation algorithm.

An Integrated Recommender System Using Semantic Web With Social Tagging System

Social tagging systems (STSs) allow collaborative users to share and annotate many types of resources with descriptive and semantically meaningful information in freely chosen text labels. STS provides three recommendations such as tag, item and user recommendations. Existing recommendation algorithms transform the three dimensional space of user, resource, and tag into two dimensions using pair relations in order to apply existing techniques. However, users may have different interests for an item, and items may have multiple facets. To circumvent this, a new system that models three types of entities user, tag and item in a STS as a 3-order tensor is proposed. The sparsity is reduced using stemming and predictions are made by applying latent semantic indexing using randomized singular value decomposition (RSVD). The proposal provides all the three recommendations using semantic web and shows notable improvements in terms of effectiveness through indices such as recall, precision, time and space.

Collaborative filtering recommendation algorithm integrating time windows and rating predictions

This paper describes a new collaborative filtering recommendation algorithm based on probability matrix factorization. The proposed algorithm decomposes the rating matrix into two nonnegative matrixes using a predictive rating model. After normalization processing, these two nonnegative matrixes provide useful probability semantics. The posterior distribution of the real part of the probability model is calculated by the variational inference method. Finally, the preferences for items that users have not rated can be predicted. The user–item rating matrix is supplemented by a preference prediction value, resulting in a dense rating matrix. Finally, time weighting is integrated into the rating matrix to construct the 3D user–item–time model, which gives the recommendation results. According to experiments using open Netflix, MovieLens, and Epinion datasets, the proposed algorithm is superior to several existing recommendation algorithms in terms of rating predictions and recommendation effects.

Improving Explainability of Recommendation System by Multi-sided Tensor Factorization

Recently, explainable recommender systems to improve their persuasiveness have attracted attentions. In this regard, some approaches extract information from posts or comments on items and apply them to simple and effective template. These information (e.g., topics and interests), however, are indirectly reflected to the existing recommendation algorithms or models therefore do not directly improve the recommendation accuracy. Moreover, extra resources in deriving information are required. Thereby, we propose a collaborative filtering approach using a tensor which is modeled considering 5Ws aspects and generate explanations by combining factorization results with templates. Quality and explanation of recommendations were evaluated on quantitative/qualitative analyses.

Recommendations based on user effective point-of-interest path

Point-of-interest (POI) recommendation has become an important service in location-based social networks. Existing recommendation algorithms provide users with a diverse pool of POIs. However, these algorithms tend to generate a list of unrelated POIs that user cannot continuously visit due to lack of appropriate associations. In this paper, we first proposed a concept that can recommend POIs by considering both category diversity features of POIs and possible associations of POIs. Then, we developed a top-k POI recommendation model based on effective path coverage. Moreover, considering this model has been proven to be a NP-hard problem, we developed a dynamic optimization algorithm to provide an approximate solution. Finally, we compared it with two popular algorithms by using two real-world datasets, and found that our proposed algorithm has better performance in terms of diversity and precision.

Personalized recommendation by matrix co-factorization with tags and time information

Personalized recommendation systems have solved the information overload problem caused by large volumes of Web data effectively. However, most existing recommendation algorithms are weak in handling the problem of rating data sparsity that characterizes most recommender systems and results in deteriorated recommendation accuracy. The results in the KDDCUP and Netflix competition have proven that the matrix factorization algorithm achieves better performance than other recommendation algorithms when the rating data is scarce. However, the highly sparse rating matrix will cause the overfitting problem in matrix factorization. Although regularization can relieve the issue of overfitting to some extent, it is still a significant challenge to train an effective model for recommender systems when the data is highly sparse. Therefore, this paper proposes a co-SVD model to enrich the single data source and mitigate the overfitting problem in matrix factorization. The user preferences are enriched not only by rating data but also the tag data; subsequently, the relevance between tags and item features are explored. Furthermore, according to the assumption that user preferences will change with time, we optimize the preference and relevance by adding the temporal influence. Based on the MovieLens benchmark datasets, the experimental results indicate that the proposed co-SVD method is more effective than other baselines. Matrix co-factorization provides an effective method to the solve data sparsity problem with additional information. The method can be used to address this problem in various expert and intelligent systems such as recommendation advertisements, e-commerce sites, and social media platforms, all of which require a relatively large amount of input data from users.

Analyzing Potential of Personal Values-Based User Modeling for Long Tail Item Recommendation

This paper examines the potential of personal values-based user modeling for long tail item recommendation. Long tail items are defined as those which are not popular but are preferred by small numbers of specific users. Although recommending long tail items to relevant users is beneficial for both the providers and consumers of such items, it is known to be a challenge for most recommendation algorithms. In particular, a long tail item is one that would be purchased and/or rated by a small number of users, so it is difficult to predict its rating accurately. This paper assumes that the influence of personal values becomes more obvious when users evaluate long tail items, and examines it through offline experiment. The Rating Matching Rate (RMRate) has been proposed in order to incorporate users’ personal values into recommender systems. As the RMRate models personal values as the weight of an item’s attribute, it is easy to incorporate into existing recommendation algorithms. An experiment was conducted to evaluate the performance of long tail item recommendation; Experimental result shows that personal values-based user modeling can recommend less popular items while maintaining precision.