Search

Aiming at the problem that the existing social recommendation algorithms ignore the potential association of users and the collaborative relationship between items, this paper proposed a new algorithm model GATHGN. This model framework unified modeled of user association and item relationship. Firstly, users’ mined explicit social relations, potential association relations and user-item association relations, so as to extract users’ social high-order characteristics and potential interest high-order characteristics. Then, the above two high-level features were aggregated based on the graph attention mechanism, and the user fusion features were updated layer by layer. Finally, the final recommendation results were calculated according to the updated user fusion characteristics and project characteristics. Experimental results on Yelp and Flickr data show that the HR and NDCG of GATHGN are significantly improved compared with the baseline algorithm. [ABSTRACT FROM AUTHOR]

In the recommendation system, the user’s ratings are often affected by the rating context, that is, the user’s previous ratings of some items will affect the objectivity of his rating of the current item. Sparse linear method treats user ratings affected by context as the same as other ratings when calculating item similarity. However, this partial ratings cannot objectively reflect the similarity between items. To solve the above problems, this paper proposed a recommendation algorithm combining rating context and item similarity based on sparse linear method. It divided the algorithm into three stages.The first stage used weighted ratings to calculate the item’s nearest neighbor for feature selection.In the second stage, it used the rating error weight to reduce the fitting of the ratings affected by the context of the algorithm model, and trained the item similarity matrix. In the third stage, it predicted the ratings according to the user’s ratings and the item similarity, and finally sorted the predicted ratings to complete the item recommendation. Experiments were conducted on four datasets of MovieLens, it used mean average precision (MAP), mean reciprocal rank (MRR) and normalized discounted cumulative gain (NDCG) to evaluate the effectiveness of the algorithm. The experimental results show that the fusion rating context will further improve the accuracy of item similarity and thus improve the performance of recommendation. [ABSTRACT FROM AUTHOR]

In recommender systems, graph convolutional networks have stronger information extraction capabilities for graphstructured data. However, existing graph convolutional networks mainly focus on enhancing the model structure, ignoring the importance of improving the sampling quality of the original samples and mining the implicit relationship between users and items. Aiming at the above problems, this paper proposed a graph contrastive learning recommendation algorithm based on mixed sampling. Firstly, the algorithm used a mixed sampling method to extract part of the information in positive samples and injected them into negative samples, thereby generating new informative hard negative samples. Secondly, to extract features on hard negative samples, it used the light graph convolution network to obtain node representations of users and items. Finally, it carried out a multi-task strategy to jointly optimize the recommendation supervision task and the contrastive learning task. The experiments on real datasets Yelp2018 and Amazon-book demonstrate that the proposed algorithm improves performance compared with other recommendation algorithms in recall and NDCG evaluation indexes. [ABSTRACT FROM AUTHOR]

In indoor localization, PDR method is one of the most convenient localization methods, but the cumulative error will get larger as the distance traveled increases. The use of Bluetooth peak correction algorithm can effectively reduce the cumulative error of PDR, but the Bluetooth signal has the problem of pseudo-peak and mis-peak in multi-Bluetooth base stations. To address the above problems, this paper proposed a PDR-based Bluetooth peak detection method. The method first calculated the prediction probability based on the PDR position estimation information and Bluetooth distribution information to filter out the closest Bluetooth in the forward direction, thus reducing false peaks. Secondly, it calculated the adaptive threshold based on the change of the signal, thus outputting the switching sate to find the wave peak, to detect the true peak and reduce the false peak. Finally, this paper used an improved Bluetooth peak correction algorithm to correct the PDR position estimation. The experimental results show that the proposed method can achieve a correct peak detection rate of 82.1% on Android smartphones . Compared with the traditional sliding window peak detection method, it can effectively reduce the interference of pseudo-peaks and false peaks, and reduce the cumulative error of PDR. [ABSTRACT FROM AUTHOR]

Most sequential recommendation systems（SRSs） assume that the next item to be predicted is related to the user’s previous input. However, in real scenarios, users may click items inconsistent with their own interests and preferences（unreliable instances） by mistake during browsing. This paper proposes a sequential recommendation algorithm based on Gaussian distribution modeling to solve this problem. Firstly, the algorithm extracts the input sequence patterns by reducing the uncertainty of input items through an uncertainty aware graph ensemble network（UAN） with multiple heads of self-attention; Secondly, the extracted input sequence pattern is modeled as a Gaussian distribution, and obtain the dynamic user’s preferences and the uncertainty of preferences in the sequence information; Then, extend the traditional recommended objective function to a sampling loss function and an uncertainty regularizer, and give each training instance appropriate uncertainty; Finally, remove the unreliable examples with high loss and low uncertainty to enhance the accuracy of sequence recommendation. The algorithm is tested on three open datasets, Book Cross, Movielens-1 M and Steam. The results show that the algorithm has achieved an improvement of about 5.3% compared with the baseline with good effects, and has obtained better sequential recommendation results. The proposed algorithm can improve the recommendation accuracy by effectively reducing the uncertainty of input sequence information. [ABSTRACT FROM AUTHOR]

Although collaborative filtering algorithms become widely applicable, they also suffer from the difficulties of scalability and extensibility. To address this problem, this paper proposed a collaborative filtering model for recommendation systems based on the classification of user complex network features（UCNFC）. Firstly, the method selects feature users based on degree values in the user set and establishes similarity thresholds to implement groups of non-feature users. Then, user-user similarity constructs the network and completes the community labeling in the network by K-core decomposition. Finally, the target users choose their neighbors within themselves group to accomplish movie rating prediction. Based on the experimental results of Movie Lens and Netflix datasets, the algorithm shows improved both temporal and spatial performance compared with the classical collaborative filtering algorithm, demonstrating more excellent scalability and extensibility. [ABSTRACT FROM AUTHOR]

Traditional collaborative filtering methods largely model in a uniform way without considering the modeling relationships at the fine-grained level of interaction intent. This paper proposes a graph neural network collaborative filtering algorithm fused with interaction intention, named INTNGCF. Firstly, the potential intention is identified by decomposing edges into multiple potential spaces; Secondly, the interaction intention fusion layer is used to determine the importance of these potential intentions. Finally, a predicted user rating of the project is generated. Compared with seven baseline models on three real datasets, the results show that the proposed algorithm has certain performance advantages compared with other advanced recommendation models. [ABSTRACT FROM AUTHOR]

The recommendation algorithm based on graph neural network can extract the association relationship between users and goods. Traditional methods can’t extract this relationship. At present, most of these algorithms ignore the general prefe-rences in the review data of users and products. In order to solve this problem, this paper proposed a new method. This me-thod used the graph neural network to extract association relations, and the advantage of deep learning to extract the general preferences, and carried out feature fusion to improve the recommendation effect. This paper conducted comparative experiments and ablation experiments on four sets of public data sets to verify the effectiveness of the proposed method. The evaluation indexes include the recall rate and normalized discounted cumulative gain. Experiments show that this method is more effective than the existing algorithms. The feature fusion of the two networks can improve the recommendation effect. [ABSTRACT FROM AUTHOR]

The function of recommendation system is to help users actively finding personalized items that meet theirs preferences and recommend them to users. Collaborative filtering algorithm is a classic algorithm in recommender system, but it is limited by cold start of data and sparsity and has disadvantages such as poor interpretability and poor model generalization ability . This paper studied its shortcomings. By taking the original score matrix in the form of user-project bipartite graph as input, designing the figure convolution neural network as a variant of graph autoencoder, which was obtained by the latent vector of user and item by iteratively aggregating neighbor node information, and combining CNN, this paper proposed a recommendation algorithm based on convolution matrix decomposition to improve the interpretability of the model and generalization ability. And also solved the auxiliary information fusion the data sparseness, and made recommendation performance improved by 1.4% and 1.7% . It provides a new idea for recommendation direction based on graph neural network in the future. [ABSTRACT FROM AUTHOR]

In order to improve the accuracy and quality of the personalized recommendation system, this paper constructed a collaborative filtering recommendation algorithm based on SVD and intuitionistic fuzzy clustering ( SVDIFC-CF) to address the information overload and data sparsity problems in traditional recommendation algorithms. The algorithm firstly introduced SVD to fill in the original matrix after dimensionality reduction. Then it used the user-item preference matrix to perform intuitionistic fuzzy clustering of users. Finally it calculated the top N users with the highest similarity to the target user, and found the items that users were most interested in as recommendation results. This paper used the Movie Lens and Jester datasets to verify the effectiveness of the algorithm. The experimental results show that compared with the traditional recommendation algorithm, this algorithm can effectively solve the problem of data sparseness and cold start, and improve the accuracy and quality of recommendation. [ABSTRACT FROM AUTHOR]

Deep matrix factorization adopts deep nonlinear mapping to break through the bottleneck of bilinear relationship affecting the performance of the recommendation system in matrix factorization. However, it doesn't take into account users' preference for unrated items, and its recommendation performance isn't advantageous for large-scale data with high sparsity. This paper proposed a recommendation algorithm for fusing matrix completion and deep matrix factorization. First, it filled the unknown elements in the original scoring matrix by the matrix completion model. Then, based on the completed matrix, it constructed the latent vectors of users and items by deep learning model. Finally, this paper tested the proposed model on the MovieLens and SUSHI datasets. Experimental results show that the proposed algorithm significantly improves the performance of the recommendation system compared to deep matrix factorization. [ABSTRACT FROM AUTHOR]

In recommender systems, extracting higher order information about the graph using methods such as graph convolutional network alleviates the cold start problem. In order to fuse the deep feature extraction capability of neural network collaborative filtering on this basis, this paper proposed a two-channel collaborative filtering recommendation algorithm (GCNCF-2C) based on graph convolution. Firstly, it divided the recommendation problem into an upstream task and a downstream task. Secondly, to form a feature representation of nodes, in the upstream task, a pre-trained encoder extracted node features and higher-order features of the graph using a one-dimensional convolutional layer containing residuals and multiple graph convolutional layers in two independent channels. Finally, the decoder performed end-to-end training by node features for rating prediction. Experiments on the datasets MovieLens-100k and MovieLens-1M show that compared to the baseline models, the algorithm improves RMSE metrics by 1.72% and 1.76% on average and MAE metrics by 2.7% and 1.98% on average on the two datasets. Meanwhile, it improves RMSE metrics by 5.9% on average in the cold-start experiments based on users and projects, which shows overall performance. [ABSTRACT FROM AUTHOR]

To address the problem that the Bayesian personalized sorting algorithm fails to fully apply the user's behavior information, which leads to a significant reduction in recommendation performance and robustness under sparse data conditions, this paper proposed the mean Bayesian personalized tanking (MBPR) algorithm to further explore the user's preference for implicit feedback information. Considering the characteristics of user interest changing over time, the MBPR algorithm introduced the forgetting function. The algorithm firstly preprocessed the historical rating records of users, then divided the items into positive and negative feedback based on the rating information of users, personalized the modeling for each user, explored the users' preferences for unengaged items, and generated a list of recommendations. To verify the recommendation performance of the proposed algorithm, this paper conducted analysis and comparison experiments on the public data sets MovieLens and Yahoo. Experimental results show that the recommendation performance and robustness of the proposed algorithm are significantly improved over the comparison algorithm. [ABSTRACT FROM AUTHOR]

Applying knowledge graph to recommendation system can make use of semantic relations between entities of knowledge graph to learn user and item representation. The embedding propagation method uses the graph structure of the knowledge graph to learn relevant features, but the semantic dependency between multi-hop entities decreases as the propagation range increases. In order to effectively improve the semantic expression ability of recommendation and improve the accuracy of recommendation, this paper proposes a knowledge-aware propagation recommendation algorithm based on users’ potential interests. The model adopts heterogeneous propagation method to disseminate item relevant knowledge and iteratively learn users’ potential interests, so as to enhance the representation ability of the model to users and items. Specifically, first, graph embedding layer generate initialize representation of users and items, and in the heterogeneous propagation layer, the knowledge-aware attention mechanism can distinguish the importance of entities in the same layer, so the model can generate the representation of target entities more accurately. Then the user’s potential interest propagation can effectively learn the user’s higher-order potential interest and enhance the semantic relevance of multi-hop entities. Finally, information decay factor is used in the prediction layer to distinguish the importance of different communication levels and generate the final representation of users and items. Experiments show that the AUC value of the model on the Last. FM and Book-Crossing increases by 2.25% and 4.71% compared with the most advanced baseline, and the F1 value increases by 3.05% and 1.20% respectively, and the Recall@K value is superior to the comparison baseline model. The proposed model can effectively improve the accuracy of recommendation. [ABSTRACT FROM AUTHOR]

Aiming at the fact that traditional linear regression recommendation algorithms don ' t take into account influencing factors such as user interest drift, activity, and credibility, this paper proposed a linear regression recommendation algorithm incorporated credibility ratings to further improve the accuracy of the algorithm and the fit to user preferences. Firstly, this paper comprehensively considered the user' s interest drift, activity and user evaluation information in the calculation method of user scoring credibility. Then it integrated this algorithm into the coefficient matrix solution process of traditional linear regression recommendation algorithm. Finally, it used the optimized linear regression recommendation algorithm to predict user scoring. In order to verify the accuracy of the algorithm, this paper compared the proposed algorithm with the traditional linear regression recommendation algorithms on Hadoop cluster and Amazon product scoring dataset. The experimental results show that the algorithm has significantly improved the processing efficiency, recommendation effect and fitting degree. [ABSTRACT FROM AUTHOR]

The imputation-based solution can alleviate the sparsity problem of recommendation system. Improved Jaccard similarity based on logistic user habit rating weighting, which is called the HabJac. By combining this metric, the K nearest neighbor ( KNN) obtains the imputation value. Furthermore, this paper proposed a new HISVD recommendation algorithm by combining the regularized singular value decomposition ( RSVD) technology to predict the unknown ratings. Firstly, the user' s habit rating was the most frequent one. Secondly, logistic weight was positively correlated with Euclidean distance between rating and habit one . The experimental results on four real data sets show that: a ) the optimal parameters of HISVD algorithm for different data sets are similar; b) HISVD surpasses the mainstream competitors for the MAE and RSME. [ABSTRACT FROM AUTHOR]

Focusing on problems of the accuracy recommendation system that the recommended commodity type is relatively single, and commodities are mostly popular goods and lack of freshness, the novelty recommendation is gradually gaining attention. However, current researches don't combine item features when designing algorithms, which make them unable to distinguish and select items with higher novelty for different users. In order to improve the performance of the recommendation system, this paper improved the method based on random walk and designed a new recommendation algorithm by fusing novelty features. This algorithm further analyzed features of items and gave the formal definition of the novelty from perspectives of user interest expansion and prediction. This paper analyzed user demands, constructed new transition probability, generated personalized recommendation lists and improved the novelty of the lists. The experimental results show that the proposed algorithm has less influence on the accuracy than existing methods and has significant improvement on novelty indexes. It concludes that by fusing novel features, this system can improve the recommendation contents effectively while taking into account the accuracy. [ABSTRACT FROM AUTHOR]

Matrix factorization is one of the most important algorithms in recommendation systems due to its simple and reliable characteristics. However, since the inner product can't fully capture the interaction between users and items, it is difficult to further enhance the performance of matrix factorization. In order to solve this problem, this paper modified the basic distance metric factorization model and proposed a collaborative filtering recommendation algorithm based on biased metric factorization and implicit feedback,and dynamically modeled users' rating timestamps, which further improved the model performance. This paper tested the most common rating prediction task in the recommendation system on three data sets. The experimental results show that the prediction accuracy of the proposed model is significantly improved. [ABSTRACT FROM AUTHOR]

Based on matrix factorization techniques, this paper implemented a modified binomial matrix decomposition algorithm in order to solve the recommender system's rating prediction problem. It supposed that the user's rating of the item was based on the binomial distribution. There were differences in the user's rating habits, popularity of the items, it would result in an offset in the user-item scoring matrix. This paper used the maximum a posteriori estimate to design model and optimized the model by a stochastic gradient descent algorithm. The experimental results show that the modified binomial matrix decomposition algorithm is superior to the traditional binomial matrix decomposition algorithm in terms of recommender accuracy and offline calculation time on the MovieLens lOOK datasets. [ABSTRACT FROM AUTHOR]

The traditional matrix decomposition model cannot fully explored the intrinsic relationship between the user and the object in the mean, bias and characteristics. This paper proposed a fitting matrix model to improve the prediction performance by constructing the user and the item matrix to represent the characteristics of the user and the item respectively. The matrix decomposition model had the advantage of accuracy in the field of recommender system, but the gradient descent method, which was the most popular method to train parameters of model, had a slow convergence speed. To resolve the above defects, this paper considered to accelerate the convergence speed using the convergence of quasi Newton method, and named the proposed algorithm as fitting matrix and two orders fusion iterative (FAST) algorithm. The experimental results show that the FAST algorithm is better than the traditional non negative matrix decomposition (NMF), singular value matrix decomposition (SVD), and the regularized singular value matrix decomposition (RSVD). FAST algorithm has a decrease with regard to the mean absolute error (MAE) and the root mean square error (RMSE), and has a significant improvement in the iterative efficiency, which alleviates the problem that the accuracy is difficult to balance with the efficiency of the iteration. [ABSTRACT FROM AUTHOR]

In recent years, as deep learning has achieved good results in many fields, deep learning has also begun to be applied to recommendation systems, such as NFM models and DeepFM models that use deep learning technology to capture high-level feature interactions. However, considering the changes in the external environment and internal perception, the user' s interest should also change dynamically over time, and the combination based on the original features may not necessarily learn effective feature interaction. This paper attempted to build a new model FG_DRFwFm, which could learn the interaction of low-level and high-level features of multiple feature domains and dealt with long-term changes in user interest. It constructed the training features by constructing new features based on the original features and splicing them together, which could better learn effective feature interaction. Finally, the proposed model compared the recommendation effect with multiple advanced CTR algorithms on the MovieLens data set. The experimental results show that the proposed model achieves better results. [ABSTRACT FROM AUTHOR]

Data sparsity and user cold start are always the problems to be solved in the recommendation system. This paper proposed a cross domain recommendation algorithm SRTCD. Considering the similarity of latent factors of user groups / items in different domains, the algorithm decomposed the rating matrix of each domain into probability matrix to obtain the latent factors of users and items. Then, it used K-means algorithm based on simulated annealing and genetic algorithm to cluster users and items respectively, and used the inner product of user category and items category as shared rating. Then, it obtained the recommendation results by using the shared rating of each domain dataset and the specific rating of the target domain dataset. Finally, it verified the proposed method by public data sets, and the results show that the performance of SRTCD is significantly better than that of common recommendation algorithms. [ABSTRACT FROM AUTHOR]

In order to improve the performance of the recommendation system, the hot research of personalized recommendation field is how to build more accurate recommendation algorithms and models with massive user browsing behavior data. Firstly, this paper briefly described the browsing behavior of users, gave the overall framework of the browsing behavior recommendation system, and summarized the development of the recommendation system based on user browsing behavior. It summarized, compared and analyzed the key techniques and single browsing behavior quantization method and mixed browsing behavior quantization method. Finally, it discussed the latest achievements of browsing behavior recommendation combined with multi-source heterogeneous data and summarized future research difficulties and development trends in this field. [ABSTRACT FROM AUTHOR]

In order to solve the problems such as sparse data, cold start and lack of diversity of recommendation results in traditional collaborative filtering recommendation algorithms, this paper proposed a collaborative filtering recommendation algorithm that integrated social networking with key users . Based on the score matrix of user projects, the proposed algorithm integrated user social networks to de rive social trust matrix, integrated key user information to obtain key user scoring matrix. And then it used these three matrix data distributions to predict user's target project with different weights score. At the same time, aiming at the massive data problem, this paper used the Spark distributed cluster to realize the parallelization of the algorithm. The experimental results show that the proposed algorithm can effectively alleviate the data sparse problem and improve the data processing speed and recommendation accuracy. [ABSTRACT FROM AUTHOR]

Shilling attack is one of serious challenges which recommender systems are facing. Malicious users can easily insert well-designed ratings into recommender systems to affect recommendation results and decrease user experiences because of the openness of recommender systems. This paper proposed a robust shilling-attack-tolerant personalized recommendation（ SATPR） algorithm based on attribution facilitated matrix completion with structural noise technology,regarded the ratings of attack users in the rating matrix as structural row noise and modeled them with L2,1-norm. This paper also introduced attributive characters of users and items to improve the accuracy of detection of shilling-attack. Experimental results show that SATPR algorithm achieves more accurate results of personalized rating prediction than traditional recommendation algorithms under shilling attacks. [ABSTRACT FROM AUTHOR]

At present, in the recommendation task based on document information, the traditional document-based hybrid recommendation algorithm still relies on a shallow linear model. When the scoring data becomes huge and complex, its recommendation performance is often not ideal. This paper proposed a deep fusion model( DeepFM), which could fully capture text information and handle complex and sparse scoring data well. DeepFM consisted of two parallel neural networks . One of them used multiple layers of perceptron to extract the row vector information of the rating matrix to obtain the user' s potential feature vector. The other one used fusion model of MLP and convolutional neural network (CNN) to extract additional textual information about the item and obtain the item' s potential feature vector. Finally, it fused the user feature vector and the item feature vector by constructing a fusion layer to obtain a prediction score. Experimental results show that DeepFM outperforms mainstream recommendation models on MovieLens dataset and Amazon dataset. [ABSTRACT FROM AUTHOR]

Concerning the problem that data sparsity and user preferences quickly change for traditional news recommendation, this paper proposed a hybrid news recommendation algorithm combining social relations and tag information. Firstly, the algorithm utilized the social relationship and hashtag information in the user's social network. Then it applied the LDA topic model to model user interest. Finally, the algorithm used a hybrid recommendation algorithm based on content and collaborative filtering to complete news recommendations. In the experiments, comparing with existing recommendation algorithms, the proposed algorithm can improve the precision by 10.7%, MRR by 4.1 %, NDCG by 10%. The proposed algorithm can improve the accuracy and the quality of news recommendation algorithm effectively. [ABSTRACT FROM AUTHOR]

Nowadays, user 's friendship relationship and itself are dynamic, therefore, the recommendation algorithm based on static social relationship is difficult to meet the needs of today's rapidly changing world. To solve the problem of low accuracy, this paper proposed a method utilized user's time-series behavior of purchasing item to mine implicit social relationship. Firstly, it integrated implicit social and similarity algorithm. Then, it proposed the improved neighbor score filling method for the sparsity of the neighbor score. Finally, it used the filled neighbor score to modify the model prediction score and generate the final prediction score. The experimental part used MovieLens dataset to evaluate the performance of proposed method and compare it with the existing algorithms. The experimental results show that the proposed algorithm is more stable and effective in predicting the true score of target users than traditional algorithms and improved algorithms. [ABSTRACT FROM AUTHOR]

The traditional collaborative filtering recommendation algorithm has the problems of low timing and high computational complexity of similar users with sparse data when there are too many users. According to the above problem, this paper proposed a spectral clustering optimization recommendation algorithm combining the minimum hash signature and temporal model prediction. Firstly, it used the MHS and Levenshtein distance measure to extract similar users from user-item scoring matrix. Then it used AIM-RT prediction algorithm with weight fitting to complete sparse matrix of similar users. Finally,it carried out similar user optimization by combining spectral clustering to find the optimal set of similar users and completed the final recommendation. Experimental analysis and verification show that the proposed recommendation algorithm can improve the overall recommendation performance in terms of computational complexity, scoring prediction accuracy, data missing filling and other aspects. [ABSTRACT FROM AUTHOR]

The traditional matrix factorization algorithm uses all the training data as a whole to obtain user interest features from the perspective of time, ignoring the drift of user interest. To address this problem, this paper proposed a recommendation algorithm based on state space model and probability matrix factorization. Firstly, this method mapped the matrix factorization of user interest feature vector and user rating data into linear Gaussian state space. Then, it used EM algorithm and Kalman filter to solve the model parameters dynamically. Finally, it obtained the user interest feature vector to confirm whether the user interest drifts or not, and made the recommendation. The experiments demonstrate that the proposed algorithm can better perceive user interest drift and improve the recommendation quality compared with the traditional matrix factorization algorithm. [ABSTRACT FROM AUTHOR]

For the high dimension and insufficient sample of side information and user comment input in the recommendation algorithm, this paper proposed an attentive collaborative side variational autoencoder recommender model (sVAE-a) for the nonlinear modeling ability of variational autoencoder and the related data enhancement ability of the attention mechanism. This model used collaborative side variational autoencoder to model the side information. At the same time, it used the attention mechanism to integrate the side information into the collaborative variational autoencoder architecture to strengthen the hidden variables and provide the decoder with cleaner features. At last, it used the variational inference to approximate the side information and the user's comments, and obtained the recommended model through the parameters training. The experimental results on the MovieLens-20M dataset show that the method can improve not only from the basic recall rate index, but also from the coverage rate and normalized damage cumulative gain (NOCG) index. The model is easy to implement. It can combine different types of input and side information, and improve recommendation performance. [ABSTRACT FROM AUTHOR]

By establishing the binary relationship between users and information products, the recommender system makes use of the data generated by user behavior to mine the objects that each user is interested in and make recommendations, userbased collaborative filtering has been a mainstream approach in recent years, but it has a limitation : recommendations need to consider all users, and a specific user is often similar to a small number of users. To solve this problem, this paper proposed a collaborative filtering algorithm based on improved Canopy clustering, which combined the user model data density, distance and user activity to calculate the weights of the user, then clustered the user model data, the idea of clustering based on Canopy made one user could belong to different classes, which fit in with situations that users might be interested in multiple areas. Finally, corresponding recommendations were made for each user in Canopy, and it predicted the objects that users might be interested in based on the clustering result and user score. By comparing with other algorithms on two real-world data sets MovieLens and million songs, it verifies that the proposed algorithm can significantly improve the accuracy of the recommender system. [ABSTRACT FROM AUTHOR]

The data sparsity and cold start problems of the collaborative filtering recommendation algorithm affect the quality of the recommendation. Traditional user-item bipartite graph trust and the similarity calculation is negatively affected by the local individual point correlation factors. Firstly, this paper proposed a bipartite graph trust recommendation based on project popularity calculated by a grey correlational analysis by degree of balance and approach. On this basis, it proposed the user's preference enhanced bipartite graph direct trust degree mechanism, and used the JMSD correlation coefficient as the global trust recommendation. The comparison experiments under the MovieLens dataset show that the improved algorithm has a lower mean absolute error( MAE) compared with the benchmark algorithm, it improves the recommendation quality and alleviates the cold start problem. [ABSTRACT FROM AUTHOR]

Aiming at the problems of privacy security and inefficient recommendation system in dynamic Internet of Things, this paper proposed a blockchain cognitive IoT framework with high efficiency and privacy. It divided the proposed framework into three layers: block chain IoT management layer, cognitive layer and requirements layer. The blockchain IoT management layer provided the required information for the cognitive layer, and then operated the manageable elements of the system safely and privacy. In the cognitive layer, the cognitive engine observed the information about the system, and then executed the appropriate algorithm to manage the system. In the requirements layer, it used the cognitive specification language( CSL) to describe the goals and behavior of the network. The cognitive recommendation system of the proposed blockchain IoT framework learned from past experiences and improved decision-making on IoT recommendations. Compared with other Internet of Things frameworks, the proposed framework and recommendation system have privacy security and high performance recommendation capabilities. [ABSTRACT FROM AUTHOR]

The context-aware recommender system is a research hotspot in the current recommendation field. The CAMF-CC is an effective model in the current field. In view of the low accuracy of the CAMF-CC algorithm in recommendation, this paper presented a new algorithm called ICAMF-CC, which was based on CAMF-CC and incorporated item bias and context weights to predict scores. The advantage of ICAMF-CC was that, on the one hand, it incorporated the influence of the item's own characteristics on the score; on the other hand, it fully considered that different contextual factors had different influences in the recommendation process and improved the accuracy of the rating prediction. Experiments on the LD0S-CoMoDa dataset show that ICAMF-CC is superior to the CAMF-CC algorithm in accuracy. [ABSTRACT FROM AUTHOR]

Neighborhood-based top-N recommendation techniques for implicit users ' inte raction data are ranking models, of which similarity functions are the important ingredients. Traditional similarity functions suffer from two major issues: sparse data and high dimensional data. Sparse data hinder the recommendation system from scoring on smooth neighborhood, high dimensional data causes the curse of dimensionality problem. This paper proposed a representation-based approach named multi object representation learning ( MO) for recommendation, MO was a bipartite node representation learning algorithm, which embedded the different level of network structures and item ordering information into nodes representations to help leverage the recommendation performance. Experimental results on three real data sets of different scales show that the algorithm has higher accuracy and recall than the commonly used recommendation model based on implicit feedback, especially for large-scale data sets, which can effectively alleviate the problems of matrix sparsity and dimensionality disaster, and improve the recommended performance. [ABSTRACT FROM AUTHOR]

The current trust-based recommendation algorithms did not fully exploit the trust relationship between users and it lacked reasonable trust transitivity rules, which greatly affected the reliability and accuracy of the recommendation algorithm. Aiming at the above problems, this paper combined user rating data with the user's social relationship to build a trust transitivity model, and proposed a recommendation algorithm based on trust transitivity. Firstly, the algorithm used the score data to calculate the implicit direct trust relationship of the users in the trust transitivity model. Secondly, it solved the indirect trust relationship of multiple trust transitivity chains by solving the ordered weighted average operator. Finally, it converge the user's trust and similarity into comprehensive similarity for predictive recommendation. The experimental results show that the proposed algorithm can effectively improve the recommendation quality of the system. [ABSTRACT FROM AUTHOR]

In the recommendation system, this paper proposed an XGBoost recommendation algorithm to integrate collaborative filtering based on the cold-start problem of users. Firstly, it used coarse grain to recall according to the collaborative filtering recommendation algorithm based on user similarity, and got a recall set of some users. Then it used XGBoost algorithm to predict the items in the recall set. Secondly, for users with cold-start problems, it could directly use XC Boost algorithm to predict the items in the candidate set. Finally, the algorithm used the online evaluation data set of CCIR2018 personalized recommendation evaluation, and put the recommendation results on the online platform provided by Zhihu for evaluation. The evaluation results show that the proposed algorithm can solve the cold-start problem of users with high efficiency and accuracy. It achieves remarkable recommendation effect in the online evaluation platform. [ABSTRACT FROM AUTHOR]

Aiming at the sparseness and low prediction accuracy of traditional collaborative filtering algorithms, this paper proposed a hybrid recommendation algorithm based on restricted Boltzmann machine and weighted Slope One. Firstly, it used the preliminary filling of the scoring matrix by the restricted Boltzmann machine to alleviate the sparseness problem of the data. Then, it introduced the project attribute information through a hybrid project similarity calculation method. Finally, it adopted the second prediction by the weighted Slope One algorithm to improve the recommended effect. Experiments on the MovieLenslOOK dataset show that the combination of the two algorithms increases the accuracy of the recommendation. [ABSTRACT FROM AUTHOR]

Most recommendation systems based on the network structure suffer from lack of diversity, so that the paper proposed a recommendation system of big data based on multi-weight projection of bipartite network. Firstly, it abstracted the basic information of datasets, and applied items-users lists as an input to Levenshtein distance algorithm to compute similarity of each property. Then, it computed the number of common neighbors of the nodes in the bipartite network, the degree of common neighbors of the nodes in the bipartite network and degree of each node in bipartite network, and also computed the triple weights of each side of the bipartite network. Lastly, it adopted the enhanced bipartite projection technique to abstract the potential links of the bipartite network to realize the link prediction based on similarity. It realized the experiments based on both of big dataset and small dataset. The results show that the proposed algorithm outperforms different kinds of recommendation systems in terms of accuracy and coverage of recommendation, at the same time, it outperforms the other recommendation system based on network structure. [ABSTRACT FROM AUTHOR]

This paper studied the algorithm of drug repositioning based on collaborative filtering. Considering the great influence of data sparsity on collaborative filtering algorithm, it proposed a hybrid recommendation algorithm based on the association of drug and disease characteristics. The algorithm not only used the data of drug and disease, but also used the information of drug structure, target protein, side effect and drug-disease feature matrix to calculate the similarity between drugs, which reduced the influence of data sparsity to the recommendation effect and improved the precision of recommendation. The results of contrastive experiment show that the algorithm has a good recommendation effect, and can explore the drug-disease combinations which have po tential relationship, and further verified that the algorithm can be effectively applied to drug repositioning. [ABSTRACT FROM AUTHOR]

Aiming at the problem of defects in the accuracy of recommendation and data sparse processing for recommendations made by explicit feedback information. This paper introduced implicit feedback information, designed and implemented a multi-dimensional recommendation algorithm( iMCF) based on implicit feedback. The algorithm included three dimensional information of user, project and implicit feedback. For the first two dimensional information, it used the similarity of cloud model to model. For the information of implicit feedback, it mainly used the probabilistic matrix decomposition model to deal with. Afterwards, it balanced the prediction scores obtained from the three dimensions according to the weights, and obtained the final prediction score and made recommendation. The experimental data shows that the algorithm's performance in terms of recall rate and accuracy is significantly improved compared to other algorithms, and it is suitable for big data environments. [ABSTRACT FROM AUTHOR]

Aiming at the scalability problem in collaborative filtering recommendation algorithm, on the basis of the original information core extraction method based on FB and RB, this paper proposed an improved extraction information core method IFB and IRB. When in search of the most similar neighbors, it proposed a concept: optimization set, and found the most similar neighbors for each user on this set. The experimental results show that this method can get more accurate recommendation results, and reduce the mean average absolute error( MAE) effectively. At the same time, it has higher precision and recall, so it has better recommendation effect. [ABSTRACT FROM AUTHOR]

For the scoring matrix as the unique algorithm input of the collaborative filtering recommendation algorithm, the differences in the quality of the data have great impact on the recommendation results, including arousing the attention to data quality, how to characterize quality differences, and how to group users and recommend on the basis of user groups with different quality data . This paper proposed a description of data quality, comprehensively considered the user behavior consistency and user information entropy to evaluate the data quality. Users of different groups could perform more accurate recommendation results based on analyzing their historical behavior. The experimental results show that the difference of data quality has an important impact on the improvement of recommendation accuracy, and at the same time demonstrates the necessity of group recommendation. The experimental results also show that the accuracy of the combination of the two aspects of user behavior consistency and user information entropy is the most significant. [ABSTRACT FROM AUTHOR]

Tradition al personalized recommendation algorithm generally suffers from the problem of data sparseness, which affects the accuracy of recommendation. The Slope One algorithm is simple and efficient, but the algorithm is only based on the user - project score matrix to analyze the data, ignoring the type characteristics of the project and the user's preference for the type of the project. In order to solve the above problems, this paper proposed the LR-Slope One algorithm. Firstly, it constructed the user's preference matrix based on user project score matrix and project type information. Secondly, it calculated the weight of each type by linear regression model and optimized the weight by random gradient descending algorithm. Finally, it predicted the score combined by Slope One, filled the scoring matrix, which improved the quality of recommendation. Experimental results show that the proposed algorithm improves the accuracy of recommendation and alleviates data sparseness effectively. [ABSTRACT FROM AUTHOR]

Most of the existing modeling methods of combining metadata are based on the same user I item attribute weights, so that the key relationships between users and items are not significant, and it is difficult to obtain better recommendation performance. To solve the above problems, this paper proposed a method of deep joint learning recommendation based on metadata and attention mechanism. It used double deep network joint learning, one of the networks implemented matrix nonlinear decomposition based on implicit feedback data to learn user/ project personalization relationship, and another network automatically captured influence the user/ item key attributes to recommend by using attention mechanism, through the user preference relation with weighted of different attributes modeling highlighted the extended model. Experimental results show that the proposed recommendation algorithm has better recommendation performance on two public datasets of MovieLens 100K and MovieLens lM. [ABSTRACT FROM AUTHOR]

Aiming at the problem that there are item deviation and user preferences in collaborative filtering recommendation for the interaction between factors related in items and users, this paper proposed a recommendation algorithm integrated item deviation and user preference. Firstly it clustered to generate item clusters on LDA topics modeling and to get user clusters by using K-means; then it generated item deviation score on the constraints of item cluster and user cluster, and obtained user preference score with probability transfer on user-item score and item type. Finally it weighted the item deviation score and user preference score linearly to form the prediction score based on the existing scoring average in the item cluster. Comparison experiments show that the new algorithm can obtain reasonable recommendation based on different neighbors and improve recommendation accuracy. [ABSTRACT FROM AUTHOR]

This paper proposed a linear regression algorithm to integrate collaborative filtering based on the data sparse influence of the traditional collaborative filtering algorithm. Firstly, it built a similarity matrix between the user and the project based on the user’s rating of the project, as well as the user and the project’s own characteristics. Secondly, based on the similarity matrix, it selected the user and project nearest neighbor set. It predicted the score that the users had graded respectively by the way of collaborative filtering algorithms based on the user and the project. And it would take the difference between predicted scores and the real scores as features to generate new training data, and regarded the new training data as the input of the linear regression model. Finally, according to the training model, it could predict the unknown score, and used the top-N algorithm to generate the recommended list. It conducted the experiment on the MovieLens data set. The experimental result shows that the proposed accuracy of the new algorithm improves compared with the traditional collaborative filtering algorithm. [ABSTRACT FROM AUTHOR]

In order to solve the problem of data sparsity in the collaborative filtering algorithm, this paper proposed an improved collaborative filtering algorithm. The algorithm filled the unrated items in the scoring matrix using the prediction value calculated by the slope-one algorithm and then gave recommendations based on the user-based collaborative filtering method based on the filled user-item scoring matrix. Using the predictive value of slope-one algorithm as the backfill value could not only reduce the sparsity of the scoring matrix, but also ensured the diversity of backfill values, so as to reduce the recommended error caused by the single fill value such as mean value and median value. It performed half off cross-validation experiments on the MovieLens-lM dataset. The results show that the collaborative filtering algorithm based on the score prediction data effectively mitigates data sparsity and has better performance recommended effect. [ABSTRACT FROM AUTHOR]

According to the sparse characteristics of big data, singular value decomposition method was introduced into collaborative filtering algorithm to calculate and verify the page interest degree of internal site users, this paper proposed a user page interest prediction algorithm based on improved collaborative filtering algorithm. The algorithm could extract the false score in the dominant user score data in the network log file, and found user page interest and influence factors. The results of MATLAB simulation show that the collaborative filtering algorithm based on singular value decomposition is accurate and efficient in predicting the interest degree of user pages under the unavoidable situation of massive data sparsity. [ABSTRACT FROM AUTHOR]