scholarly journals SVD++ Recommendation Algorithm Based on Backtracking

Information ◽  
2020 ◽  
Vol 11 (7) ◽  
pp. 369
Author(s):  
Shijie Wang ◽  
Guiling Sun ◽  
Yangyang Li
Keyword(s):  
Line Search ◽  
Optimal Solution ◽  
Recall Rate ◽  
Personalized Recommendation ◽  
Recommendation Algorithm ◽  
Backtracking Line Search ◽  
Low Efficiency ◽  
Public Datasets ◽  
Value Decomposition ◽  
Local Gradient

Collaborative filtering (CF) has successfully achieved application in personalized recommendation systems. The singular value decomposition (SVD)++ algorithm is employed as an optimized SVD algorithm to enhance the accuracy of prediction by generating implicit feedback. However, the SVD++ algorithm is limited primarily by its low efficiency of calculation in the recommendation. To address this limitation of the algorithm, this study proposes a novel method to accelerate the computation of the SVD++ algorithm, which can help achieve more accurate recommendation results. The core of the proposed method is to conduct a backtracking line search in the SVD++ algorithm, optimize the recommendation algorithm, and find the optimal solution via the backtracking line search on the local gradient of the objective function. The algorithm is compared with the conventional CF algorithm in the FilmTrust, MovieLens 1 M and 10 M public datasets. The effectiveness of the proposed method is demonstrated by comparing the root mean square error, absolute mean error and recall rate simulation results.

scholarly journals Personalized News Recommendation and Simulation Based on Improved Collaborative Filtering Algorithm

Complexity ◽  
2020 ◽  
Vol 2020 ◽  
pp. 1-12
Author(s):  
Kunni Han
Keyword(s):  
Collaborative Filtering ◽  
Nearest Neighbor ◽  
Data Preprocessing ◽  
Recall Rate ◽  
Personalized Recommendation ◽  
Recommendation Algorithm ◽  
Data Set ◽  
Filtering Algorithm ◽  
Collaborative Filtering Algorithm ◽  
Recommendation Accuracy

Faced with massive amounts of online news, it is often difficult for the public to quickly locate the news they are interested in. The personalized recommendation technology can dig out the user’s interest points according to the user’s behavior habits, thereby recommending the news that may be of interest to the user. In this paper, improvements are made to the data preprocessing stage and the nearest neighbor collection stage of the collaborative filtering algorithm. In the data preprocessing stage, the user-item rating matrix is filled to alleviate its sparsity. The label factor and time factor are introduced to make the constructed user preference model have a better expression effect. In the stage of finding the nearest neighbor set, the collaborative filtering algorithm is combined with the dichotomous K-means algorithm, the user cluster matching the target user is selected as the search range of the nearest neighbor set, and the similarity measurement formula is improved. In order to verify the effectiveness of the algorithm proposed in this paper, this paper selects a simulated data set to test the performance of the proposed algorithm in terms of the average absolute error of recommendation, recommendation accuracy, and recall rate and compares it with the user-based collaborative filtering recommendation algorithm. In the simulation data set, the algorithm in this paper is superior to the traditional algorithm in most users. The algorithm in this paper decomposes the sparse matrix to reduce the impact of data sparsity on the traditional recommendation algorithm, thereby improving the recommendation accuracy and recall rate of the recommendation algorithm and reducing the recommendation error.

Personalized Recommendation Based on Contextual Awareness and Tensor Decomposition

2018 ◽  
Vol 16 (3) ◽  
pp. 39-51
Author(s):  
Zhenjiao Liu ◽  
Xinhua Wang ◽  
Tianlai Li ◽  
Lei Guo
Keyword(s):  
Recommendation System ◽  
Tensor Decomposition ◽  
Personalized Recommendation ◽  
Third Order ◽  
Recommendation Algorithm ◽  
Singular Value Decomposition Method ◽  
Contextual Awareness ◽  
Value Decomposition ◽  
Item Item

In order to solve users' rating sparsely problem existing in present recommender systems, this article proposes a personalized recommendation algorithm based on contextual awareness and tensor decomposition. Via this algorithm, it was first constructed two third-order tensors to represent six types of entities, including the user-user-item contexts and the item-item-user contexts. And then, this article uses a high order singular value decomposition method to mine the potential semantic association of the two third-order tensors above. Finally, the resulting tensors were combined to reach the recommendation list to respond the users' personalized query requests. Experimental results show that the proposed algorithm can effectively improve the effectiveness of the recommendation system. Especially in the case of sparse data, it can significantly improve the quality of the recommendation.

Personalized Recommendation Algorithm Based on Product Reviews

2018 ◽  
Vol 16 (3) ◽  
pp. 22-38
Author(s):  
Zhibo Wang ◽  
Mengyuan Wan ◽  
Xiaohui Cui ◽  
Lin Liu ◽  
Zixin Liu ◽  
...  
Keyword(s):  
Recommendation System ◽  
Information Overload ◽  
Recall Rate ◽  
Personalized Recommendation ◽  
The Internet ◽  
Product Reviews ◽  
Recommendation Algorithm ◽  
Preference Model ◽  
Product Features ◽  
Huge Data

Under the background of leap-forward development for the internet, e-commerce has played an important role in people's daily life, but huge data sizes have also brought problems, such as information overload which can be solved by using a recommendation system effectively. However, with the development of the e-commerce, the amount of the product catalogs and users becomes larger, which causes lower performance of the traditional recommendation system. This article comes up with a personalized recommendation algorithm based on the data mining of product reviews to optimize the performance of the new recommendation system. Features of the product were extracted, for which the users' sentiment polarity was analyzed. This article develops a recommendation system based on the user's preference model and the product features to get the recommendation result. Experimental results show that a personalized recommendation has significantly improved the accuracy and recall rate when compared with a traditional recommendation algorithm.

Optimized Collaborative Filtering Recommendation Based on User’s Social Relationships

2014 ◽  
Vol 1044-1045 ◽  
pp. 1484-1488
Author(s):  
Yue Kun Fan ◽  
Xin Ye Li ◽  
Meng Meng Cao
Keyword(s):  
Collaborative Filtering ◽  
Social Relationships ◽  
Digital Libraries ◽  
Social Relationship ◽  
Nearest Neighbor ◽  
Mean Absolute Error ◽  
Absolute Error ◽  
Personalized Recommendation ◽  
Recommendation Algorithm

Currently collaborative filtering is widely used in e-commerce, digital libraries and other areas of personalized recommendation service system. Nearest-neighbor algorithm is the earliest proposed and the main collaborative filtering recommendation algorithm, but the data sparsity and cold-start problems seriously affect the recommendation quality. To solve these problems, A collaborative filtering recommendation algorithm based on users' social relationships is proposed. 0n the basis of traditional filtering recommendation technology, it combines with the interested objects of user's social relationship and takes the advantage of the tags to projects marked by users and their interested objects to improve the methods of recommendation. The experimental results of MAE ((Mean Absolute Error)) verify that this method can get better quality of recommendation.

Sign in / Sign up

Export Citation Format

Share Document