A personalized recommendation method and recommendation device based on user behavior
A recommendation method and technology of a recommendation device, applied in special data processing applications, instruments, electrical digital data processing, etc., can solve the problems of low recommendation accuracy and only consider user rating behavior, so as to improve service quality, experience and feeling. , the effect of improving accuracy and efficiency
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0035] The present invention provides a personalized recommendation method based on user behavior, see figure 1 , the method includes the following steps:
[0036] 101: Model the user's attention behavior, and obtain the user's attention behavior matrix;
[0037] Through the establishment and extraction of the user model, the user's attention behavior matrix can be obtained, that is, an e×f matrix M is established, e is the trustee user, f is the trustee user, and the corresponding M ij Indicates the attention behavior of user i to user j.
[0038] 102: Introduce the user's rating of the project and the attention behavior among users into the calculation of the homogeneity measure, and use the Pearson coefficient to obtain the improved homogeneity measure;
[0039] Among them, the user-based personalized recommendation algorithm introduces the user's attention behavior into the calculation of the homogeneity measure, introduces the user's attention behavior into the Pearson ...
Embodiment 2
[0046] The scheme in embodiment 1 is described in detail below in conjunction with specific calculation formulas and examples, see the following description for details:
[0047] 201: Generate a user-to-item rating matrix R through user modeling;
[0048] In the process of personalized recommendation, user modeling must be carried out first, and the rating relationship and rating value of the user to the item are used in this process. Generate user-item rating matrix R through user modeling. Among them, R is an n×d scoring matrix, n is the number of users, d is the number of items, and the corresponding r is Indicates user i's rating on item s, and the rating value can be a binary attribute value or a user-defined rating level. Since the number of items is very large, users usually only rate a small number of items, which will cause the generated scoring matrix R to be very sparse. If such scoring matrix is directly calculated, it will bring huge overhead to the system. ...
Embodiment 3
[0077] The following combined with specific examples, figure 2 with image 3 The feasibility of the schemes in Examples 1 and 2 is verified, that is, through comparative experiments with existing homogeneity measurement methods, the improvement in accuracy and efficiency of this method is verified, as described below for details:
[0078] Conduct experiments on the values of the influencing factors α and β, and select the composition scheme of the homogeneity coefficient that can achieve the best accuracy of trust prediction. The experimental results are as follows: figure 2 shown.
[0079] It can be seen from the experimental results that when α is 0.3 and β is 0.9, the constructed homogeneity coefficient makes the trust prediction accuracy reach the optimal value. It can be seen that the change of the value of α does not have a great impact on the accuracy rate, so it can be seen that the importance of user comment behavior on the homogeneity measurement between users ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More