Data Fusion Sorting Method Based on Social Network Recommendation Platform

Abstract

The invention discloses a data fusion sorting method based on a social network recommendation platform, which mainly solves the problems of recommendation accuracy rate and recommendation recall rate of data fusion sorting. The scheme includes: 1) collecting data; 2) classifying the data after data cleaning and data standardization to establish a project document set; 3) obtaining the feature vector of the project document in the training sample set; 4) obtaining the feature of the project document in the test sample set 5) Construct neural network feature fusion scoring model; 6) Construct ranking model. The invention considers feature information of various aspects of the project, constructs a neural network feature fusion scoring model, can be used for project recommendation on a social network recommendation platform, and improves data fusion speed, recommendation accuracy rate and recommendation recall rate.

Description

technical field [0001] The invention belongs to the field of computer technology, and further relates to a data fusion and sorting method based on a social network recommendation platform in the field of network application technology. According to the accuracy rate, recall rate, and speed requirements of the target user for data fusion and sorting recommendation, the present invention uses different feature calculation methods to process each feature of the data separately, obtain the feature vector of each data, and input the feature vector of each data The feature fusion scoring model constructed by the neural network obtains the score of each data, sorts the scores of each data, and obtains the ranking results to recommend to the target user. Background technique [0002] Data fusion and sorting technology refers to the use of computers to extract features from document observation data, and automatically coordinate and fuse them according to the importance of each featu...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the unified evaluation index cannot comprehensively evaluate the importance of all the features of the data, resulting in poor universality of the data evaluation index, so the accuracy and recall rate of the recommendation after the data fusion and sorting are low. Suitable for social network recommendation platform

The disadvantage of this method is that the generalization ability of the fusion ranking model adopted is low, and an efficient feature set cannot be quickly derived from the training set.

When the amount of data reaches a saturated state, more data cannot improve its performance, so it is not suitable for social network recommendation platforms with huge data

Images