Personalized Recommendation Method and System Based on Crowd Sensing in Cloud Edge

Abstract

The recommendation system is based on the recommendation technology and data mining
theory. According to the user's interest characteristics, the user may mine the resources that may
be of interest or need from the massive information, and make corresponding recommendations
to the user. The traditional collaborative filtering recommendation technology has certain
deficiencies in the big data environment. The purpose of this article is to study the problem of
personalized recommendation method and system based on crowd sensing in the cloud edge
environment, improve the traditional collaborative filtering algorithm based on crowd sensing,
and at the same time, after the Map Reduce of classic Item CF algorithm., A parallel
recommendation engine based on the Hadoop open source framework was established, and the
effectiveness of the system was verified through the recommendation work on the online
education platform. The research results show that: 22000 data are calculated on 2 slave nodes,
67000 data are calculated on 4 slave nodes, and 170,000 data is calculated on 6 slave nodes, and
the parallel efficiency E is maintained at about 0.78. This shows that when the data size
increases, you can always compensate for the performance loss by adding computing nodes, that
is, the Item-CF-MR algorithm has good scalability. Finally, an experiment is performed on the
data set, and comparison of the experimental results shows that the improved personalized
recommendation method and system recommendation result based on crowd sensing in the
cloud edge environment has higher accuracy.