Position big data statistical publishing method and system

Abstract

The invention relates to a position big data statistical publishing method and system. The method comprises the following steps: dividing collected historical position big data into a plurality of first sets according to time units; dividing each first set into a plurality of second sets according to the release time interval; dividing the two-dimensional space region corresponding to each second set into a plurality of grids; recording whether the grids in the coverage area are merged or not according to a set grid area merging condition; taking a record whether to be merged as a division structure matrix; forming a three-dimensional space-time sequence by the division structure matrixes at the moment of the same release time, and training a deep learning model by taking the three-dimensional space-time sequence as input and taking the corresponding division structure matrixes as output; obtaining a prediction division structure according to the trained deep learning model; obtaining a noise adding release statistical value of each sub-region in the prediction division structure; and according to the publishing time sequence, publishing the noise adding publishing statistical value corresponding to each prediction division structure matrix. According to the invention, the execution efficiency of the release algorithm is improved.

Description

technical field [0001] The present invention relates to the technical field of data privacy protection, in particular to a method and system for publishing location big data statistics. Background technique [0002] The statistical release of location big data can provide users with accurate and timely traffic and road congestion information, help people plan reasonable travel time and routes, obtain high-precision LBS services, and help reduce traffic congestion and unnecessary waste of resources . However, improper release of location information and reverse analysis and reasoning can easily lead to the leakage of personal privacy such as the user's specific location, movement trajectory, living habits, health status, hobbies, and economic conditions, and even endanger property and life safety. Therefore, solving the privacy protection problem in the release and use of location big data has become the most urgent task restricting the development of location big data appli...

AI Technical Summary

Problems solved by technology

[0004] The existing two-dimensional space statistical division release method mainly has the following disadvantages: 1) It is difficult to determine the condition for dividing the stop when dividing the two-dimensional space from top to bottom, which leads to the introduction of more noise errors when the division level is too fine, and the division level is insufficient Reduced the precision of the range count query service when

2) The existing partition structure cannot balance the noise error and the uniform assumption error well, resulting in the low accuracy of the range count query for publishing location statistics; 3) In order to obtain a more reasonable partition structure, it is often necessary to scan the location big data multiple times and iterative analysis, the execution efficiency of the partition release algorithm is not high

Images