A visual query method and device for combining data from multiple data sources

Abstract

The invention discloses a visual query method and device for merging data from multiple data sources, belonging to the field of visual query and privacy protection, comprising the following steps: 1) establishing a server between N participants; 2) according to visual requests , each participant calculates the local data D i , i∈[1, N]; 3) Exchange a random number vector between each participant, and transfer the local data D i , the received random number vector and the sent random number vector are fused; at the same time, the fusion results of the N participants are uploaded to the server; 4) the server merges the fusion results uploaded by the N participants and sends them to the server. The front end returns the combined results for visual analysis by the front end. The data of multiple clients is merged without iteration, and accurate visual query results can be obtained with only one processing, and the privacy of the data can be guaranteed not to be leaked.

Description

technical field [0001] The invention relates to the field of visual query and privacy protection, in particular to a visual query method and device for combining data from multiple data sources. Background technique [0002] In the era of big data, the emphasis on data privacy and security has become a worldwide trend. Therefore, there is a great risk in collecting sensitive data on a large scale. With the EU's implementation of the General Data Protection Regulation (GDPR) on May 25, 2018, countries are strengthening data security and privacy protection, paying more attention to users' personal privacy and data security. However, in most industries, data exists in the form of isolated islands. Due to problems such as industry competition, privacy security, and complex administrative procedures, there are many obstacles to data integration even between different departments of the same company. The cost of integrating data that is scattered across various locations and inst...

AI Technical Summary

Problems solved by technology

[0004] However, the above patents only realize the exchange of data between internal and external networks, and it is a one-to-one exchange. In the field of visual query, people often need to perform partial visual analysis from the merged data of multiple data sources, and need Relatively Accurate Query Results

Due to the existence of the above-mentioned data island problem, it is a very difficult task to merge different data sources while satisfying the accuracy of the merged results under the premise of ensuring privacy

There are many existing methods such as differential privacy and federated learning. The former adds a large amount of randomization to each data source, resulting in a sharp decline in its usability, especially for some complex queries, sometimes the randomization results will be largely concealed. Real results; the latter requires multiple iterations and repeated training, which is expensive and does not guarantee absolute accuracy

Images