Processing method, device and system for ensuring consistency of calculated data

Abstract

The invention provides a processing method, device and system for ensuring the consistency of calculated data. The method comprises the steps of reading off-line data in an off-line database and determining delay time of off-line data synchronization delay according to storage statistic time of the off-line data; reading real-time data of current statistic time from a preset real-time flow calculation result database and reading real-time data corresponding to the delay time as compensation data of business calculation; conducting the business calculation on the basis of the off-line data, the real-time data and the compensation data. According to the processing method, device and system for ensuring the consistency of the calculated data, the problem can be solved that uncertainty and delay of data synchronization time result in data inconsistency, the calculation accuracy of recommendation ranking scores is ensured, the performance losses of the business calculation are reduced, and the calculation efficiency is improved.

Description

technical field [0001] The application belongs to the technical field of database data processing, and in particular relates to a processing method, device and system for ensuring the consistency of calculation data. Background technique [0002] In many current business systems, such as product recommendation, essence post recommendation, etc., the statistics and calculation of many businesses often need to use the combination of historical data stored offline and current real-time data to calculate the output results to ensure the accuracy of statistics and calculation results. accuracy. [0003] For example, in a product recommendation system, based on the user's product browsing history and the currently browsed product, it is possible to accurately recommend to the user the product that the user may need / is looking for. In the product recommendation system, the stored characteristic data of product browsing times usually includes both the historical viewing times hpv o...

AI Technical Summary

Problems solved by technology

[0005] In the above-mentioned post recommendation system using historical offline data and real-time data, due to the fast update frequency of posts, especially the interactive behavior of posts in recent days has a great impact on the calculation of post quality scores, if the statistics of day t are calculated In the middle of the results, the statistical data such as the t-1th day is missing, which will obviously seriously affect the accuracy of the final quality score calculation of the post

Similarly, in systems that require synchronizing historical offline data and real-time data and then combining statistical calculation output results, such as user product recommendation and hot keyword search, due to delays or uncertainties in data synchronization time, etc., often lead to calculation and processing of data. Inconsistency, lack of delayed synchronization data, the accuracy of the calculation results cannot be guaranteed, and even the data calculation cannot be completed normally

Images