Log content audit optimization method and device

Abstract

The invention provides a method and a device for auditing and optimizing log contents. The method comprises the following steps: abstracting multiple groups of input vectors respectively corresponding to contents under multiple dimensions from the contents of the log under the multiple dimensions; calculating to obtain multiple output vectors corresponding to the multiple groups of input vectors according to the multiple groups of input vectors and a preset matrix matched with the dimension corresponding to each group of input vectors; and inputting an expected output vector, training the expected output vector by a neural network, and correcting the preset matrix matched with the corresponding dimension of the output vector according to the difference of the output vector and the expected output vector corresponding to the output vector. The method provided by the invention is used for realizing an auditing and training mechanism by the neural network technology, so as to achieve the purpose that the audit is more and more accurate and the labor cost is lower and lower; and thus, the workload of auditors is not only greatly reduced, moreover, the audit efficiency and accuracy are greatly improved, and the audit cost and the storage cost are saved.

Description

technical field [0001] The present invention relates to a log content audit optimization method and device, in particular to a log content audit optimization method and device applied to a monitoring platform. Background technique [0002] Log analysis technology is widely used in many fields such as monitoring platform and user behavior analysis. However, the log is not a rigorous and reliable data source, it is just a loose and variable text string. As far as the string splicing code has no constraints, in this loose scenario, there may be many reasons why the log parsing may not meet expectations, for example: the agreed format has been modified; input in the log wrong delimiter; one of the codes is logically shared by other codes, and other types of logs are output mixedly; the encoding configuration of a certain link from the code to the log is abnormal, resulting in Chinese garbled characters, etc. However, currently, for loose and random String concatenation, hard to...

AI Technical Summary

Problems solved by technology

It is not difficult to see that for dimension values ​​of this scale, the method of manual review requires high human resources and costs, and cannot guarantee its review effect. At the same time, a large amount of illegal data will waste a lot of computing and storage resources. At the same time Also seriously affects user experience and data reliability

Images