Rapid dirty data detection and processing method for log collection

The invention relates to a dirty data rapid detection and processing method for log collection. Comprising the following steps: S1, collecting logs of various different network equipment types, and converging the logs to message middleware; s2, analyzing rule matching is conducted on logs in the message middleware according to analyzing rules in an analyzing library, fields are extracted if matching succeeds, and the step S3 is executed; if the matching is unsuccessful, determining the data as dirty data, and executing the step S5; s3, classifying the extracted fields according to a field classification rule in a field classification library; s4, finding the probability of the classification value through breadth-first traversal for the classified fields according to the corresponding classification value, finding the probability of the classification value of all the fields under the corresponding log in the same way, sequentially judging the probability of the classification value of each field, and if the