Method for extracting feature string, device, network apparatus, and storage medium

A device for extracting a feature string, comprising:

• a processor and a memory, wherein the processor is configured to read and execute computer programs stored on the memory to:
• count each first session corresponding to each first five-tuple information according to the each first five-tuple information corresponding to an application, for the each first session, determine data packets having feature strings to be extracted in the first session, and determine participles of the data packets, and take each determined participle as each candidate feature string;
• determine a first-order Markov transition probability matrix containing a transition probability of each two adjacent characters according to an existing rule base and a current genetic variation coefficient;
• for the each candidate feature string, determine the transition probability of each two adjacent characters in the candidate feature string according to the first-order Markov transition probability matrix;
• and determine a transition entropy of the candidate feature string according to the transition probability of each two adjacent characters and a logarithm of the transition probability;
• for the each candidate feature string, judge whether the transition entropy of the candidate feature string is greater than a preset threshold, when the transition entropy of the candidate feature string is not greater than the preset threshold, label the candidate feature string as a feature string to be discarded, when the transition entropy of the candidate feature string is greater than the preset threshold, label the candidate feature string as a first usable feature string;
• and
• judge whether at least one first usable feature string exists as a valid feature string, when the at least one first usable feature string exists as the valid feature string, take the valid feature string as a target feature string extracted from the data packet;
• when the at least one first usable feature string does not exist as the valid feature string, reset the genetic variation coefficient, and trigger operations of:
• determining the first-order Markov transition probability matrix containing the transition probability of each two adjacent characters according to the existing rule base and the current genetic variation coefficient.