Network anomaly data detection method and device as well as computer equipment and storage medium

1. A network anomaly data detection method, comprising:

receiving access request data transmitted by a client;

searching historical access request data corresponding to a user session identifier in the access request data;

acquiring a header character string of the access request data;

performing word segmentation processing on the header character string according to a preset step length so as to obtain a word segmentation set;

obtaining a word segmentation weight matrix according to the historical access request data and the word segmentation set;

inputting the word segmentation weight matrix into an anomaly data detection model so as to obtain a data anomaly probability;

judging whether anomaly data exists in the header character string according to the data anomaly probability; and

correcting an initial anomaly data detection model in response.

2. The method according to claim 1 , after receiving the access request data transmitted by the client, further comprising:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier comprises a request session identifier; and

generating a new request session identifier when the user fingerprint identifier is inconsistent with the user fingerprint identifier of last received access request data, and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

3. The method according to claim 1 , after receiving the access request transmitted by the client, further comprising:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier comprises a request session identifier;

acquiring a session length of the current session when the user fingerprint identifier is inconsistent with the user fingerprint identifier of last access request; and

executing a step of searching historical access request data corresponding to the user session identifier in the access request data when the session length does not exceed a preset session length threshold.

4. The method according to claim 3 , further comprising:

generating a new request session identifier when the session length exceeds the preset session length threshold; and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

5. The method according to claim 1 , wherein the obtaining a word segmentation weight matrix according to the historical access request data and the word segmentation set comprises:

calculating a word frequency of each segmented word in the header character string in the word segmentation set;

calculating an inverse document frequency of each segmented word in the historical access request data in the word segmentation set;

calculating to obtain a word weight of each segmented word according to the calculated word frequency of each segmented word and the inverse document frequency; and

generating a word segmentation weight matrix of the word segmentation set according to the word weight of each segmented word.

6. The method according to claim 1 , further comprising:

acquiring expected values of an access request data sample set and each access request data sample in the access request data sample set;

calculating a word segmentation weight matrix of each access request data sample;

inputting the expected value of each access request data sample and the word segmentation weight matrix into the initial anomaly data detection model for training and obtaining an optimal regression coefficient matrix; and

correcting the initial anomaly data detection model according to the optimal regression coefficient matrix so as to obtain an anomaly data detection model.

7. A computer equipment, comprising a memory and one or more processors, wherein computer readable instructions are stored in the memory; and when the computer readable instructions are executed by the one or more processors, the one or more processors execute the following steps:

receiving access request data transmitted by a client;

searching historical access request data corresponding to a user session identifier in the access request data;

acquiring a header character string of the access request data;

performing word segmentation processing on the header character string according to a preset step length so as to obtain a word segmentation set;

obtaining a word segmentation weight matrix according to the historical access request data and the word segmentation set;

inputting the word segmentation weight matrix into an anomaly data detection model so as to obtain a data anomaly probability;

judging whether anomaly data exists in the header character string according to the data anomaly probability; and

correcting an initial anomaly data detection model in response.

8. The computer equipment according to claim 7 , wherein while executing the computer readable instructions, the processors further execute the following steps:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier includes a request session identifier; and

generating a new request session identifier when the user fingerprint identifier is inconsistent with the user fingerprint identifier of the last received access request data, and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

9. The computer equipment according to claim 7 , wherein while executing the computer readable instructions, the processors further execute the following steps:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier comprises a request session identifier;

acquiring a session length of the current session when the user fingerprint identifier is inconsistent with the user fingerprint identifier of last access request; and

executing a step of searching historical access request data corresponding to the user session identifier in the access request data when the session length does not exceed a preset session length threshold.

10. The computer equipment according to claim 9 , wherein while executing the computer readable instructions, the processors further execute the following steps:

generating a new request session identifier when the session length exceeds the preset session length threshold; and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

11. The computer equipment according to claim 7 , wherein while executing the computer readable instructions, the processors realize the step of obtaining the word segmentation weight matrix according to the historical access request data and the word segmentation set, further comprising:

calculating a word frequency of each segmented word in the header character string in the word segmentation set;

calculating an inverse document frequency of each segmented word in the historical access request data in the word segmentation set;

calculating to obtain a word weight of each segmented word according to the calculated word frequency of each segmented word and the inverse document frequency; and

generating a word segmentation weight matrix of the word segmentation set according to the word weight of each segmented word.

12. The computer equipment according to claim 7 , wherein while executing the computer readable instructions, the processors further execute the following steps:

acquiring expected values of an access request data sample set and each access request data sample in the access request data sample set;

calculating a word segmentation weight matrix of each access request data sample;

inputting the expected value of each access request data sample and the word segmentation weight matrix into the initial anomaly data detection model for training and obtaining an optimal regression coefficient matrix; and

correcting the initial anomaly data detection model according to the optimal regression coefficient matrix so as to obtain an anomaly data detection model.

13. One or more nonvolatile computer readable storage media stored with the computer readable instructions, wherein when the computer readable instructions are executed by one or more processors, the one or more processors execute the following steps:

receiving access request data transmitted by a client;

searching historical access request data corresponding to a user session identifier in the access request data;

acquiring a header character string of the access request data;

performing word segmentation processing on the header character string according to a preset step length so as to obtain a word segmentation set;

obtaining a word segmentation weight matrix according to the historical access request data and the word segmentation set;

inputting the word segmentation weight matrix into an anomaly data detection model so as to obtain a data anomaly probability;

judging whether anomaly data exists in the header character string according to the data anomaly probability; and

correcting an initial anomaly data detection model in response.

14. The storage media according to claim 13 , wherein when executed by the processors, the computer readable instructions also execute the following steps:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier includes a request session identifier; and

generating a new request session identifier when the user fingerprint identifier is inconsistent with the user fingerprint identifier of the last received access request data, and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

15. The storage media according to claim 13 , wherein when executed by the processors, the computer readable instructions also execute the following steps:

reading a user session identifier in the access request data;

comparing a user fingerprint identifier in the user session identifier with a user fingerprint identifier of last received access request data when the user session identifier comprises a request session identifier;

acquiring a session length of the current session when the user fingerprint identifier is inconsistent with the user fingerprint identifier of last access request; and

executing a step of searching historical access request data corresponding to the user session identifier in the access request data when the session length does not exceed a preset session length threshold.

16. The storage media according to claim 14 , wherein when executed by the processors, the computer readable instructions also execute the following steps:

generating a new request session identifier when the session length exceeds the preset session length threshold; and transmitting the new request session identifier to the client when response data corresponding to the access request data is returned to the client.

17. The storage media according to claim 13 , wherein while realized by the processors to realize the step of obtaining the word segmentation weight matrix according to the historical access request data and the word segmentation set, the computer readable instructions further comprising:

calculating a word frequency of each segmented word in the header character string in the word segmentation set;

calculating an inverse document frequency of each segmented word in the historical access request data in the word segmentation set;

calculating to obtain a word weight of each segmented word according to the calculated word frequency of each segmented word and the inverse document frequency; and

generating a word segmentation weight matrix of the word segmentation set according to the word weight of each segmented word.

18. The storage media according to claim 14 , wherein while executed by the processors, the computer readable instructions also execute the following steps:

acquiring expected values of an access request data sample set and each access request data sample in the access request data sample set;

calculating a word segmentation weight matrix of each access request data sample;

inputting the expected value of each access request data sample and the word segmentation weight matrix into the initial anomaly data detection model for training and obtaining an optimal regression coefficient matrix; and

correcting the initial anomaly data detection model according to the optimal regression coefficient matrix so as to obtain an anomaly data detection model.