Method and device for determining data anomaly

Patent metadata

Field	Value
Publication number	US-2020213341-A1
Application number	US-202016810961-A
Country	US
Kind code	A1
Filing date	Mar 6, 2020
Priority date	Dec 29, 2017
Publication date	Jul 2, 2020
Grant date	—

Abstract

Computer-implemented methods, non-transitory, computer-readable media, and computer-implemented systems for determination of anomalous data are provided. In a computer-implemented method, a plurality of data packets is received within a predetermined time period, the plurality of data packets comprising a data structure. A historical distribution of historical data including the data structure as the data packets is determined. The plurality of data packets is compared to the historical distribution to generate a comparison result. If it is determined that data anomaly exists in the plurality of data packets according to the comparison result, an alert indicating the data anomaly is generated.

First claim

What is claimed is:

1 . A computer-implemented method for determination of anomalous data, comprising:
retrieving, by one or more processors, a plurality of data packets within a predetermined time period, the plurality of data packets comprising a data structure; determining, by the one or more processors, a historical distribution, wherein the historical distribution is of historical data comprising the data structure as the data packets; comparing, by the one or more processors, the plurality of data packets with the historical distribution to generate a comparison result; determining, by the one or more processors, that a data anomaly exists in the plurality of data packets according to the comparison result; and in response to determining that the data anomaly exists in the plurality of data packets, generating, by the one or more processors, an alert indicating the data anomaly.

2 . The computer-implemented method of claim 1 , wherein comparing the plurality of data packets with the historical distribution comprises:
generating a plurality of distribution state parameters by substituting the plurality of data packets into the historical distribution; and comparing the plurality of distribution state parameters with a predetermined threshold related to a distribution state to determine a number of data packets having a distribution state parameter exceeding the predetermined threshold.

3 . The computer-implemented method of claim 1 , wherein comparing the plurality of data packets with the historical distribution comprises:
determining a data distribution state of the plurality of data packets as a current distribution; and comparing the current distribution with the historical distribution.

4 . The computer-implemented method of claim 3 , wherein comparing the current distribution with the historical distribution comprises:
determining a distribution center of the current distribution; determining a distribution center of the historical distribution; determining an offset between the distribution center of the current distribution and the distribution center of the historical distribution; determining that the offset between the distribution center of the current distribution and the distribution center of the historical distribution exceeds a predetermined offset threshold; and in response to determining that the offset exceeds the predetermined offset threshold, determining that the data anomaly exists in the plurality of data packets.

5 . The computer-implemented method of claim 3 , wherein comparing the current distribution with the historical distribution comprises:
determining a first distribution state parameter of a randomly selected data packet in the current distribution; determining a second distribution state parameter of the randomly selected data packet in the historical distribution; determining a difference value between the first distribution state parameter and the second distribution state parameter; determining that the difference value exceeds a predetermined difference threshold; and in response to determining that the difference value exceeds the predetermined difference threshold, determining that the data anomaly exists in the plurality of data packets.

6 . The computer-implemented method of claim 1 , wherein the historical distribution comprises a historical probability distribution determined by processing the historical data with a hybrid Gaussian model.

7 . The computer-implemented method of claim 1 , wherein the historical distribution comprises a historical clustering distribution determined by processing the historical data with a clustering algorithm.

8 . A non-transitory, computer-readable medium storing one or more instructions executable by a computer system to perform one or more operations for determination of anomalous data, comprising:
retrieving a plurality of data packets within a predetermined time period, the plurality of data packets comprising a data structure; determining a historical distribution, wherein the historical distribution is of historical data comprising the data structure as the data packets; comparing the plurality of data packets with the historical distribution to generate a comparison result; determining that a data anomaly exists in the plurality of data packets according to the comparison result; and in response to determining that the data anomaly exists in the plurality of data packets, generating an alert indicating the data anomaly.

9 . The non-transitory, computer-readable medium of claim 8 , wherein comparing the plurality of data packets with the historical distribution comprises:
generating a plurality of distribution state parameters by substituting the plurality of data packets into the historical distribution; and comparing the plurality of distribution state parameters with a predetermined threshold related to a distribution state to determine a number of data packets having a distribution state parameter exceeding the predetermined threshold.

10 . The non-transitory, computer-readable medium of claim 8 , wherein comparing the plurality of data packets with the historical distribution comprises:
determining a data distribution state of the plurality of data packets as a current distribution; and comparing the current distribution with the historical distribution.

11 . The non-transitory, computer-readable medium of claim 10 , wherein comparing the current distribution with the historical distribution comprises:
determining a distribution center of the current distribution; determining a distribution center of the historical distribution; determining an offset between the distribution center of the current distribution and the distribution center of the historical distribution; determining that the offset between the distribution center of the current distribution and the distribution center of the historical distribution exceeds a predetermined offset threshold; and in response to determining that the offset exceeds the predetermined offset threshold, determining that the data anomaly exists in the plurality of data packets.

12 . The non-transitory, computer-readable medium of claim 10 , wherein comparing the current distribution with the historical distribution comprises:
determining a first distribution state parameter of a randomly selected data packet in the current distribution; determining a second distribution state parameter of the randomly selected data packet in the historical distribution; determining a difference value between the first distribution state parameter and the second distribution state parameter; determining that the difference value exceeds a predetermined difference threshold; and in response to determining that the difference value exceeds the predetermined difference threshold, determining that the data anomaly exists in the plurality of data packets.

13 . The non-transitory, computer-readable medium of claim 8 , wherein the historical distribution comprises a historical probability distribution determined by processing the historical data with a hybrid Gaussian model.

14 . The non-transitory, computer-readable medium of claim 8 , wherein the historical distribution comprises a historical clustering distribution determined by processing the historical data with a clustering algorithm.

15 . A computer-implemented system for determination of anomalous data, comprising:
one or more computers; and one or more computer memory devices interoperably coupled with the one or more computers and having tangible, non-transitory, mac

Assignees

• Alibaba Group Holding Ltd

Inventors

• Wu Yuehao
• Xie Ling
• Xu Xin

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F18/23

  Clustering techniques · CPC title

• G06F16/90335

  Query processing · CPC title

• H04L43/16

  Threshold monitoring · CPC title

• H04L41/142 Primary

  using statistical or mathematical methods · CPC title