CN111245738B

CN111245738B - Flow dyeing method

Info

Publication number: CN111245738B
Application number: CN202010018573.2A
Authority: CN
Inventors: 蔺昊; 钟远君
Original assignee: Shenzhen Inveno Technology Co ltd
Current assignee: Shenzhen Inveno Technology Co ltd
Priority date: 2020-01-08
Filing date: 2020-01-08
Publication date: 2023-05-26
Anticipated expiration: 2040-01-08
Also published as: CN111245738A

Abstract

The flow dyeing method provided by the invention comprises the following steps: dividing the flow to be processed into a plurality of layers according to a preset layering rule; acquiring a characteristic value in the flow to be processed; in each level, the flow to be treated is dyed in combination with the characteristic value to obtain an experiment number. According to the flow dyeing method, the flow is layered by using a multi-layer mode, the flow can be split according to the characteristics of the user, the number of parallel experiments can be increased, and more detailed experiments are carried out on different types of users.

Description

Flow dyeing method

Technical Field

The invention belongs to the technical field of big data processing, and particularly relates to a flow dyeing method.

Background

The flow dyeing means that corresponding flow dyeing rules are set according to a flow protocol, the designated flow is dyed and marked, and the mark is carried in the whole call chain. Specific traffic can be tracked and routed through the dye traffic.

The flow dyeing is widely applied to an AB TEST experiment, two schemes are established for one optimization target in the AB TEST experiment, one part of users use the scheme A, the other part of users use the scheme B, then user data generated by the two schemes are collected, and the advantages and disadvantages of the two schemes are judged according to the user data, so that the selection of the schemes is decided, and the conversion rate is improved. In the above process, flow staining is required, see fig. 1 and 2, i.e. different flows (or different users) are stained with different experiment numbers. And then the service user obtains corresponding experimental configurations according to different returned experimental numbers, and finally evaluates which experimental configuration has the best effect according to the effects and indexes of the different experimental configurations.

However, in the existing AB TEST experiment, sufficient flow is required to be ensured for two groups of experiments, if the experiment is too many at the same time, the flow of each group of experiments is reduced, so that the credibility of experimental effect evaluation is reduced. In the prior art, all flow is regarded as one layer, so that the number of experiments performed simultaneously is reduced. In addition, the prior art can only split according to some simple flow characteristics (such as model and application version), but can not split according to the characteristics of users.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a flow dyeing method which uses a multi-layer mode to layer the flow and can also split the flow according to the user characteristics.

A flow dyeing method comprising the steps of:

dividing the flow to be processed into a plurality of layers according to a preset layering rule;

acquiring a characteristic value in the flow to be processed;

in each level, the flow to be treated is dyed in combination with the characteristic value to obtain an experiment number.

Preferably, the layering rule specifically includes:

classifying the flow to be treated into four levels: a UI layer, an algorithm layer, a content layer, and an advertisement layer.

Preferably, the obtaining the experiment number specifically includes:

and respectively obtaining different experiment numbers output by the four layers.

Preferably, the layering rules further classify the traffic to be processed into independent traffic layers.

Preferably, the obtaining the experiment number specifically includes:

and obtaining four identical experiment numbers output by the independent flow layers.

Preferably, in each level, dyeing the flow to be treated in combination with the characteristic value to obtain an experiment number specifically includes:

setting a plurality of dyeing characteristic values and corresponding experiment numbers in each layer;

when the characteristic value of the flow to be processed is obtained, identifying the dyeing characteristic value which is consistent with the characteristic value in the level, and outputting the experiment number corresponding to the dyeing characteristic value as the dyeing result of the flow to be processed in the level.

Preferably, after the flow to be processed is divided into a plurality of levels according to a preset layering rule, the method further comprises:

and carrying out hash operation on each level, so that the flow to be processed can be shunted to different experiment numbers for dyeing according to the preset duty ratio dyeing configuration.

Preferably, the hash operation is performed on each level, so that the flow to be processed can be shunted to different experiment numbers for dyeing according to a preset duty ratio dyeing rule, which specifically includes:

setting a plurality of experiment numbers and a dyeing range containing a plurality of numerical values in each layer;

in each level, assigning all values in a dyeing range to different experiment numbers according to the duty ratio dyeing configuration so as to obtain a dyeing value of each experiment number;

when the level receives the flow to be processed, a random algorithm is used for outputting a random number;

calculating the random number so that the calculated random number falls into the dyeing range of the hierarchy;

and identifying a dyeing numerical value which is consistent with the calculated random number, and outputting an experiment number corresponding to the dyeing numerical value as a dyeing result of the flow to be treated at the level.

Preferably, in each level, assigning all the values in the dyeing range to different experiment numbers according to the duty dyeing configuration, so as to obtain the dyeing value of each experiment number specifically includes:

obtaining the dyeing duty ratio of each experiment number according to the duty ratio dyeing configuration;

and sequentially and continuously distributing the numerical values in the dyeing range to different experiment numbers according to the dyeing duty ratio of the different experiment numbers so as to obtain the dyeing numerical value of each experiment number.

According to the technical scheme, the flow dyeing method provided by the invention adopts a multi-layer mode to layer the flow, can split the flow according to the characteristics of the user, can increase the number of parallel experiments, and can carry out more detailed experiments on different types of users.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. Like elements or portions are generally identified by like reference numerals throughout the several figures. In the drawings, elements or portions thereof are not necessarily drawn to scale.

Fig. 1 is a data interaction diagram of a conventional dyeing method provided in the background art.

Fig. 2 is a flowchart of a conventional dyeing method provided in the background art.

Fig. 3 is a flowchart of a flow dyeing method according to a first embodiment of the present invention.

Fig. 4 is a schematic diagram of a flow hierarchy according to a first embodiment of the present invention.

Fig. 5 is a flowchart of a method for splitting according to user characteristics according to a second embodiment of the present invention.

Fig. 6 is a flow chart of a method for duty dyeing according to a second embodiment of the present invention.

Fig. 7 is a data interaction diagram of an experiment number application provided in a third embodiment of the present invention.

FIG. 8 is a schematic diagram of an external module for experiment number application according to a third embodiment of the present invention

Detailed Description

Embodiments of the technical scheme of the present invention will be described in detail below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and thus are merely examples, and are not intended to limit the scope of the present invention. It is noted that unless otherwise indicated, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

It should be understood that the terms "comprises" and "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

As used in this specification and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".

Embodiment one:

a flow staining method, see fig. 3, comprising the steps of:

acquiring a characteristic value in the flow to be processed;

Specifically, in order to solve the problems that single-layer dyeing in the prior art is easy to cause experiment increase and flow reduction, the flow is divided into a plurality of layers, and the flow is layered in a multi-layer mode. Meanwhile, the dyeing can be combined with the user characteristics, and the number of parallel experiments can be increased by shunting according to the user characteristics, so that more detailed experiments are carried out on different types of users. The method can be implemented in a web page or SaaS.

Referring to fig. 4, the method may classify traffic into four levels: a UI layer, an algorithm layer, a content layer, and an advertisement layer. Or to categorize traffic into separate traffic layers. When classifying the flow into four levels, the method outputs four different experiment numbers after dyeing the flow. When the flow is classified into an independent flow layer, the method outputs four identical experiment numbers after dyeing the flow. Wherein the dyeing process is the same for each level. The hierarchy of traffic layering is determined based on the business scope of the user and the customer base.

Embodiment two:

example two dyeing methods are provided on the basis of example one.

1. And (5) characteristic dyeing.

In each level, dyeing the flow to be processed by combining the characteristic values to obtain experiment numbers specifically comprises the following steps:

Specifically, the method may dye according to a characteristic value generated during use, for example, a user characteristic (for example, a user click, a user request) generated during use by a user, a region characteristic, and the like. For example, an experiment number corresponding to a dyeing characteristic value a (using an algorithm a) is 1001 and an experiment number corresponding to a dyeing characteristic value B (using an algorithm B) is 1002 in the algorithm layer. When the flow A to be processed is received, the characteristic value of the flow A to be processed is identified as using the algorithm A, and the output experiment number is 1001 after the flow A to be processed is dyed by the algorithm layer.

For example, the UI layer is also set with a test number 2001 corresponding to the dyeing characteristic value a (UI using red) and a test number 2002 corresponding to the dyeing characteristic value B (UI using black). When the flow A to be processed is received, the characteristic value of the flow A to be processed is identified as a UI using black, and the output experiment number is 2001 after the flow A to be processed is dyed by the UI layer. It follows that the same flow to be treated can be dyed through a plurality of levels.

Fig. 5 provides a method for offloading according to user characteristics, in which when receiving a traffic to be processed with a user flag, user data is acquired through redis, and then the traffic to be processed with user data is calculated through user characteristic configuration, so as to obtain user characteristics. And finally judging whether the user characteristics are matched with the dyeing characteristic values, and if so, outputting corresponding experiment numbers. In the matching process, the dyeing matching method is divided into independent flow layer dyeing matching and other level dyeing matching.

2. Duty cycle dyeing.

Referring to fig. 6, after the flow to be processed is divided into a plurality of levels according to a preset layering rule, the method further includes:

Specifically, the duty dyeing configuration is used to set the dyeing duty of each experiment number, for example, when the flow rate duty of experiment number a is set to 30% and the flow rate duty of experiment number B is set to 70%, that is, 30% of the flow rate is assigned to experiment number a and 70% of the flow rate is assigned to experiment number B. The user can set the dyeing duty ratio of each experiment number according to the self requirement, so that the method can split the flow according to the duty ratio dyeing configuration. According to the method, hash operation is carried out on each level according to the duty ratio dyeing configuration, so that the flow is shunted to different experiment numbers according to the duty ratio dyeing configuration. When hash operation is performed, an independent traffic layer and other layers are required to be separated for operation.

a: setting a plurality of experiment numbers and a dyeing range containing a plurality of numerical values in each layer;

b: in each level, assigning all values in a dyeing range to different experiment numbers according to the duty ratio dyeing configuration so as to obtain a dyeing value of each experiment number; the method specifically comprises the following steps:

Specifically, the method provides a simple diversion method, the dyeing range is set, the numerical values in the dyeing range are distributed to different experiment numbers, and the dyeing duty ratio of the experiment numbers is controlled by distributing different numbers of numerical values to the different experiment numbers. For example: assuming 1000 values in the dyeing range, 300 values are assigned to experiment number a and the remaining 700 values are assigned to experiment number B, then it is indicated that experiment number a needs to be divided into 30% and experiment number B needs to be divided into 70%. The embodiment also provides a more convenient distribution method, which sequentially and continuously distributes the numerical values in the dyeing range to different experiment numbers according to the dyeing duty ratio of the different experiment numbers. For example, the first 300 of 1000 values in the dyeing range are assigned to experiment number A, and the second 700 are assigned to experiment number B, so that the sequential assignment is easy to manage and difficult to confuse.

C: when the level receives the flow to be processed, a random algorithm is used for outputting a random number;

in particular, when the stochastic algorithm is sufficiently stochastic, the effect of the final split is more likely to be close to the duty cycle dyeing configuration. For example, the sha1 algorithm may be used to obtain a 32-bit 16-ary number, which is then left over by 1000 to obtain a value of 0-999.

D: calculating the random number so that the calculated random number falls into the dyeing range of the hierarchy;

e: and identifying a dyeing numerical value which is consistent with the calculated random number, and outputting an experiment number corresponding to the dyeing numerical value as a dyeing result of the flow to be treated at the level.

Specifically, for example, assuming that when a flow to be processed is received, a sha1 algorithm is used to obtain a 16-bit number of 32 bits, then the number is obtained by taking the remainder of 1000 to obtain 246, and the experiment number corresponding to the number is experiment number a, then the experiment number output by the flow to be processed is experiment number a. If the random number is obtained after the remainder 1000, and the experiment number corresponding to the value is experiment number B, the experiment number of the flow output to be processed is experiment number B.

For a brief description of the method provided in the embodiments of the present invention, reference may be made to the corresponding content in the foregoing method embodiments where the description of the embodiments is not mentioned.

Embodiment III:

embodiment III the application of the experiment number is increased on the basis of the embodiment.

Fig. 7 illustrates, by way of example, a news client, an application of the above-described dyeing result, comprising the steps of:

step 1: the user opens the news client for the first time.

Step 2: the news client requests information (with user id, model, application version, etc. user characteristics) from the information server.

Step 3: the information server requests a dyeing service (with user features).

Step 4: the staining service returns the experiment number according to the user characteristics (4, this step has not been able to perform user characteristic staining due to initial use).

Step 5: the information server determines the dyeing configuration used by the user (for example, using the recommendation algorithm a or UI presentation using UI a) according to the obtained experiment number, and then returns the corresponding content to the user.

Step 6: the user uses the user data (such as the number of clicks and the number of impressions) generated after the application to report to the log service, and then the log service processes the user data offline to obtain the user day level data (such as daily request, impressions, clicks and numbers) and stores the user day level data to redis.

Step 7: and when the user uses the application again, repeating the steps 1-3. In flow 4, the staining service may perform finer staining (i.e., user feature staining) based on the user day data.

Step 8: repeating the steps 1-6.

Fig. 8 provides an external connection module of a dyeing service, including:

an Apollo module. Dyeing center. The experiment numbers for configuring each layer and the corresponding dyeing configuration (e.g., UI layer has 2 experiment numbers, 1001, 1002, algorithm layer has 3 experiment numbers, 2001, 2002, 2003, where 1001 requires 30% traffic, 1002 requires 70% traffic, 2001 needs to be new users, 2002 requires 10% traffic, 2003 requires 90% traffic).

And the Redis module is used for storing the user day level data obtained in the figure 7.

Eureka module, registry. The dyeing service is registered in the service, and other services (for example, information service) of the registry can know the address of the dyeing service, so that a request can be initiated to the dyeing service.

The falcon-agent module is used for collecting various monitoring indexes of the dyeing service, such as the number of received requests, the time consumption of each request, the number of abnormal service occurrence and the like.

And the alarm module is used for sending alarm information to related personnel when a certain index collected by the falcon-agent module exceeds a threshold value (for example, the time is more than 10 ms).

And the Grafana (monitoring UI) module displays all indexes collected by the falcon-agent module in real time through a graphical interface.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention, and are intended to be included within the scope of the appended claims and description.

Claims

1. A flow dyeing method, characterized by comprising the steps of:

acquiring a characteristic value in the flow to be processed;

in each level, dyeing the flow to be processed by combining the characteristic values to obtain experiment numbers;

the method further comprises the following steps after dividing the flow to be processed into a plurality of layers according to a preset layering rule:

carrying out hash operation on each level, so that the flow to be processed can be shunted to different experiment numbers for dyeing according to a preset duty ratio dyeing configuration;

the hash operation is performed on each level, so that the flow to be processed can be shunted into different experiment numbers for dyeing according to a preset duty ratio dyeing rule, and the steps specifically include:

identifying a dyeing numerical value which is consistent with the calculated random number, and outputting an experiment number corresponding to the dyeing numerical value as a dyeing result of the flow to be processed at the level;

in each level, assigning all values in the dyeing range to different experiment numbers according to the duty ratio dyeing configuration, so as to obtain the dyeing value of each experiment number specifically includes:

2. The flow dyeing method according to claim 1, wherein,

the layering rule specifically comprises:

3. A flow dyeing method according to claim 2, characterized in that,

the obtaining experiment numbers specifically comprises the following steps:

4. A flow dyeing method according to claim 2, characterized in that,

the layering rules also classify the traffic to be processed into independent traffic layers.

5. The flow dyeing method according to claim 4, wherein,

the obtaining experiment numbers specifically comprises the following steps:

6. The flow staining method according to any one of claims 1 to 5 wherein, in each level, staining the flow to be treated in combination with the characteristic value to obtain an experiment number specifically comprises: