CN112734505B

CN112734505B - User behavior analysis method and device and electronic equipment

Info

Publication number: CN112734505B
Application number: CN202110365756.6A
Authority: CN
Inventors: 贺园
Original assignee: Beijing Qingsongchou Information Technology Co ltd
Current assignee: Beijing Easy Yikang Information Technology Co ltd
Priority date: 2021-04-06
Filing date: 2021-04-06
Publication date: 2021-07-23
Anticipated expiration: 2041-04-06
Also published as: CN112734505A

Abstract

The invention provides a method and a device for analyzing user behaviors and electronic equipment, wherein when the user behaviors are analyzed, the user behavior data are analyzed, and the statistical result of user input data is also analyzed, namely the user behaviors are analyzed from two dimensions of the user behaviors and the user input data. Furthermore, when determining the user behavior analysis result, the user behavior analysis method not only refers to the weight values of the behavior analysis features and the feature values of the data analysis features, but also refers to the weight values of the preset advertisement analysis feature combinations, namely, the weights are analyzed from multiple dimensions, and the accuracy of the user behavior analysis can be further improved.

Description

User behavior analysis method and device and electronic equipment

Technical Field

The invention relates to the technical field of computers, in particular to a user behavior analysis method and device and electronic equipment.

Background

With the continuous development of the internet industry, interface display is more and more abundant. Generally, in some areas of the page, such as the lower right corner, an advertisement interface is popped up, and a user can realize the access of the advertisement by clicking the advertisement, such as the access of insurance advertisement.

In practical application, operation behaviors of a user aiming at the advertisement, such as clicking, can be collected, then the collected user behaviors are analyzed, a user behavior analysis result is obtained, and then advertisement putting control is carried out based on the user behavior analysis result. Currently, the accuracy of user behavior analysis is low, and therefore the accuracy of advertisement putting based on a user behavior analysis result is low.

Disclosure of Invention

In view of this, the present invention provides a user behavior analysis method, a user behavior analysis device, and an electronic device, so as to solve the problem that the accuracy of user behavior analysis is low, and thus the accuracy of advertisement delivery based on the user behavior analysis result is low.

In order to solve the technical problems, the invention adopts the following technical scheme:

a user behavior analysis method, comprising:

acquiring user behavior data and user input data corresponding to a preset advertisement, and determining a statistical result of the user input data; the user input data is data input in a target interface which is jumped after the user clicks the preset advertisement;

acquiring a preset advertisement analysis feature set; the preset advertisement analysis feature set comprises at least one behavior analysis feature and at least one data analysis feature;

performing characteristic analysis on the user behavior data to obtain a characteristic value of each behavior analysis characteristic, and performing characteristic analysis on a statistical result of the user input data to obtain a characteristic value of each data analysis characteristic;

determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determination rule corresponding to the preset advertisement analysis feature set, wherein the preset advertisement analysis feature combination comprises at least one behavior analysis feature and/or at least one data analysis feature;

and determining a user behavior analysis result based on the weight value corresponding to each characteristic value in the preset advertisement analysis characteristic set and the weight value corresponding to the preset advertisement analysis characteristic combination.

Optionally, after determining a user behavior analysis result based on the weight value corresponding to each feature value in the preset advertisement analysis feature set and the weight value corresponding to the preset advertisement analysis feature combination, the method further includes:

determining a channel identifier corresponding to the user behavior analysis result;

acquiring a user behavior analysis result corresponding to the same channel identifier;

and determining a channel operation behavior analysis result corresponding to the channel identification based on a user behavior analysis result corresponding to the same channel identification.

Optionally, determining a statistical result of the user input data comprises:

acquiring historical user input data corresponding to the user input data in a preset historical time period;

acquiring a plurality of preset indexes and calculation rules of the indexes;

according to the calculation rule of each index, calculating the user input data and the historical user input data to obtain the index value of each index;

and determining the index value of each index as a statistical result of the user input data.

Optionally, determining, according to a feature weight determination rule corresponding to the preset advertisement analysis feature set, a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination includes:

acquiring a characteristic weight determination rule corresponding to the preset advertisement analysis characteristic set;

determining intervals where each feature value in the preset advertisement analysis feature set is located based on the feature weight determination rule, and determining a weight value corresponding to the intervals as a weight value corresponding to the feature value;

acquiring a preset advertisement analysis feature combination; the preset advertisement analysis feature combination comprises at least one behavior analysis feature and/or at least one data analysis feature;

acquiring a weight value corresponding to a characteristic value of the behavior analysis characteristic and/or the data analysis characteristic in the preset advertisement analysis characteristic combination;

and determining a weight value corresponding to the preset advertisement analysis feature combination based on the weight value corresponding to the characteristic value of the behavior analysis feature and/or the data analysis feature in the preset advertisement analysis feature combination.

Optionally, determining a user behavior analysis result based on a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to the preset advertisement analysis feature combination, including:

calculating the sum of the weighted values of all the characteristic values in the preset advertisement analysis characteristic set and the weighted value of a preset advertisement analysis characteristic combination;

determining a weight value corresponding to the sum;

and determining a weight interval corresponding to the weight value, and taking a grade mark corresponding to the weight interval as a user behavior analysis result.

A user behavior analysis device, comprising:

the statistical result determining module is used for acquiring user behavior data and user input data corresponding to preset advertisements and determining statistical results of the user input data; the user input data is data input in a target interface which is jumped after the user clicks the preset advertisement;

the characteristic acquisition module is used for acquiring a preset advertisement analysis characteristic set; the preset advertisement analysis feature set comprises at least one behavior analysis feature and at least one data analysis feature;

the characteristic determining module is used for performing characteristic analysis on the user behavior data to obtain a characteristic value of each behavior analysis characteristic, and performing characteristic analysis on a statistical result of the user input data to obtain a characteristic value of each data analysis characteristic;

a weight determination module, configured to determine, according to a feature weight determination rule corresponding to the preset advertisement analysis feature set, a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination, where the preset advertisement analysis feature combination includes at least one behavior analysis feature and/or at least one data analysis feature;

and the behavior analysis module is used for determining a user behavior analysis result based on the weight value corresponding to each characteristic value in the preset advertisement analysis characteristic set and the weight value corresponding to the preset advertisement analysis characteristic combination.

Optionally, the user behavior analysis apparatus further includes: the channel identification determining module is used for determining a channel identification corresponding to the user behavior analysis result;

the result counting module is used for acquiring a user behavior analysis result corresponding to the same channel identifier;

and the channel analysis module is used for determining a channel operation behavior analysis result corresponding to the channel identifier based on a user behavior analysis result corresponding to the same channel identifier.

Optionally, the statistical result determining module includes:

the data acquisition submodule is used for acquiring historical user input data corresponding to the user input data in a preset historical time period;

the index acquisition submodule is used for acquiring a plurality of preset indexes and the calculation rule of each index;

the index determining submodule is used for calculating the user input data and the historical user input data according to the calculation rule of each index to obtain the index value of each index;

and the statistical result determining submodule is used for determining the index value of each index as the statistical result of the user input data.

Optionally, the weight determining module includes:

the rule obtaining submodule is used for obtaining a characteristic weight determining rule corresponding to the preset advertisement analysis characteristic set;

a first weight determining submodule, configured to determine, based on the feature weight determining rule, an interval in which each feature value in the preset advertisement analysis feature set is located, and determine a weight value corresponding to the interval as a weight value corresponding to the feature value;

the combination obtaining submodule is used for obtaining a preset advertisement analysis characteristic combination; the preset advertisement analysis feature combination comprises at least one behavior analysis feature and/or at least one data analysis feature;

the weight obtaining submodule is used for obtaining a weight value corresponding to a characteristic value of the behavior analysis characteristic and/or the data analysis characteristic in the preset advertisement analysis characteristic combination;

and the second weight determining submodule is used for determining a weight value corresponding to the preset advertisement analysis feature combination based on the weight value corresponding to the characteristic value of the behavior analysis feature and/or the data analysis feature in the preset advertisement analysis feature combination.

An electronic device, comprising: a memory and a processor;

wherein the memory is used for storing programs;

the processor calls a program and is used to:

Compared with the prior art, the invention has the following beneficial effects:

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, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.

Fig. 1 is a flowchart of a method for analyzing user behavior according to an embodiment of the present invention;

fig. 2 is a flowchart of another method for analyzing user behavior according to an embodiment of the present invention;

FIG. 3 is a scene diagram of a feature provided by an embodiment of the invention;

fig. 4 is a flowchart of a method of analyzing user behavior according to another embodiment of the present invention;

fig. 5 is a flowchart of a method of analyzing user behavior according to another embodiment of the present invention;

fig. 6 is a schematic structural diagram of a user behavior analysis apparatus according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

At present, in order to improve brand exposure of enterprises, insurance advertisements acquire customers through advertisement drainage, and the purpose of single conversion is achieved. However, in the actual delivery process, part of the channels are subjected to cheating methods, such as providing a large amount of false flow or providing sub-optimal flow, so that the subsequent conversion of the single product is extremely low, and the input and output are seriously unbalanced.

The common cheating methods include the following methods:

1. and displaying the cheat. That is, a developer superimposes a plurality of advertisements on the same position of a page, an actual user can only see one advertisement, or tens of thousands of advertisement spots are densely arranged in the page, the exposure of the plurality of advertisement spots is realized by the same page loading, but the actual user cannot notice the corresponding advertisement at all.

2. And clicking to cheat. One is error click, which can be passed only by clicking in the operation flow of the user, resulting in a large number of clicks, and the other simulates the user through an automatic script or program, even stimulates the user to click, thereby achieving the aims of generating a large amount of useless advertisement clicks and consuming the budget of an advertiser.

3. Installation/activation cheating, simulation downloading by simulation of improper equipment such as a simulation machine or a test machine, modification of equipment information by manual or technical means, interception and modification of SDK sending information and the like.

4. The traffic is attributed to cheating. The advertisement platform allocates the budget of the advertiser according to the last click model to redirect the advertisement, which causes the waste of budget.

To avoid the above-mentioned cheating methods, the present invention provides a method to identify false or sub-full traffic to avoid wasting advertising budget and to properly plan advertising budget.

The inventor finds that in order to identify false traffic or sub-full traffic, user behavior can be analyzed, the quality of the traffic at the place can be determined according to the user behavior analysis result, the quality of each channel providing the traffic can be further determined, and advertisement putting control can be performed according to the user behavior analysis result.

However, the inventor has found that if only the user behavior is analyzed, the characteristics and service data of the service itself, such as insurance service, cannot be effectively utilized in the advertisement service. Therefore, when the inventor analyzes the user behaviors, the user input data is introduced, such as the input information of the policyholder, the input information of the policyholder and the like, so that the data characteristics of the advertisement service are introduced into the user behavior analysis, the accuracy of the user behavior analysis is improved, the accuracy of flow analysis and the accuracy of channel analysis are further improved, the channel cheating behaviors are identified, the quality of one channel is evaluated, the occurrence probability of the channel cheating behaviors is further reduced, enterprises are helped to realize the control of the delivery cost, and the delivery efficiency is guaranteed.

Specifically, when the user behavior is analyzed, not only the user behavior data but also the statistical result of the user input data is analyzed, that is, the user behavior is analyzed from two dimensions of the user behavior and the user input data, and compared with a mode of only analyzing the user behavior data, the accuracy of user behavior analysis can be improved. Furthermore, when determining the user behavior analysis result, the user behavior analysis method not only refers to the weight values of the behavior analysis features and the feature values of the data analysis features, but also refers to the weight values of the preset advertisement analysis feature combinations, namely, the weights are analyzed from multiple dimensions, and the accuracy of the user behavior analysis can be further improved.

On the basis of the above, an embodiment of the present invention provides a user behavior analysis method, and an execution subject of the user behavior analysis method in this embodiment may be a processor, such as a processor of an advertiser.

Referring to fig. 1, the user behavior analysis method may include:

s11, obtaining user behavior data and user input data corresponding to the preset advertisement, and determining the statistical result of the user input data.

And the user input data is data input in a target interface which is jumped after the user clicks the preset advertisement.

In practical applications, the display interface displays an interface for an advertisement, and the advertisement can be accessed by clicking on the advertisement interface. After clicking the advertisement interface, the user can jump to the detailed interface corresponding to the advertisement, and in this embodiment, the interfaces jumping after clicking the advertisement are collectively referred to as target interfaces. Taking insurance advertisement as an example, after clicking the insurance advertisement interface, skipping to the detailed interface of the insurance advertisement, such as detailed information of insurance and a purchase link of insurance, if the user clicks the purchase link, skipping further to the purchase interface, and the user operates in sequence according to the operation flow, so that the insurance can be purchased.

In this embodiment, the environmental information, the operation behavior, the order information, the transaction information, the continuous and refunding fees, the refunding and refunding fees and the consultation complaint data of the user in the whole process of exposure, click and order placing can be collected through the crawler. The user equipment environment information and the operation behavior can be referred to as user behavior data, and the order information, the transaction information, the continuous fee, the refund fee and the consultation complaint data can be referred to as user input data. The user input data includes data input by the user in each interface, such as information of the applicant, information of the applicant and the like, jumping to after clicking the advertisement.

After obtaining the user input data, it is further required to obtain a statistical result of the user input data based on the user input data, and specifically, referring to fig. 2, determining the statistical result of the user input data may include:

and S21, acquiring historical user input data corresponding to the user input data in a preset historical time period.

In practical applications, the historical user input data may be order information, transaction information, renewal fees, refund fees, and consulting complaint data.

The historical user input data in this embodiment is data of the same user and the same advertisement in a preset historical time period at the current time, for example, the historical user input data of the ad B by the user a. Wherein the preset historical time period may be the previous year.

And S22, acquiring a plurality of preset indexes and calculation rules of the indexes.

The multiple indexes in this embodiment may include a deduction cancellation rate within N days after the application, a guarantee withdrawal rate within N days after the application, an N-term continuous rate after the application, a deduction cancellation rate of the policy, a failure rate of the policy, and an annual premium real yield of the policy.

The calculation rule of the index may be a conventional calculation rule of each index.

And S23, calculating the user input data and the historical user input data according to the calculation rule of each index to obtain the index value of each index.

In practical application, data required for calculating the index is extracted from the user input data and the historical user input data based on the calculation rule of the index, and then the index value of the index is calculated according to the calculation rule of the index.

And S24, determining the index value of each index as the statistical result of the user input data.

After the index values of the indexes are obtained through calculation, the obtained index values are the statistical results of the user input data.

And S12, acquiring a preset advertisement analysis feature set.

In this embodiment, the preset advertisement analysis feature set includes at least one behavior analysis feature and at least one data analysis feature.

The behavior analysis feature, the data analysis feature and the preset advertisement analysis feature combination in the embodiment are determined by technical staff in an offline research and development mode.

Specifically, based on the online user data, operations such as offline feature calculation, feature exploration verification and the like are performed to determine which features are selected and determine feature weight determination rules of the features. And then performing online test, searching based on a rule engine to obtain online user data, performing effect monitoring, effect evaluation and cross validation on the selected features and the feature weight determination rule, if the validation is passed, performing online use, and if the validation is not passed, continuously adjusting the selected features and the feature weight determination rule of the features.

The behavior analysis features in this embodiment may be user basic features (such as user mobile phone number features (such as virtual number segment, abnormal number, blank number, and black list), user operation behavior features (such as advertisement click, browsing behavior (such as page dwell time and dwell area), form filling/submitting (form filling time and space from page entering to form filling), skip payment, and guarantee cancelling (guarantee-guarantee cancelling interval)), user interface brushing behavior features (such as page PV and interface request difference, browsing behavior loss), device environment features (such as protocol IP, UA, resolution, user browser model, user browsing area, simulator information of devices (such as electric quantity, central processing unit CPU, memory, and so on), device environment features (such as protocol IP, UA, resolution, user browser model, user browsing area, and simulator information of devices (such as electric quantity, central processing unit CPU, memory, and so on) Storage space), etc.). Referring to fig. 3, fig. 3 gives an example of some of the features.

The data analysis features may be:

the characteristics of the policyholder: the accumulated insuring singular number of the insurant, the accumulated value risk singular number of the insurant, the accumulated insuring user number of the insurant and the invalid insuring singular number of the insurant;

insured characteristics: the accumulated insuring odd number of the insured person, the accumulated value risk odd number of the insured person, the accumulated number of the insured person who is invested, the number of channels of the insured person insurance source, the number of SKU products of the insured person who is insuring value risk, and the number of the corresponding mobile phone number of the insured person;

service characteristics: the deduction rate of the withdrawal within N days after the insurance is put into operation, the withdrawal rate within N days after the insurance is put into operation, the continuous rate within N days after the insurance is put into operation, the deduction rate of the withdrawal of the insurance policy, the failure rate of the insurance policy, the annual insurance fee actual yield of the insurance policy, the average number of days for insurance, and the like.

The service characteristics are used for comprehensively judging the quality of the user acquired by the delivery channel so as to judge the quality of the channel.

The indexes in the service characteristics can reflect which characteristics can effectively identify the cheating order, ensure that the characteristics are not optional and are full of the Chinese yam, and eliminate the characteristics with randomness and contingency in the distinguishing effect. The obvious cheating distinguishing characteristics should meet the condition that the cheating list and the non-cheating list have obvious difference on evaluation indexes. The specific contents of various indexes of the two types of products, namely the value insurance product and the gift insurance product, are described in detail:

1. evaluation indexes of the value risk product are as follows:

TABLE 1

The data required in the index calculation rule are all obtained based on user input data and the historical user input data statistics.

2. Evaluation indexes of the gift products are as follows:

TABLE 2

S13, performing feature analysis on the user behavior data to obtain a feature value of each behavior analysis feature, and performing feature analysis on the statistical result of the user input data to obtain a feature value of each data analysis feature.

In practical application, after the behavior analysis features and the data analysis features are determined, feature analysis is performed on user behavior data according to the analysis mode of each feature to obtain a feature value of each behavior analysis feature, and feature analysis is performed on the statistical result of the user input data to obtain a feature value of each data analysis feature.

For example, taking the IP in the device environment feature as an example, a corresponding analysis method is preset, and if the IP is analyzed to be a blacklisted IP, the feature value may be divided into two types, yes and no.

S14, determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determination rule corresponding to the preset advertisement analysis feature set.

In this embodiment, in order to focus on the association relationship between different features, a preset advertisement analysis feature combination is set, and when feature values of features in the preset advertisement analysis feature combination are all preset feature values, it is indicated that the user has a high possibility of belonging to a cheating user.

In practical application, for a certain feature, a plurality of intervals are preset, each interval corresponds to a weight value, in the actual judgment process, which interval the obtained weight belongs to is determined, and the weight value corresponding to the interval is determined as the weight value of the feature.

It should be noted that, in the case that whether a feature satisfies a certain condition, if so, the determination corresponds to a weight value, and if not, the determination corresponds to another weight value, which is a special case in the above-mentioned section.

In addition, the preset advertisement analysis feature combination comprises at least one behavior analysis feature and/or at least one data analysis feature. The preset advertisement analysis feature combination has three implementation modes, which are respectively:

1. comprises a plurality of behavior analysis characteristics;

2. comprises a plurality of data analysis features;

3. including at least one behavioral analysis characteristic and at least one data analysis characteristic.

In any of the three cases, the predetermined advertisement analysis feature combination is made to include a plurality of features.

For example, the preset advertisement analysis feature combination comprises a user operation behavior feature and an applicant feature.

The weight value corresponding to the preset advertisement analysis feature combination is determined according to the weight value of each feature included in the preset advertisement analysis feature combination.

S15, determining a user behavior analysis result based on the weight values corresponding to the feature values in the preset advertisement analysis feature set and the weight values corresponding to the preset advertisement analysis feature combination.

In practical applications, step S15 may include:

1) and calculating the sum of the weighted value of each characteristic value in the preset advertisement analysis characteristic set and the weighted value of a preset advertisement analysis characteristic combination.

Specifically, the weight value of each feature value is added to the weight value of a preset advertisement analysis feature combination to obtain a sum.

2) Determining a weight value corresponding to the sum.

In practical applications, there are two implementations for determining the weight value corresponding to the sum.

The first realization mode is as follows:

the sum is directly determined as the weight value.

The second implementation manner is as follows:

and (4) calculating the difference between 100 and the sum by adopting a subtraction system and default to 100, and determining the difference as the weight value.

3) And determining a weight interval corresponding to the weight value, and taking a grade mark corresponding to the weight interval as a user behavior analysis result.

Specifically, the weight values obtained by calculation are preset with weight intervals of 0 to less than or equal to 0 point, 0 to 20 points, 20 to 40 points, 40 to 60 points and more than 60 points.

The different weight intervals correspond to corresponding grade marks, the grade marks in the embodiment may be colors, and the colors corresponding to the different weight intervals are different. For example, the color of the red pigment is black at a value of less than or equal to 0, red at a value of 0-20, orange at a value of 20-40, yellow at a value of 40-60, and green at a value of more than 60.

Taking the above-mentioned division system as an example, the higher the score is, the lower the probability of cheating by the user behavior this time is, and the lower the score is, the higher the probability of cheating by the user behavior this time is.

Taking the above-mentioned manner of directly determining the total as the weight value as an example, the higher the score is, the higher the probability of the user behavior cheating is indicated, and the lower the score is, the lower the probability of the user behavior cheating is indicated.

The grade mark is finally output to the display interface, and in addition, the weight interval corresponding to the weight value can also be output, so that related personnel can directly know the final score of the user.

When analyzing the user behavior, a day update method is used, that is, the user behavior generated on the previous day is analyzed on the second day.

In this embodiment, when analyzing the user behavior, not only the user behavior data but also the statistical result of the user input data may be analyzed, that is, the user behavior may be analyzed from two dimensions of the user behavior and the user input data. Furthermore, when determining the user behavior analysis result, the user behavior analysis method not only refers to the weight values of the behavior analysis features and the feature values of the data analysis features, but also refers to the weight values of the preset advertisement analysis feature combinations, namely, the weights are analyzed from multiple dimensions, and the accuracy of the user behavior analysis can be further improved.

In addition, the embodiment of the invention can analyze cheating characteristics of different channels and identify channel user quality, thereby improving timeliness of detection of cheating behaviors, improving accuracy of detection of the cheating behaviors, providing channel cheating evidence, effectively reducing channel cheating phenomena, helping enterprises reduce putting cost, acquiring high-value users and improving operation efficiency.

In the foregoing embodiment, reference is made to "determining, according to a feature weight determination rule corresponding to the preset advertisement analysis feature set, a weight value corresponding to each feature value in the preset advertisement analysis feature set, and a weight value corresponding to a preset advertisement analysis feature combination", and a specific implementation process of the embodiment is described, with reference to fig. 4, where the implementation process may include:

and S31, acquiring a characteristic weight determination rule corresponding to the preset advertisement analysis characteristic set.

In practical application, the feature weight determination rule corresponding to the preset advertisement analysis feature set includes a feature weight determination rule of each feature in the preset advertisement analysis feature set.

The feature weight determination rule generally determines whether a feature satisfies a certain condition, and if so, the feature corresponds to a certain weight, and if not, the feature corresponds to other weights.

For example, taking the device environment characteristics as an example, an example of the characteristic weight determination rule is given.

1) For the IP features in the device environment features, the corresponding weight determination rule is as follows:

whether the current weight value is a black list IP or not is judged, if yes, the weight value is set, the weight value represents a set weight value, and if not, the weight value is set as & & & &, & & represents another set weight value;

the weight value may be determined by the following determination conditions in the weight determination rule, and a different result of each determination condition described below corresponds to a corresponding weight value.

Such as: whether the IP is the proxy IP or not, whether the regional distribution of the IP accords with the distribution of the prior data or not and whether the IP concentration is greater than the concentration threshold or not.

2) For the UA feature in the device environment feature, determining a weight value through the following judgment conditions in the weight determination rule, wherein the different results of each judgment condition correspond to the corresponding weight value:

whether the browser model and version are normal, whether the device and operating system environment are abnormal, and whether various application interfaces are hijacked.

For other features in the device environment features, the weight determination process is the same as described above, and the following judgment conditions in the weight determination rule are given.

3) Resolution ratio: whether the resolution is abnormal or not and whether the resolution is consistent with the model of the equipment or not;

4) IMEI: whether the IMEI is a blacklist or not and whether the distribution of manufacturers represented by the IMEI is random or not; whether the IMEI coding rule is met or not;

5) and OS: whether the distribution condition of the operating system version accords with certain randomness and statistics or not and whether the distribution condition accords with prior distribution or not;

6) model: whether the brand distribution of the machine types accords with prior data or not and whether the concentration of the machine types is too high or not;

7) accessing: whether the order placing is successful is the first access of the user, whether the user has no access within three days after the order placing is successful, whether automatic withholding is cancelled on the day after the insurance is applied, whether the user has no access record on the day of the insurance application, whether the browsing behavior is abnormal, whether the referrer front page is abnormal, whether the average user access depth is abnormal, whether the average user access duration is abnormal, whether the channel landing page jumping-out rate is abnormal, whether the insurance application page staying duration is abnormal, and whether the insurance application page access frequency is abnormal.

Each of the above-mentioned device environment features determines whether a corresponding condition is satisfied, and if the condition is satisfied or not, there is a corresponding weight value, and the above-mentioned embodiments are all the determination of whether, and in addition, it may also be a determination of multiple sections, for example, it is determined whether a feature value of a feature is in a QW section, a WE section, or an ER section, and each section corresponds to a corresponding weight value.

It should be noted that, the above feature weight determination rule generally determines whether a feature satisfies a certain condition, where the condition includes a threshold, and the threshold is set in multiple ways, which are:

1. detection of abnormal values by statistical methods (3. sigma., quartile spread, etc.)

2. Distance-based outlier detection

3. Abnormal value detection based on classification models, namely, a method through machine learning. (based on the existing data, then build the model, get the abnormal behavior model feature library)

The invention mainly adopts a 3 sigma detection method, namely, a mean value plus three times of standard deviation to set each characteristic threshold value.

And all the thresholds are updated periodically, are updated once in a half year to a year, monitor the effectiveness of the characteristics according to the indexes, and indicate that the thresholds of the characteristics need to be updated when the weight values of the cheat policy and the normal policy in a certain characteristic are similar.

The process of judging whether the threshold value of the feature needs to be updated is as follows:

TABLE 3

As shown in table 3, for a certain feature in the features 1-n in the preset advertisement analysis feature set, the corresponding unmarked policy (the feature value representing that the form corresponds to the feature that is not malicious) and the marked policy (the feature value representing that the form corresponds to the feature that is malicious) are compared in number, if the difference is small, the feature threshold is invalid, and the feature threshold should be updated based on all current historical policy data.

Table 3 summarizes all feature-calculated policies of the previous month by using monthly update, and number 1 each month, to verify the distinguishing effect of the features of the previous month.

S32, determining the interval where each characteristic value in the preset advertisement analysis characteristic set is based on the characteristic weight determination rule, and determining the weight value corresponding to the interval as the weight value corresponding to the characteristic value.

Please refer to the detailed description of step S31 for the specific process of determining the weight value.

S33, acquiring a preset advertisement analysis feature combination; the preset advertisement analysis feature combination comprises at least one behavior analysis feature and/or at least one data analysis feature.

For a specific explanation of the preset advertisement analysis feature combination, please refer to the corresponding explanation in the above embodiments, which is not described herein again.

S34, obtaining a weight value corresponding to the characteristic value of the behavior analysis characteristic and/or the data analysis characteristic in the preset advertisement analysis characteristic combination.

The weight values corresponding to the feature values of the behavior analysis features and/or the data analysis features in the preset advertisement analysis feature combination may be determined in step S31.

S35, determining a weight value corresponding to the preset advertisement analysis feature combination based on the weight value corresponding to the feature value of the behavior analysis feature and/or the data analysis feature in the preset advertisement analysis feature combination.

Specifically, after determining a weight value corresponding to a feature value of the behavior analysis feature and/or the data analysis feature in a preset advertisement analysis feature combination, it is determined whether each weight value is a prescribed weight value, if each weight value is prescribed, the preset advertisement analysis feature combination corresponds to one weight value, and if at least one weight value is not prescribed, the preset advertisement analysis feature combination corresponds to another weight value.

For example, the following gives four different combinations of predetermined advertisement analysis characteristics. It should be noted that, each example is exemplified by satisfying the characteristic weight determination rule, but not exemplified by a weight value, and each condition described below may be modified to a corresponding weight value. In addition, for the preset advertisement analysis feature combination, there are three implementation forms, 1, a plurality of behavior analysis features, 2, a plurality of data analysis features, 3, at least one behavior analysis feature and at least one data analysis feature.

In each of the following examples, the front of the semicolon is the behavior analysis feature, and the back of the semicolon is the data analysis feature.

1) The mobile phone number is abnormal, the mobile phone number is a virtual number section, and the number of concerned public numbers is abnormal; the user has the exception of the monthly settlement single number, the exception of the insurance reimbursement single number, the exception of the insurance successful insurance single number (without gift risk), the exception of the number of insured persons, the insurance reimbursement within T +7 days of the insurance application, and the insurance reimbursement after the payment of one time.

2) The ordering time distribution is abnormal; canceling automatic withholding on the ordering day, abnormal order source ratio, abnormal ordering conversion, abnormal number of strokes of per-person orders, abnormal relation of insured persons and abnormal behavior of canceling insured persons.

3) The regional distribution of the mobile phone numbers of the applicant is abnormal, the distribution of the operators of the mobile phone numbers of the applicant is abnormal, and whether the public number ratio is abnormal or not is concerned; the insured life cancels the automatic replacement of the buckle.

4) The mobile phone number of the insured person is null, the rate of the alien in the attribution area of the mobile phone of the insured person is abnormal, the rate of the alien in the territory of the insured person is abnormal, and the rate of the alien in the attribution area of the mobile phone of the insured person is abnormal; the premium variation coefficient of the insured life is abnormal. In this embodiment, when determining the weight value, the weight value of the feature value of each feature is determined, and the weight value of the feature combination is also determined, so that the analysis of the cheating behavior of the user can be performed based on the weight value of the feature combination and the weight value of a certain feature.

In another embodiment of the present invention, a channel may be analyzed based on an analysis result of the user behavior to determine whether the channel quality is good or bad, whether a cheating user is provided, or whether a high-quality customer is provided. If the policy, the user and the channel are judged to be abnormal according to the statistical rules, for example, the conversion rate of a single channel is within a certain interval range, if a certain flow source has too low or too high conversion rate, the flow source has cheating suspicion, and the user and the channel are scored according to different rules to obtain channel cheating.

Specifically, after step S14, the method may further include:

and S41, determining a channel identifier corresponding to the user behavior analysis result.

Specifically, each user carries a channel identifier, such as channel a, channel b, channel c, etc., when accessing the advertisement, where the channels may be different websites.

After the user behavior analysis result is determined, the corresponding relationship between the user behavior analysis result of the same user and the channel identifier can be established based on the user, and then the channel identifier corresponding to the user behavior analysis result can be determined.

And S42, obtaining a user behavior analysis result corresponding to the same channel identification.

Specifically, the same channel may provide multiple users performing advertisement access, so in this embodiment, multiple user behavior analysis results corresponding to the same channel identifier are summarized.

S43, determining a channel operation behavior analysis result corresponding to the channel identification based on the user behavior analysis result corresponding to the same channel identification.

Specifically, the channel operation behavior analysis result corresponding to the channel identifier may be determined according to the proportion of different colors in the user behavior analysis result corresponding to the same channel identifier.

If the above-mentioned division system is taken as an example, if most of the user behavior analysis results are black, it is indicated that the channel provides more cheating users, and the channel operation behavior analysis result of the channel is a channel with a higher possibility of being a cheating channel.

If the analysis result of most user behaviors is green, the channel provides more high-quality users, and the analysis result of the channel operation behaviors of the channel is a channel which has little possibility of cheating.

In the embodiment, the channel operation behavior analysis result of the channel can be determined based on the user behavior analysis results of the multiple users corresponding to the same channel, so that an advertiser can know cheating conditions of each channel in time and adjust advertisement putting proportion of each channel in time.

Optionally, on the basis of the embodiment of the user behavior analysis method, another embodiment of the present invention provides a user behavior analysis apparatus, and referring to fig. 6, the user behavior analysis apparatus may include:

the statistical result determining module 11 is configured to obtain user behavior data and user input data corresponding to a preset advertisement, and determine a statistical result of the user input data; the user input data is data input in a target interface which is jumped after the user clicks the preset advertisement;

the feature acquisition module 12 is configured to acquire a preset advertisement analysis feature set; the preset advertisement analysis feature set comprises at least one behavior analysis feature and at least one data analysis feature;

a feature determining module 13, configured to perform feature analysis on the user behavior data to obtain a feature value of each behavior analysis feature, and perform feature analysis on a statistical result of the user input data to obtain a feature value of each data analysis feature;

a weight determining module 14, configured to determine, according to a feature weight determining rule corresponding to the preset advertisement analysis feature set, a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination, where the preset advertisement analysis feature combination includes at least one behavior analysis feature and/or at least one data analysis feature;

and the behavior analysis module 15 is configured to determine a user behavior analysis result based on a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to the preset advertisement analysis feature combination.

Further, the user behavior analysis device further includes: the channel identification determining module is used for determining a channel identification corresponding to the user behavior analysis result;

Further, the statistical result determination module comprises:

Further, the weight determination module comprises:

Further, the behavior analysis module includes:

the weight calculation submodule is used for calculating the sum of the weight values of all the characteristic values in the preset advertisement analysis characteristic set and the weight values of the preset advertisement analysis characteristic combination;

a third weight determination submodule for determining a weight value corresponding to the sum;

and the behavior analysis submodule is used for determining a weight interval corresponding to the weight value and taking the grade mark corresponding to the weight interval as a user behavior analysis result.

It should be noted that, for the working processes of each module and sub-module in this embodiment, please refer to the corresponding description in the above embodiments, which is not described herein again.

Optionally, on the basis of the embodiment of the user behavior analysis method and apparatus, another embodiment of the present invention provides an electronic device, including: a memory and a processor;

wherein the memory is used for storing programs;

the processor calls a program and is used to:

Further, after determining a user behavior analysis result based on the weight values corresponding to the feature values in the preset advertisement analysis feature set and the weight values corresponding to the preset advertisement analysis feature combination, the method further includes:

Further, determining statistics of the user input data includes:

acquiring a plurality of preset indexes and calculation rules of the indexes;

Further, determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determination rule corresponding to the preset advertisement analysis feature set includes:

Further, determining a user behavior analysis result based on a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to the preset advertisement analysis feature combination, including:

determining a weight value corresponding to the sum;

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A user behavior analysis method is characterized by comprising the following steps:

acquiring user behavior data and user input data corresponding to a preset advertisement, and determining a statistical result of the user input data so as to introduce service data characteristics of a service corresponding to the advertisement into user behavior analysis and improve the accuracy of the user behavior analysis; the user input data and the user behavior data are data with different dimensions respectively, the user behavior data at least comprise behavior data of operation behaviors executed by a user aiming at an advertisement interface of the preset advertisement and an objective interface which is jumped after the user clicks the preset advertisement, and the user input data are data input by the user in the objective interface which is jumped after the user clicks the preset advertisement;

performing characteristic analysis on the user behavior data to obtain a characteristic value of each behavior analysis characteristic, and performing characteristic analysis on a statistical result of the user input data to obtain a characteristic value of each data analysis characteristic; the data analysis characteristics at least comprise service characteristics, and the service characteristics comprise preset service parameter values related to service types of services corresponding to advertisements;

determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determination rule corresponding to the preset advertisement analysis feature set, wherein the preset advertisement analysis feature combination comprises at least one behavior analysis feature and at least one data analysis feature so as to analyze the behavior of a user on the preset advertisement by combining an incidence relation between different features;

2. The method according to claim 1, wherein after determining a user behavior analysis result based on the weight value corresponding to each feature value in the preset advertisement analysis feature set and the weight value corresponding to the preset advertisement analysis feature combination, the method further comprises:

3. The method of claim 1, wherein determining statistics of the user input data comprises:

acquiring a plurality of preset indexes and calculation rules of the indexes;

4. The method according to claim 1, wherein determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determination rule corresponding to the preset advertisement analysis feature set includes:

acquiring a preset advertisement analysis feature combination; the preset advertisement analysis feature combination comprises at least one behavior analysis feature and at least one data analysis feature;

acquiring weight values corresponding to the characteristic values of the behavior analysis characteristic and the data analysis characteristic in the preset advertisement analysis characteristic combination;

and determining a weight value corresponding to the preset advertisement analysis feature combination based on the weight values corresponding to the characteristic values of the behavior analysis feature and the data analysis feature in the preset advertisement analysis feature combination.

5. The method according to claim 1, wherein determining a user behavior analysis result based on a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to the preset advertisement analysis feature combination comprises:

calculating the sum of the weighted value of each characteristic value in the preset advertisement analysis characteristic set and the weighted value of a preset advertisement analysis characteristic combination;

determining a weight value corresponding to the sum;

6. A user behavior analysis apparatus, comprising:

the statistical result determining module is used for acquiring user behavior data and user input data corresponding to preset advertisements, and determining statistical results of the user input data so as to introduce service data characteristics of services corresponding to the advertisements into user behavior analysis and improve accuracy of the user behavior analysis; the user input data and the user behavior data are data with different dimensions respectively, the user behavior data at least comprise behavior data of operation behaviors executed by a user aiming at an advertisement interface of the preset advertisement and an objective interface which is jumped after the user clicks the preset advertisement, and the user input data are data input by the user in the objective interface which is jumped after the user clicks the preset advertisement;

the characteristic determining module is used for performing characteristic analysis on the user behavior data to obtain a characteristic value of each behavior analysis characteristic, and performing characteristic analysis on a statistical result of the user input data to obtain a characteristic value of each data analysis characteristic; the data analysis characteristics at least comprise service characteristics, and the service characteristics comprise preset service parameter values related to service types of services corresponding to advertisements;

the weight determining module is used for determining a weight value corresponding to each feature value in the preset advertisement analysis feature set and a weight value corresponding to a preset advertisement analysis feature combination according to a feature weight determining rule corresponding to the preset advertisement analysis feature set, wherein the preset advertisement analysis feature combination comprises at least one behavior analysis feature and at least one data analysis feature so as to analyze the behavior of a user on the preset advertisement by combining the incidence relation between different features;

7. The apparatus according to claim 6, further comprising: the channel identification determining module is used for determining a channel identification corresponding to the user behavior analysis result;

8. The apparatus according to claim 6, wherein the statistical result determining module comprises:

9. The apparatus of claim 6, wherein the weight determination module comprises:

the combination obtaining submodule is used for obtaining a preset advertisement analysis characteristic combination; the preset advertisement analysis feature combination comprises at least one behavior analysis feature and at least one data analysis feature;

the weight obtaining submodule is used for obtaining a weight value corresponding to the characteristic value of the behavior analysis characteristic and the characteristic value of the data analysis characteristic in the preset advertisement analysis characteristic combination;

and the second weight determining submodule is used for determining a weight value corresponding to the preset advertisement analysis feature combination based on the weight values corresponding to the characteristic values of the behavior analysis feature and the data analysis feature in the preset advertisement analysis feature combination.

10. An electronic device, comprising: a memory and a processor;

wherein the memory is used for storing programs;

the processor calls a program and is used to: