Channel behavior quality assessment method, channel quality assessment method and related equipment

Abstract

The embodiment of the invention provides a channel behavior quality evaluation method, a channel quality evaluation method and related equipment. The channel behavior quality evaluation method comprises the steps of obtaining equipment behavior similarity among multiple pieces of equipment corresponding to a channel according to channel behavior data of each channel; determining a channel cohesion value and a channel diversity value corresponding to each channel according to each equipment behavior similarity corresponding to each channel; and determining a channel behavior quality score corresponding to each channel according to the channel benchmark value, the channel cohesiveness value and the channel diversity value corresponding to each channel. According to the technical scheme of the embodiment of the invention, behavior related factors such as the behavior data and the multi-dimensional evaluation indexes are introduced to enrich a buying quantity channel quality evaluation system, the evaluation accuracy is maximized, the channel selection process is optimized, and thus the scientificity and efficiency of the advertisement putting and popularization effect are improved.

Description

Technical Field

[0001] This application relates to the fields of computer and communication technologies. Specifically, it relates to a method for evaluating the quality of channel behavior, a method for evaluating channel quality, and related devices. Background Art

[0002] Currently, the evaluation system for evaluating the quality of buying channels mainly comprehensively measures from three major dimensions: new user acquisition, retention, and revenue and expenditure. This evaluation system fails to fully consider the diversity and consistency of the traffic within the channels and is prone to ignoring the interference of false traffic. Therefore, solving how to accurately evaluate the quality of buying channels, identify false traffic, and select high-quality traffic has become a technical problem. Summary of the Invention

[0003] Embodiments of this application provide a method for evaluating the quality of channel behavior, a method for evaluating channel quality, and related devices, which can thus at least to a certain extent accurately evaluate the quality of buying channels, identify false traffic, and select high-quality traffic.

[0004] Other features and advantages of this application will become apparent through the following detailed description, or will be partially learned through the practice of this application.

[0005] According to one aspect of the embodiments of this application, a method for evaluating the quality of channel behavior is provided, including: respectively obtaining the device behavior similarity between multiple devices corresponding to each channel according to the channel behavior data of each channel; determining the channel cohesion value and channel diversity value corresponding to each channel according to the device behavior similarity corresponding to each channel; determining the channel behavior quality score corresponding to each channel according to the channel benchmark value, channel cohesion value, and channel diversity value corresponding to each channel.

[0006] In some embodiments of this application, the step of respectively obtaining the device behavior similarity between multiple devices corresponding to each channel according to the channel behavior data of each channel specifically includes: respectively extracting device behavior representations from the channel behavior data of each channel through a behavior characterization extraction model to obtain device behavior chains; respectively performing same-dimensional processing on each device behavior chain corresponding to each channel to obtain corresponding same-dimensional device behavior chains; respectively comparing each same-dimensional device behavior chain corresponding to each channel to obtain the device behavior similarity between devices.

[0007] In some embodiments of this application, the step of respectively performing same-dimensional processing on each device behavior chain corresponding to each channel to obtain corresponding same-dimensional device behavior chains specifically includes: respectively performing filling processing on each device behavior chain corresponding to each channel to obtain filled device behavior chains; respectively performing same-position alignment processing on each device behavior chain corresponding to each channel to obtain same-dimensional device behavior chains.

[0008] In some embodiments of the present application, the behavior characterization extraction model includes an encoder and a decoder, and the method further includes: obtaining a set of device behavior data sequence samples, where the set of device behavior data sequence samples contains multiple device behavior data sequence samples, and each device behavior data sequence sample is labeled with a corresponding behavior characterization sequence; inputting the device behavior data sequence samples into the encoder to obtain a behavior sequence characterization; inputting the behavior sequence characterization into the decoder to obtain a behavior characterization vector; determining a loss function based on the finally obtained behavior characterization vector and the labeled behavior characterization sequence; and updating the parameters of the encoder and the decoder based on the loss function until the loss function converges.

[0009] In some embodiments of the present application, determining the channel cohesion value and the channel diversity value corresponding to each channel according to the device behavior similarities corresponding to each channel specifically includes: respectively taking the mean of the device behavior similarities corresponding to each channel to obtain the channel cohesion value; and respectively taking the standard deviation of the device behavior similarities corresponding to each channel to obtain the channel diversity value.

[0010] In some embodiments of the present application, the channel benchmark values include a channel cohesion benchmark value and a channel diversity benchmark value. Determining the channel behavior quality score corresponding to each channel according to the channel benchmark values, the channel cohesion value, and the channel diversity value corresponding to each channel specifically includes: determining the channel behavior cohesion score corresponding to each channel according to the channel cohesion benchmark value and the channel cohesion value corresponding to each channel, where the channel cohesion benchmark value is the cohesion value determined from natural traffic; determining the channel behavior diversity score corresponding to each channel according to the channel diversity benchmark value and the channel diversity value corresponding to each channel, where the channel diversity benchmark value is the diversity value determined from natural traffic; and obtaining the channel behavior quality score corresponding to each channel according to the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel.

[0011] In some embodiments of the present application, determining the channel behavior cohesion score corresponding to each channel according to the channel cohesion benchmark value and the channel cohesion value corresponding to each channel specifically includes: comparing the magnitudes of the channel cohesion benchmark value and the channel cohesion value corresponding to each channel; if the channel cohesion value corresponding to this channel is smaller than the channel cohesion benchmark value, then using the channel cohesion value as the cohesion behavior amplitude difference; if the channel cohesion value corresponding to this channel is larger than the channel cohesion benchmark value, then using the difference between the unit value and the channel cohesion value as the cohesion behavior amplitude difference; determining the maximum cohesion behavior amplitude difference and the minimum cohesion behavior amplitude difference among the cohesion behavior amplitude differences corresponding to each channel; and obtaining the channel behavior cohesion score according to the maximum cohesion behavior amplitude difference, the minimum cohesion behavior amplitude difference, and the channel cohesion value corresponding to each channel.

[0012] In some embodiments of the present application, determining the channel behavior diversity score corresponding to each channel according to the channel diversity benchmark value and the channel diversity value corresponding to each channel specifically includes: comparing the magnitudes of the channel diversity benchmark value and the channel diversity value corresponding to each channel; if the channel diversity value corresponding to this channel is smaller than the channel diversity benchmark value, then using the channel diversity value as the diversity behavior amplitude difference; if the channel diversity value corresponding to this channel is larger than the channel diversity benchmark value, then using the difference between the unit value and the channel diversity value as the diversity behavior amplitude difference; determining the maximum diversity behavior amplitude difference and the minimum diversity behavior amplitude difference among the diversity behavior amplitude differences corresponding to each channel; and obtaining the channel behavior diversity score according to the maximum diversity behavior amplitude difference, the minimum diversity behavior amplitude difference, and the channel diversity value corresponding to each channel.

[0013] In some embodiments of the present application, obtaining the channel behavior quality score corresponding to each channel according to the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel specifically includes: performing weighted summation on the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel to obtain the channel behavior quality score corresponding to each channel.

[0014] According to one aspect of the embodiments of the present application, a method for evaluating channel quality is provided. The method for evaluating channel behavior quality includes: obtaining channel evaluation data, where the channel evaluation data includes channel new data, channel retention data, channel revenue data, and channel behavior data; determining a channel new quality score according to the channel new data; determining a channel retention quality score according to the channel retention data; determining a channel revenue quality score according to the channel revenue data; determining a channel behavior quality score by using the method for evaluating channel behavior quality as described above according to the channel behavior data; and determining a channel quality evaluation score according to the channel new quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

[0015] In some embodiments of the present application, the determining a channel new quality score according to the channel new data specifically includes: determining a new channel new score and an existing channel new score respectively according to the channel new data; and determining a channel new quality score according to the new channel new score and the existing channel new score.

[0016] In some embodiments of the present application, the determining a channel retention quality score according to the channel retention data specifically includes: determining a next-day retention rate, a weekly retention rate, and a long-term retention rate according to the channel retention data; and determining a channel retention quality score according to the next-day retention rate, the weekly retention rate, and the long-term retention rate.

[0017] In some embodiments of the present application, the determining a channel revenue quality score according to the channel revenue data specifically includes: determining an active device revenue, a channel revenue, and a channel cost according to the channel revenue data; and determining a channel revenue quality score according to the active device revenue, the channel revenue, and the channel cost.

[0018] In some embodiments of the present application, the determining a channel quality evaluation score according to the channel new quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score specifically includes: performing a weighted sum of the channel new quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score to obtain a channel behavior quality evaluation score.

[0019] According to one aspect of the embodiments of the present application, a channel behavior quality evaluation device is provided. The channel behavior quality evaluation device includes: a device similarity determination module, configured to obtain the device behavior similarity between multiple devices corresponding to each channel respectively according to the channel behavior data of each channel; a cohesion and diversity determination module, configured to determine the channel cohesion value and the channel diversity value corresponding to each channel according to the device behavior similarity corresponding to each channel; a behavior score determination module, configured to determine the channel behavior quality score corresponding to each channel according to the channel benchmark value, the channel cohesion value, and the channel diversity value corresponding to each channel.

[0020] In some embodiments of the present application, the device similarity determination module specifically includes: a behavior characterization extraction sub-module, configured to extract device behavior characterizations from the channel behavior data of each channel respectively through a behavior characterization extraction model to obtain device behavior chains; a device same-dimension processing sub-module, configured to perform same-dimension processing on each device behavior chain corresponding to each channel respectively to obtain corresponding same-dimension device behavior chains; a device similarity determination sub-module, configured to compare each same-dimension device behavior chain corresponding to each channel respectively to obtain the device behavior similarity between devices.

[0021] In some embodiments of the present application, the device same-dimension processing sub-module specifically includes: a behavior chain completion unit, configured to perform completion processing on each device behavior chain corresponding to each channel respectively to obtain completed device behavior chains; a same-position alignment unit, configured to perform same-position alignment processing on each device behavior chain corresponding to each channel respectively to obtain same-dimension device behavior chains.

[0022] In some embodiments of the present application, the behavior characterization extraction model includes an encoder and a decoder. The channel behavior quality evaluation device further includes: a sample acquisition module, configured to acquire a sample set of device behavior data sequences, where the sample set of device behavior data sequences contains multiple device behavior data sequence samples, and each device behavior data sequence sample is marked with a corresponding behavior characterization sequence; a sample encoding module, configured to input the device behavior data sequence sample into the encoder to obtain a behavior sequence characterization; a sample decoding module, configured to input the behavior sequence characterization into the decoder to obtain a behavior characterization vector; a loss function module, configured to determine a loss function based on the finally obtained behavior characterization vector and the marked behavior characterization sequence; a parameter update module, configured to update the parameters of the encoder and the decoder based on the loss function until the loss function converges.

[0023] In some embodiments of the present application, the cohesion diversity determination module specifically includes: a channel cohesion value sub-module, configured to calculate the average value of the device behavior similarities corresponding to each channel, so as to obtain a channel cohesion value; a channel diversity value sub-module, configured to calculate the standard deviation of the device behavior similarities corresponding to each channel, so as to obtain a channel diversity value.

[0024] In some embodiments of the present application, the channel benchmark value includes a channel cohesion benchmark value and a channel diversity benchmark value. The behavior score determination module specifically includes: a behavior cohesion score sub-module, configured to determine the channel behavior cohesion score corresponding to each channel according to the channel cohesion benchmark value and the channel cohesion value corresponding to each channel, where the channel cohesion benchmark value is the cohesion value determined from the natural traffic; a behavior diversity score sub-module, configured to determine the channel behavior diversity score corresponding to each channel according to the channel diversity benchmark value and the channel diversity value corresponding to each channel, where the channel diversity benchmark value is the diversity value determined from the natural traffic; a channel behavior quality score sub-module, configured to obtain the channel behavior quality score corresponding to each channel according to the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel.

[0025] In some embodiments of the present application, the behavior cohesion score sub-module specifically includes: a cohesion benchmark comparison unit, configured to compare the magnitudes of the channel cohesion benchmark value and the channel cohesion value corresponding to each channel; a first cohesion amplitude unit, configured to use the channel cohesion value as the cohesion behavior amplitude difference if the channel cohesion value corresponding to the channel is smaller than the channel cohesion benchmark value; a second cohesion amplitude unit, configured to use the difference between the unit value and the channel cohesion value as the cohesion behavior amplitude difference if the channel cohesion value corresponding to the channel is larger than the channel cohesion benchmark value; a cohesion amplitude extreme value unit, configured to determine the maximum cohesion behavior amplitude difference and the minimum cohesion behavior amplitude difference among the cohesion behavior amplitude differences corresponding to each channel; a cohesion score calculation unit, configured to obtain the channel behavior cohesion score according to the maximum cohesion behavior amplitude difference, the minimum cohesion behavior amplitude difference, and the channel cohesion value corresponding to each channel.

[0026] In some embodiments of the present application, the behavior diversity sub-module specifically includes: a diversity benchmark comparison unit, configured to compare the magnitude of the channel diversity benchmark value and the channel diversity value corresponding to each channel; a first diversity amplitude unit, configured to use the channel diversity value as the diversity behavior amplitude difference if the channel diversity value corresponding to the channel is smaller than the channel diversity benchmark value; a second diversity amplitude unit, configured to use the difference between the unit value and the channel diversity value as the diversity behavior amplitude difference if the channel diversity value corresponding to the channel is larger than the channel diversity benchmark value; a diversity amplitude extreme value unit, configured to determine the maximum diversity behavior amplitude difference and the minimum diversity behavior amplitude difference among the diversity behavior amplitude differences corresponding to each channel; a diversity score calculation unit, configured to obtain the channel behavior diversity score according to the maximum diversity behavior amplitude difference, the minimum diversity behavior amplitude difference, and the channel diversity value corresponding to each channel.

[0027] In some embodiments of the present application, the channel behavior quality sub-module specifically includes: a channel behavior quality score unit, configured to perform weighted summation on the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel to obtain the channel behavior quality score corresponding to each channel.

[0028] According to one aspect of the embodiments of the present application, a channel quality evaluation system is provided. The channel quality evaluation system includes: a channel quality data acquisition device, configured to acquire channel evaluation data, where the channel evaluation data includes channel new data, channel retention data, channel revenue data, and channel behavior data; a channel new quality evaluation device, configured to determine a channel new quality score according to the channel new data; a channel retention quality evaluation device, configured to determine a channel retention quality score according to the channel retention data; a channel revenue quality evaluation device, configured to determine a channel revenue quality score according to the channel revenue data; a channel behavior quality evaluation device, configured to determine a channel behavior quality score by using the channel behavior quality evaluation method as described above according to the channel behavior data; a channel quality evaluation score device, configured to determine a channel quality evaluation score according to the channel new quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

[0029] In some embodiments of the present application, the channel quality data acquisition device specifically includes: a channel new score module, configured to respectively determine a new channel new score and an existing channel new score according to the channel new data; a new quality score module, configured to determine a channel new quality score according to the new channel new score and the existing channel new score.

[0030] In some embodiments of the present application, the channel retention quality evaluation device specifically includes: a retention parameter determination module, configured to determine the next-day retention rate, the weekly retention rate, and the long-term retention rate according to the channel retention data; a retention quality score module, configured to determine the channel retention quality score according to the next-day retention rate, the weekly retention rate, and the long-term retention rate.

[0031] In some embodiments of the present application, the channel revenue quality evaluation device specifically includes: a revenue-cost determination module, configured to determine the active device revenue, the channel revenue, and the channel cost according to the channel revenue data; a revenue quality score module, configured to determine the channel revenue quality score according to the active device revenue, the channel revenue, and the channel cost.

[0032] In some embodiments of the present application, the channel quality evaluation score device specifically includes: a channel quality evaluation score module, configured to perform a weighted sum of the channel new user quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score to obtain the channel behavior quality evaluation score.

[0033] According to one aspect of the embodiments of the present application, there is provided a computer program product, including one or more computer programs, which when executed by one or more processors, implement the channel behavior quality evaluation method as described in the above embodiments or implement the channel quality evaluation method as described in the above embodiments.

[0034] According to one aspect of the embodiments of the present application, there is provided a computer-readable medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the channel behavior quality evaluation method as described in the above embodiments or implements the channel quality evaluation method as described in the above embodiments.

[0035] According to one aspect of the embodiments of the present application, there is provided an electronic device, including: one or more processors; a storage device, configured to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the channel behavior quality evaluation method as described in the above embodiments or implement the channel quality evaluation method as described in the above embodiments.

[0036] In the technical solutions provided by some embodiments of the present application, by analyzing the similarity of device behaviors, the consistency of traffic within channels is evaluated to help identify false traffic. At the same time, through two dimensions of behavior consistency and behavior difference, combined with a comparison with a benchmark value, a comprehensive quality score for each channel is obtained, which can more comprehensively analyze the quality of channel traffic, help make optimization decisions, effectively solve the technical problems that cannot be refined by traditional evaluation methods, enrich the evaluation system for the quality of traffic acquisition channels by introducing behavior-related factors such as behavior data and multi-dimensional evaluation indicators, maximize the accuracy of evaluation, optimize the channel selection process, and thus improve the scientific nature and efficiency of advertising placement and promotion effects.

[0037] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present application and used together with the specification to explain the principles of the present application. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts. In the drawings:

[0039] Figure 1 A schematic diagram showing an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied is shown.

[0040] Figure 2 A flowchart showing the process of a method for evaluating the quality of channel behaviors provided by an embodiment of the present application is shown.

[0041] Figure 3 Shows according to Figure 2 A specific implementation flowchart of step S510 in the method for evaluating the quality of channel behaviors corresponding to the embodiment is shown.

[0042] Figure 4 Shows according to Figure 2 A specific implementation flowchart of step S530 in the method for evaluating the quality of channel behaviors corresponding to the embodiment is shown.

[0043] Figure 5 A flowchart showing the process of a method for evaluating the quality of channels provided by an embodiment of the present application is shown.

[0044] Figure 6 A schematic diagram showing the structure of a device for evaluating the quality of channel behaviors provided by an embodiment of the present application is shown.

[0045] Figure 7 A schematic diagram showing the structure of a device for evaluating the quality of channels provided by an embodiment of the present application is shown.

[0046] Figure 8 The figure shows a schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0047] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more complete and comprehensive, and will fully convey the concept of the example embodiments to those skilled in the art.

[0048] In addition, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of this application. However, those skilled in the art will realize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, well-known methods, devices, implementations, or operations are not shown or described in detail to avoid obscuring aspects of this application.

[0049] The block diagrams shown in the drawings are only functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0050] The flowcharts shown in the drawings are only illustrative and do not necessarily include all the content and operations / steps, nor do they necessarily have to be executed in the described order. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to the actual situation.

[0051] Figure 1 The figure shows a schematic diagram of an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied.

[0052] As Figure 1 shown, the system architecture can include terminal devices (such as Figure 1 one or more of the smart phone 101, tablet computer 102, and portable computer 103 shown in the figure, and of course it can also be a desktop computer, etc.), a network 104, and a server 105. The network 104 is used as a medium to provide a communication link between the terminal device and the server 105. The network 104 can include various connection types, such as wired communication links, wireless communication links, etc.

[0053] It should be understood that Figure 1The numbers of the terminal devices, networks, and servers therein are merely illustrative. According to implementation requirements, there can be any number of terminal devices, networks, and servers. For example, the server 105 can be a server cluster composed of multiple servers, etc.

[0054] Users can use the terminal device to interact with the server 105 through the network 104 to receive or send messages, etc. The server 105 can be a server that provides various services. For example, the user uses the terminal device 103 (which can also be the terminal device 101 or 102) to upload the channel behavior data of each channel to the server 105. The server 105 can respectively obtain the device behavior similarities between multiple devices corresponding to the channel according to the channel behavior data of each channel; determine the channel cohesion value and channel diversity value corresponding to each channel according to the device behavior similarities corresponding to each channel; and determine the channel behavior quality score corresponding to each channel according to the channel benchmark value, channel cohesion value, and channel diversity value corresponding to each channel.

[0055] It should be noted that the channel behavior quality assessment method provided by the embodiments of the present application is generally executed by the server 105. Correspondingly, the channel behavior quality assessment device is generally set in the server 105. However, in other embodiments of the present application, the terminal device can also have a similar function as the server, so as to execute the channel behavior quality assessment solution provided by the embodiments of the present application.

[0056] The implementation details of the technical solutions of the embodiments of the present application are elaborated in detail below:

[0057] Figure 2 shows a flowchart of a channel behavior quality assessment method according to an embodiment of the present application. The channel behavior quality assessment method can be executed by a server, and the server can be Figure 1 the server shown in Figure 2 As shown, the channel behavior quality assessment method at least includes:

[0058] S510, respectively obtain the device behavior similarities between multiple devices corresponding to the channel according to the channel behavior data of each channel.

[0059] S520, determine the channel cohesion value and channel diversity value corresponding to each channel according to the device behavior similarities corresponding to each channel.

[0060] S530, determine the channel behavior quality score corresponding to each channel according to the channel benchmark value, channel cohesion value, and channel diversity value corresponding to each channel.

[0061] In the embodiments of the present application, by analyzing the similarity of device behavior and evaluating the consistency of traffic within the channel, false traffic is identified. At the same time, by comparing the two dimensions of behavior consistency and behavior difference, and combining benchmark values, the comprehensive quality score of each channel is obtained, which can more comprehensively analyze the quality of channel traffic, help make optimization decisions, and effectively solve technical problems that traditional evaluation methods have not been able to refine. By introducing behavior-related factors such as behavioral data and multi-dimensional evaluation indicators to enrich the quality evaluation system of the buying volume channel, the accuracy of the evaluation is maximized, the channel selection process is optimized, and the scientificity and efficiency of advertising and promotion effects are improved.

[0062] In S510, by analyzing the device behavior similarity, it is possible to preliminarily identify whether there are large-scale false behavior patterns (e.g., automated non-real user behaviors) in the channel. Channels with higher device behavior similarity often indicate that the traffic of the channel is more real, stable, and has higher quality.

[0063] Specifically, in some embodiments, the specific implementation of step S100 can be found in Figure 3 . Figure 3 is based on Figure 2 The detailed description of step S100 in the channel behavior quality evaluation method shown in the corresponding embodiment, in the channel behavior quality evaluation method, step S100 may include the following steps:

[0064] S512, extracting device behavior representations from the channel behavior data of each channel through a behavior representation extraction model to obtain a device behavior chain.

[0065] S514, performing same-dimensional processing on each device behavior chain corresponding to each channel to obtain a corresponding same-dimensional device behavior chain.

[0066] S516, respectively compare the device behavior chains of the same dimension corresponding to each channel to obtain the device behavior similarity between the devices.

[0067] In this embodiment, first, the behavior characterization extraction model can accurately extract the device behavior characteristics, avoiding the complexity and noise interference caused by directly processing the original data. Through the extraction of behavior characteristics and the construction of device behavior chains, it can effectively reflect the interaction process of the device in the channel, thereby improving the accuracy of the behavior similarity calculation. Secondly, the same-dimensional processing ensures the dimensional consistency of device behavior data from different channels. The same-dimensional processing eliminates the dimensional differences between different channels and devices, so that the subsequent similarity calculation can eliminate the interference caused by the data differences between channels, thereby making the behavior similarity comparison between different devices more reasonable and fair. Finally, through the calculation of behavioral similarity, the quality of the traffic in the channel can be accurately evaluated and false traffic can be effectively identified.

[0068] False traffic often manifests as randomness or abnormality in device behavior. Therefore, by comparing device behavior chains and calculating similarity, devices with abnormal behavior can be quickly identified, thus filtering out invalid traffic.

[0069] This embodiment can handle data differences from different channels and different devices and is applicable to a wider range of scenarios. Whether it is a channel with a large data dimension difference or a situation with complex device behavior, efficient and accurate evaluation can be achieved through the same-dimension processing and behavior chain comparison techniques.

[0070] In S512, through the behavior characterization extraction model, meaningful behavior features can be effectively extracted from complex raw data to construct the behavior chain of the device. The behavior chain is a tool that can accurately represent the behavior characteristics of the device and provides a clear basis for subsequent behavior similarity calculation.

[0071] Among them, the above-mentioned behavior characterization extraction model includes an encoder and a decoder, and its specific training method may include the following steps:

[0072] Obtain a sample set of device behavior data sequences, where the sample set of device behavior data sequences contains multiple device behavior data sequence samples, and each device behavior data sequence sample is marked with a corresponding behavior characterization sequence.

[0073] Input the device behavior data sequence sample into the encoder to obtain a behavior sequence characterization.

[0074] Input the behavior sequence characterization into the decoder to obtain a behavior characterization vector.

[0075] Based on the finally obtained behavior characterization vector and the marked behavior characterization sequence, determine the loss function.

[0076] Based on the loss function, update the parameters of the encoder and the decoder until the loss function converges.

[0077] In this embodiment, first, a plurality of device behavior data sequence samples are obtained, and each sample is labeled with a corresponding behavior characterization sequence. The device behavior data sequence samples usually contain a series of behavior records of a device within a certain time period, and these data records may be the operation behaviors of the device, timestamps, behavior types, etc. Each device behavior data sequence is labeled with a corresponding "behavior characterization sequence", and this sequence reflects the characteristics of the device behavior. The device behavior data sequence samples are input into the encoder, and the encoder converts the original device behavior data sequence into a low-dimensional behavior sequence characterization through a deep learning model. The encoder usually adopts deep neural network architectures similar to LSTM, GRU, or Transformer, etc., which can effectively capture the temporal characteristics and long-term and short-term dependencies of the data. The behavior sequence characterization processed by the encoder is input into the decoder, and the task of the decoder is to further convert the behavior sequence characterization output by the encoder into a final behavior characterization vector. The decoder is symmetric with the encoder, and common architectures include the Transformer decoder or the LSTM decoder, and its purpose is to further map the compressed characterization into a vector that can reflect the behavior characteristics. The design of the loss function is based on the difference between the finally obtained behavior characterization vector and the labeled behavior characterization sequence. Usually, the mean square error (MSE), cross-entropy loss, or other loss functions suitable for sequence generation are used to measure the prediction error of the model by calculating the difference between the output behavior characterization and the target behavior characterization sequence. Through the backpropagation algorithm, the parameters of the encoder and the decoder are updated according to the gradients calculated by the loss function. The parameter update process continues until the loss function converges, that is, the difference between the output of the model and the labeled behavior characterization sequence reaches the minimum value.

[0078] Through the joint training of the encoder and the decoder in this embodiment, the device behavior data can be extracted into a more concise and effective characterization vector from the complex original data, thus avoiding the problems of high dimensionality and sparsity of the original data. The design of the loss function and the gradient descent optimization enable the model to continuously learn the rules in the device behavior data and finally obtain a characterization vector that can accurately reflect the device behavior characteristics.

[0079] Through the high-quality behavior characterization vector, the model can calculate the behavior similarity more accurately, provide more accurate data support for the evaluation of channel behavior quality, and improve the ability to identify false traffic.

[0080] In S514, the purpose of the same-dimensional processing is to make the device behavior data from different channels have consistency, thus avoiding the errors caused by data dimension mismatch, significantly improving the accuracy of subsequent device behavior similarity calculation, and ensuring the reliability of the channel evaluation results.

[0081] Specifically, in some embodiments, the specific implementation of step S514 can be referred to the following embodiments. This embodiment is based on Figure 2 the detailed description of step S514 in the channel behavior quality assessment method shown in the corresponding embodiment. In the channel behavior quality assessment method, step S514 may include the following steps:

[0082] Perform padding processing on each device behavior chain corresponding to each channel to obtain padded device behavior chains.

[0083] Perform same-position alignment processing on each device behavior chain corresponding to each channel to obtain same-dimensional device behavior chains.

[0084] In this embodiment, through padding processing and same-position alignment processing, the problems of length differences and time misalignments between different channels and device behavior chains are effectively solved, ensuring the consistency of device behavior chains in length and position, thereby improving the accuracy of behavior similarity calculation and the precision of channel quality assessment. Ultimately, false traffic can be identified more reliably, and a more accurate basis can be provided for channel quality assessment.

[0085] Specifically, the padding processing ensures the consistency of the behavior chains of all devices in length, enabling data to be compared in the same dimension, thereby eliminating the interference caused by inconsistent lengths of device behavior chains. At the same time, through the same-position alignment processing, it is ensured that the behavior data from different devices can be compared at the same time node, making the comparison between device behavior chains more fair and accurate.

[0086] The device behavior chains after padding and alignment have a unified length and position, avoiding errors caused by inconsistent data, thereby improving the precision of behavior similarity calculation. Ultimately, the quality of the channel traffic evaluated is more accurate, and the ability to identify false traffic is stronger.

[0087] Whether it is the difference in the dimensions of behavior data between channels or the problems of the length and time misalignment of behavior data between devices, through the application of padding and alignment technologies, various complex situations can be handled, enabling this method to operate stably in various scenarios.

[0088] In S516, by accurately calculating the device behavior similarity, it is possible to determine whether the behaviors of different devices within the same channel are consistent. If the behavior similarity of most devices is relatively high, it indicates that the traffic of this channel is relatively real and consistent; if the similarity is relatively low, it may indicate the existence of false traffic or low-quality traffic.

[0089] In S520, the calculation of the cohesion and diversity values helps to refine the evaluation of channel quality, rather than relying solely on simple new user or ROI metrics. This can effectively reduce the impact of false traffic and make the evaluation of channel quality more accurate. For example, if a channel has a very low cohesion and a too high diversity, it may mean that the quality of the traffic in this channel is not high (for example, there is a large amount of false traffic or bot traffic). On the contrary, a channel with high cohesion and moderate diversity may represent high-quality and real user traffic.

[0090] Specifically, in some embodiments, the specific implementation of step S520 can refer to the following embodiments. This embodiment is based on Figure 2 the detailed description of step S520 in the channel behavior quality evaluation method shown in the corresponding embodiment. In the channel behavior quality evaluation method, step S520 may include the following steps:

[0091] Respectively take the mean value of the device behavior similarities corresponding to each channel to obtain the channel cohesion value.

[0092] Respectively take the standard deviation of the device behavior similarities corresponding to each channel to obtain the channel diversity value.

[0093] In the embodiment, cohesion reflects the degree of similarity of user behaviors within the same channel. Specifically, by analyzing the device behavior similarities within the channel, a cohesion index is calculated, which measures whether the user behaviors within this channel are consistent. The higher the cohesion value, the more unified the user behaviors within the channel and the higher the traffic quality. In this embodiment, the cohesion value is obtained by taking the mean value.

[0094] Suppose for a certain channel A, calculate the similarity values between all its devices, and find the average of these similarity values, then the cohesion value of this channel is obtained. Through this value, it is possible to clearly know the degree of consistency of the device behaviors in this channel.

[0095] For example: A certain channel A has 5 devices, and the similarities between the devices are 0.8, 0.9, 0.85, 0.87, 0.91 respectively. Find their mean value, that is, the cohesion value x cluster :

[0096]

[0097] And the diversity value reflects the differences in user behaviors within the same channel. By calculating the differences between the device behaviors within the channel, the diversity index of this channel can be obtained. A higher diversity value indicates that the user behaviors in this channel cover different user needs, which may mean that the traffic sources of this channel are relatively wide. In this embodiment, the cohesion value is obtained by taking the standard deviation.

[0098] Similarly, for Channel A, the device behavior similarity values are 0.8, 0.9, 0.85, 0.87, 0.91. First, calculate the mean value of 0.866. Then, calculate the sum of the squared deviations of each device behavior similarity from the mean value. Finally, obtain the standard deviation of 0.039, which is the diversity value of this channel.

[0099] In this embodiment, by calculating the mean value and standard deviation of the device behavior similarity, the channel cohesion value and the channel diversity value are obtained respectively, thus effectively solving the problems of how to quantify the channel quality and evaluate the device behavior characteristics. Through the channel cohesion value and the channel diversity value, the performance characteristics of each channel can be better understood, and based on this, optimization decisions can be made to achieve more efficient management and resource allocation.

[0100] In some other embodiments, the channel cohesion value and the channel diversity value can also be obtained by classifying the device behavior through a clustering algorithm (such as K-means or DBSCAN), clustering similar device behaviors together.

[0101] The channel cohesion value and the channel diversity value can also be obtained based on methods such as spectral clustering of the similarity matrix, behavior path analysis method, behavior pattern matching and analytic hierarchy process, graph-based centrality analysis, behavior frequency distribution method, information entropy method, etc.

[0102] In S530, the channel behavior quality score provides a comprehensive and quantitative evaluation result, which can help enterprises select those excellent channels and avoid the interference of low-quality or false traffic channels. The scoring mechanism can help enterprises evaluate the channels from multiple dimensions (such as traffic consistency, breadth, diversity, etc.), and optimize the advertising investment and traffic selection according to the score, so as to improve the return on investment (ROI).

[0103] Specifically, in some embodiments, the specific implementation manner of step S530 can refer to Figure 4 . Figure 4 is based on Figure 2 the detailed description of step S530 in the channel behavior quality assessment method shown in the corresponding embodiment. In the channel behavior quality assessment method, step S530 may include the following steps:

[0104] S532, according to the channel cohesion benchmark value and the channel cohesion value corresponding to each channel, determine the channel behavior cohesion score corresponding to each channel, where the channel cohesion benchmark value is the cohesion value determined from the natural traffic.

[0105] S534, according to the channel diversity benchmark value and the channel diversity value corresponding to each channel, determine the channel behavior diversity score corresponding to each channel, where the channel diversity benchmark value is the diversity value determined from the natural traffic.

[0106] S536. Obtain the channel behavior quality scores corresponding to each channel based on the channel behavior cohesion scores and the channel behavior diversity scores corresponding to each channel.

[0107] In this embodiment, by introducing a channel benchmark value, the cohesion score and the diversity score of each channel are compared with the benchmark value determined in the natural traffic, so that the evaluation of channel behavior not only depends on the mean and standard deviation of internal data, but also takes into account the external environment (i.e., the performance in natural traffic). This method can effectively improve the reliability and pertinence of channel evaluation, and thus provide a more accurate basis for subsequent strategy optimization.

[0108] In S532, the channel cohesion benchmark value is the cohesion value obtained from the natural traffic data, representing a standard or ideal channel behavior cohesion value.

[0109] Specifically, in some embodiments, the specific implementation manner of step S532 can refer to the following embodiments. This embodiment is a detailed description of step S532 in the channel behavior quality evaluation method shown in the corresponding embodiment. In the channel behavior quality evaluation method, step S532 may include the following steps: Figure 4 Compare the magnitudes of the channel cohesion benchmark value and the channel cohesion values corresponding to each channel.

[0110] If the channel cohesion value corresponding to the channel is smaller than the channel cohesion benchmark value, use the channel cohesion value as the cohesion behavior amplitude difference.

[0111] If the channel cohesion value corresponding to the channel is larger than the channel cohesion benchmark value, use the difference between the unit value and the channel cohesion value as the cohesion behavior amplitude difference.

[0112] Determine the maximum cohesion behavior amplitude difference and the minimum cohesion behavior amplitude difference among the cohesion behavior amplitude differences corresponding to each channel.

[0113] Obtain the channel behavior cohesion scores according to the maximum cohesion behavior amplitude difference, the minimum cohesion behavior amplitude difference, and the channel cohesion values corresponding to each channel.

[0114]

[0115] ​In this embodiment, by introducing the amplitude difference mechanism, not only the numerical differences are considered, but also the situations where the cohesion values are too large or too small are handled, thus avoiding the deviation caused by simply relying on numerical comparison and improving the accuracy of evaluation. At the same time, by introducing the comparison of the maximum and minimum amplitude differences, different types of channels can be reasonably evaluated. Whether the cohesion value is greater than or less than the benchmark value, it can be appropriately adjusted. That is, in this embodiment, by dynamically calculating the amplitude difference, it can adapt to the performance of different channels, effectively cope with different types of data and distributions that may be encountered in the actual environment, and provide more practical evaluation results.

[0116] Specifically, the channel behavior cohesion score can be obtained by the following formula:

[0117]

[0118]

[0119] where x cluster is the channel cohesion value corresponding to each channel, std cluster is the channel cohesion benchmark value, flag = 'up' indicates the situation where the channel cohesion value is greater than the channel cohesion benchmark value, flag = 'down' indicates the situation where the channel cohesion value is less than the channel cohesion benchmark value, margin cluster is the cohesion behavior amplitude difference, is the maximum cohesion behavior amplitude difference in the case of flag = 'uo', is the minimum cohesion behavior amplitude difference in the case of flag = 'up', is the maximum cohesion behavior amplitude difference in the case of flag = 'down', is the minimum cohesion behavior amplitude difference in the case of flag = 'down', Qscore cluster is the channel behavior cohesion score.

[0120] In the above formula, when the cohesion value of a channel is less than the benchmark value, this may indicate that the device behavior consistency of this channel is poor. At this time, using the cohesion value itself as the amplitude difference can more accurately reflect the magnitude of the gap. When the cohesion value of a channel is greater than the benchmark value, it generally means that the device behavior consistency of this channel is strong. At this time, not only the magnitude of the gap needs to be reflected, but also excessive amplification of the gap should be avoided. Therefore, the difference is calculated by the method of "unit value minus cohesion value" to balance this situation. At the same time, by distinguishing different situations of flag = 'up' and flag = 'down' to calculate the amplitude difference of cohesion behavior, the evaluation of each channel not only considers its difference from the benchmark value, but also can reasonably handle the situation where the cohesion value is too large or too small, thus more comprehensively reflecting the quality of channel behavior.

[0121] In some other embodiments, the cohesion score of channel behavior can also be directly obtained from the ratio of the channel cohesion value to the channel cohesion benchmark value, or directly obtained from the difference between the channel cohesion value and the channel cohesion benchmark value.

[0122] For example, in some embodiments, the cohesion score of channel behavior can be obtained by the following formula:

[0123]

[0124] where x cluster is the channel cohesion value corresponding to each channel, std cluster is the channel cohesion benchmark value, and Qscore cluster is the cohesion score of channel behavior.

[0125] In S534, the channel diversity benchmark value is still the diversity standard value of channel behavior obtained from natural traffic data, representing the ideal diversity level.

[0126] Specifically, in some embodiments, the specific implementation manner of step S534 can refer to the following embodiments. This embodiment is a detailed description of step S534 in the channel behavior quality evaluation method shown in the corresponding embodiment. In the channel behavior quality evaluation method, step S534 may include the following steps: Figure 4 Compare the magnitude of the channel diversity benchmark value and the channel diversity value corresponding to each channel.

[0127] If the channel diversity value corresponding to this channel is smaller than the channel diversity benchmark value, then use this channel diversity value as the diversity behavior amplitude difference.

[0128]

[0129] ​If the channel diversity value corresponding to the channel is greater than the channel diversity benchmark value, the difference between the unit value and the channel diversity value is used as the diversity behavior amplitude difference.

[0130] Determine the maximum diversity behavior amplitude difference and the minimum diversity behavior amplitude difference among the diversity behavior amplitude differences corresponding to each channel.

[0131] Based on the maximum diversity behavior amplitude difference, the minimum diversity behavior amplitude difference, and the channel diversity values corresponding to each channel, obtain the channel behavior diversity score.

[0132] In this embodiment, by comparing the magnitudes of the channel diversity value and the channel diversity benchmark value, the diversity differences of each channel can be quantified, avoiding over-amplifying or ignoring the differences. Further, through the calculation of the amplitude difference and the determination of the maximum and minimum differences, a diversity score is finally obtained. The behavior diversity score calculated through this difference can accurately reflect the advantages and disadvantages of the channel in terms of diversity, and can accurately and efficiently evaluate the quality of channel behavior. The behavior diversity score is a comprehensive evaluation score, which helps to optimize the behavior management and decision-making of the channel.

[0133] Specifically, the channel behavior diversity score can be obtained by the following formula:

[0134]

[0135] where x diversity is the channel diversity value corresponding to each channel, std diversity is the channel diversity benchmark value, flag = 'up' represents the case where the channel diversity value is greater than the channel diversity benchmark value, flag = 'down' represents the case where the channel diversity value is less than the channel diversity benchmark value, margin cluster is the diversity behavior amplitude difference, is the maximum diversity behavior amplitude difference in the case of flag = 'up', is the minimum diversity behavior amplitude difference in the case of flag = 'uo', is the maximum diversity behavior amplitude difference in the case of flag = 'down', is the minimum diversity behavior amplitude difference in the case of flag = 'down', Qscore diversity is the channel behavior diversity score.

[0136] In the above formula, if the diversity level of a channel is lower than the benchmark value, the difference is directly the diversity value of that channel, which is used as the measure of the amplitude of its diversity behavior, facilitating the rapid assessment of channels with lower diversity. When the diversity value of a channel is greater than the benchmark value, a unit difference is used for calculation, avoiding the over-inflation of differences higher than the benchmark and ensuring a reasonable and balanced evaluation result. At the same time, by differentiating the cases of flag = 'up' and flag = 'down' to calculate the difference in the amplitude of diversity behavior, the evaluation of each channel not only considers its difference from the benchmark value but also reasonably handles the situation where the diversity value is too large or too small, thus more comprehensively reflecting the quality of channel behavior.

[0137] In some other embodiments, the diversity score of channel behavior can also be directly obtained from the ratio of the channel diversity value to the channel diversity benchmark value, or directly from the difference between the channel diversity value and the channel diversity benchmark value.

[0138] For example, in some embodiments, the diversity score of channel behavior can be obtained from the following formula:

[0139]

[0140] where x diversity is the channel diversity value corresponding to each channel, std diversity is the channel diversity benchmark value, and Qscore diversity is the diversity score of channel behavior.

[0141] In S536, the cohesion score and the diversity score are combined to obtain a comprehensive channel behavior quality score. The comprehensive channel behavior quality score can be calculated through weighted average, multiplication, or other mathematical methods. The purpose is to give a more accurate comprehensive quality score for evaluating the performance of each channel through the evaluation of the two dimensions of cohesion and diversity, making the channel evaluation result more comparable and referenceable.

[0142] Specifically, in some embodiments, the specific implementation of step S536 can refer to the following embodiments. This embodiment is a detailed description of step S536 in the channel behavior quality evaluation method shown in the corresponding embodiment. In the channel behavior quality evaluation method, step S536 may include the following steps: Figure 4 Weighted sum the channel behavior cohesion score and the channel behavior diversity score corresponding to each channel to obtain the channel behavior quality score corresponding to each channel.

[0143]

[0144] In this embodiment, when evaluating the quality of channel behavior, a single dimension (cohesion or diversity) may not comprehensively reflect the overall performance of the channel. Cohesion mainly reflects the consistency or concentration of channel behavior, while diversity reflects the extensiveness or dispersion of channel behavior, and the two have different focuses. By means of weighted summation, the impacts of these two aspects can be comprehensively considered, making the evaluation results more comprehensive and accurate. At the same time, the impacts of the cohesion and diversity scores of different channels on quality evaluation may vary. Using weighted summation can flexibly adjust the weights of cohesion and diversity according to different business requirements, thereby achieving personalized evaluation.

[0145] Specifically, the channel behavior quality score can be obtained by the following formula:

[0146] Qscore = weight × Qscore cluster +(1 - weight) × Qscore diversity

[0147] where Qscore is the channel behavior quality score, weight is the weight ratio, Qscore cluster is the channel behavior cohesion score, and Qscore diversity is the channel behavior diversity score.

[0148] In the above formula, the weight ratio can be a preset fixed value, such as 0.4, 0.5, 0.6, etc.

[0149] In some other embodiments, the weight ratio can also be dynamically determined according to the channel benchmark value, the channel cohesion value, and the channel diversity value.

[0150] For example, the absolute value of the difference between the channel cohesion value and the channel benchmark value and the absolute value of the difference between the channel diversity value and the channel benchmark value can be normalized to obtain the weight.

[0151] Another example is that the ratio of the channel cohesion value to the channel benchmark value and the ratio of the channel diversity value to the channel benchmark value can be normalized to obtain the weight.

[0152] Still another example is that the channel benchmark value, the channel cohesion value, and the channel diversity can be input into the corresponding weight model, and the weight model outputs the corresponding weight value.

[0153] Figure 5 FIG. shows a flowchart of a channel quality evaluation method according to an embodiment of the present application. This channel quality evaluation method can be executed by a server, and the server can also be Figure 1 the server shown in Figure 5 As shown, this channel quality evaluation method at least includes:

[0154] S100. Obtain channel evaluation data, where the channel evaluation data includes channel new user data, channel retention data, channel revenue data, and channel behavior data.

[0155] S200. Determine the channel new user quality score based on the channel new user data.

[0156] S300. Determine the channel retention quality score based on the channel retention data.

[0157] S400. Determine the channel revenue quality score based on the channel revenue data.

[0158] S500. Determine the channel behavior quality score according to the channel behavior data by using the channel behavior quality evaluation method described above.

[0159] S600. Determine the channel quality evaluation score based on the channel new user quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

[0160] In the embodiments of the present application, by combining data from multiple dimensions such as new users, retention, revenue, and behavior, the problem that a single - dimension evaluation cannot comprehensively evaluate the channel quality is solved, making the channel quality evaluation more comprehensive and accurate. At the same time, by integrating the scores of different dimensions, the data of these dimensions are effectively combined, and a unified quality score is formed through a suitable weighting method, avoiding the complexity of calculation and data redundancy, and improving the evaluation efficiency and the practicality of the results. At the same time, the use of weighted summation can also flexibly adjust the weights of each dimension according to different business requirements, enabling the evaluation results to better adapt to specific business scenarios.

[0161] In S100, the acquisition of channel evaluation data usually depends on extracting relevant data from multiple data sources such as channel operation systems, user behavior analysis systems, and financial systems, which may include the number of user registrations, the number of active users, retention rates, revenue data, and specific behavior logs of the channels.

[0162] Among them, the channel new user data can reflect the ability to introduce new users and expand channels. The channel retention data can reflect the loyalty and long - term attractiveness of users to the channels. The channel revenue data can measure the economic benefits or profitability of the channels.

[0163] In S200, the new user quality score may be quantitatively scored based on factors such as the number of new users, user growth rate, user source, etc., in combination with certain criteria or models. Usually, this data will be compared with historical data to evaluate the stability and sustainability of the new user acquisition ability.

[0164] Specifically, in some embodiments, the specific implementation of step S200 can be referred to the following embodiments. This embodiment is based on Figure 5 the detailed description of step S200 in the channel behavior quality assessment method shown in the corresponding embodiment. In the channel behavior quality assessment method, step S200 may include the following steps:

[0165] Determine the newly added score of the new channel and the newly added score of the existing channel respectively according to the newly added data of the channel.

[0166] Determine the newly added quality score of the channel according to the newly added score of the new channel and the newly added score of the existing channel.

[0167] In this embodiment, by distinguishing the newly added scores of the new channel and the existing channel and calculating the newly added scores of the new channel and the existing channel respectively, the newly added quality of different types of channels can be evaluated more precisely, avoiding the errors caused by the unified evaluation of channel quality, and also being able to more carefully reflect the actual performance of different channel types in terms of newly added data, thus improving the scientificity, accuracy and operability of the evaluation results.

[0168] The new channel generally refers to a newly added channel. The quality evaluation of its newly added data may be affected by various factors, such as the initial activity of users, the growth rate of data, the integrity of data, etc. According to the above factors, the corresponding newly added score of the new channel is obtained to reflect the quality of the newly added data of the new channel. For example, some key indicators (such as the growth rate of data, the number of active users, etc.) can be set to measure the quality of the newly added data, so as to obtain the newly added score of the new channel.

[0169] The existing channels usually refer to the channels that have existed for some time. The quality of their newly added data may be different from that of the new channels. The newly added score of the existing channels may be affected by the long-term operation effect of the channels. Therefore, it is necessary to evaluate the quality of their newly added data according to the historical data of the existing channels, the comparison between the newly added data and their original data, etc.

[0170] For example, factors such as the comparison between the newly added data of the channel and its past performance, the growth rate of data, and user feedback can be compared to obtain the newly added score of the existing channel.

[0171] By comprehensively considering the newly added scores of the new channel and the existing channel, the final newly added quality score of the channel is determined. Specifically, it can be achieved through methods such as weighted average, ranking method, segmented scoring, etc. The specific calculation method depends on the specific implementation plan.

[0172] In S300, the retention quality score usually depends on indicators such as user activity, usage frequency, and retention rate. According to the continuous retention situation of users within a certain time period, the retention quality score of the channel can be calculated. The model may use different time periods (such as 1 week, 1 month, 1 year) to calculate the retention rate.

[0173] Specifically, in some embodiments, the specific implementation of step S300 can refer to the following embodiments. This embodiment is based on Figure 5 the detailed description of step S300 in the channel behavior quality evaluation method shown in the corresponding embodiment. In the channel behavior quality evaluation method, step S300 may include the following steps:

[0174] Determine the next-day retention rate, weekly retention rate, and long-term retention rate according to the channel retention data.

[0175] Determine the channel retention quality score according to the next-day retention rate, the weekly retention rate, and the long-term retention rate.

[0176] In this embodiment, the retention data can help the channel operator discover the time nodes or stages with a low retention rate, and then take targeted optimization measures to improve the overall performance of the channel. By analyzing the next-day retention rate, weekly retention rate, and long-term retention rate of the channel, the retention situation of channel users can be measured from multiple perspectives. This multi-dimensional evaluation method can accurately reflect the user stickiness and long-term value of the channel at different time stages, avoiding the deviation that may be brought by single-dimensional indicators. The retention rates in different time dimensions can reflect the channel performance at different levels. For example, the next-day retention rate focuses on the short-term stickiness of users, the weekly retention rate reflects the medium-term user retention ability of the channel, and the long-term retention rate can indicate the long-term operation effect of the channel. By integrating the data of these dimensions, a more comprehensive and reliable channel retention quality score can be obtained.

[0177] Among them, the next-day retention rate usually refers to the proportion of new users who continue to use the channel on the second day after installation or the first contact with the channel. It reflects the initial interest and short-term stickiness of users in the channel. A high next-day retention rate means that the channel has a strong short-term attraction to users. The weekly retention rate represents the proportion of users who continue to use the channel within a specific week after the first use of the channel. The weekly retention rate can better reflect the medium-term stickiness of users to the channel. It not only considers whether users maintain their interest in the channel, but also can evaluate the long-term attraction of the channel content or service to users. The long-term retention rate usually refers to the proportion of users who remain active after a long time (such as several months) after the first contact with the channel. This indicator reflects the long-term value of the channel. A higher long-term retention rate usually means that the channel has a good performance in continuous optimization and user experience.

[0178] Specifically, each retention rate can be standardized to eliminate the scale differences between data, enabling fair comparison of retention rates across different time dimensions. Then, according to business requirements, different weights may be assigned to retention rates of different dimensions. For example, short-term retention rates (such as the next-day retention rate) may be more important for certain industries, while long-term retention rates can better reflect the overall stability of a channel. Therefore, different weight coefficients can be set based on channel characteristics and goals. Finally, by combining the weighted retention rate data, methods such as weighted average, segmented scoring, and ranking can be used to comprehensively calculate the final retention quality score of the channel. The finally obtained channel retention quality score, as part of the channel behavior quality assessment, provides quantitative feedback on its long-term performance and potential.

[0179] In S400, the calculation of the revenue quality score is usually based on the direct economic benefits of the channel, such as indicators like sales volume, profit, and cost-effectiveness. Revenue data may be sourced from sales platforms, financial systems, etc., and the score will consider factors such as the growth, stability, and cost-effectiveness of the revenue.

[0180] Specifically, in some embodiments, the specific implementation manner of step S400 can refer to the following embodiments. This embodiment is based on Figure 5 the detailed description of step S400 in the channel behavior quality assessment method shown in the corresponding embodiment. In the channel behavior quality assessment method, step S400 may include the following steps:

[0181] Determine the active device revenue, channel revenue, and channel cost based on the channel revenue data.

[0182] Determine the channel revenue quality score based on the active device revenue, the channel revenue, and the channel cost.

[0183] In this embodiment, by considering multiple dimensions such as active device revenue, channel revenue, and channel cost and reasonably combining the data, a more comprehensive revenue quality assessment can be provided, reducing biases. At the same time, the revenue quality of the channel can be accurately quantified, thus providing more scientific and actionable assessment results for decision-makers. This embodiment can identify which channels have high revenue quality and which channels may have problems such as ineffective cost control or unstable revenue, thereby helping operators optimize the channel structure and resource allocation.

[0184] Specifically, active device revenue refers to the revenue contributed by active devices related to a channel (such as user terminal devices), usually considering factors such as device activity and usage, such as the online duration and usage frequency of the device. Channel revenue is the overall revenue generated by all devices through this channel, usually including revenue generated through sales, advertising, or other business models. Channel cost is all the costs incurred for the operation of this channel, including data such as device investment, maintenance costs, and promotion costs that can reveal the operation efficiency of the channel and its cost control situation.

[0185] In S500, specifically, reference can be made to Figures 2 to 4 the channel behavior quality assessment method mentioned, and this specification will not elaborate on it here.

[0186] In S600, usually, in the way of weighted summation, scores such as new addition, retention, revenue, and behavior are weighted according to preset weights, and finally the overall quality score of the channel is obtained. The setting of the weighting coefficient can be adjusted according to business requirements to highlight the importance of certain dimensions. For example, for some rapidly expanding channels, the new addition score may account for a larger proportion; for some channels with long-term operation, the retention and revenue scores may be more critical.

[0187] Specifically, in some embodiments, the specific implementation manner of step S600 can refer to the following embodiments. This embodiment is based on Figure 5 the detailed description of step S600 in the channel behavior quality assessment method shown in the corresponding embodiment. In the channel behavior quality assessment method, step S600 may include the following steps:

[0188] Perform weighted summation on the channel new addition quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score to obtain the channel behavior quality assessment score.

[0189] In this embodiment, by performing weighted summation on the quality scores of different dimensions (new addition quality, retention quality, revenue quality, and behavior quality), the comprehensive performance of the channel can be evaluated more comprehensively and accurately. This method effectively solves the limitations of single-index evaluation, and through flexible weight allocation, the final evaluation result better meets the actual operation requirements, thus providing more scientific support for decision-makers and helping to optimize channel management and resource allocation.

[0190] In some embodiments, the weighting coefficients of each dimension can be preset fixed empirical values, such as 0.25, 0.2, 0.25, 0.3.

[0191] In other embodiments, the weighting coefficients of each dimension can also be dynamically determined according to the channel benchmark value, the channel cohesion value, and the channel diversity value.

[0192] For example, the weight coefficient of each dimension can be determined according to the mean and dispersion degree of the quality scores of each dimension, and the weight coefficient of a dimension is positively correlated with its dispersion degree.

[0193] The following introduces the device embodiments of the present application, which can be used to execute the channel behavior quality evaluation method and the channel quality evaluation method in the above embodiments of the present application. For the details not disclosed in the device embodiments of the present application, please refer to the embodiments of the above channel behavior quality evaluation method and channel quality evaluation method of the present application.

[0194] Figure 6 The block diagram of a channel behavior quality evaluation device according to an embodiment of the present application is shown.

[0195] Refer to Figure 6 As shown, a channel behavior quality evaluation device 150 according to an embodiment of the present application includes: a device similarity determination module 151, a cohesion diversity determination module 152, and a behavior score determination module 153.

[0196] Among them, the device similarity determination module 151 is used to obtain the device behavior similarity between multiple devices corresponding to each channel respectively according to the channel behavior data of each channel; the cohesion diversity determination module 152 is used to determine the channel cohesion value and the channel diversity value corresponding to each channel according to the device behavior similarities corresponding to each channel; the behavior score determination module 153 is used to determine the channel behavior quality score corresponding to each channel according to the channel benchmark value, the channel cohesion value, and the channel diversity value corresponding to each channel.

[0197] In some embodiments of the present application, the device similarity determination module specifically includes: a behavior characterization extraction sub-module, which is used to extract device behavior characterizations from the channel behavior data of each channel respectively through a behavior characterization extraction model to obtain device behavior chains; a device same-dimension processing sub-module, which is used to perform same-dimension processing on each device behavior chain corresponding to each channel respectively to obtain corresponding same-dimension device behavior chains; a device similarity determination sub-module, which is used to compare each same-dimension device behavior chain corresponding to each channel respectively to obtain the device behavior similarity between devices.

[0198] In some embodiments of the present application, the device same-dimension processing sub-module specifically includes: a behavior chain complementing unit, which is used to perform complementing processing on each device behavior chain corresponding to each channel respectively to obtain complemented device behavior chains; a same-position alignment unit, which is used to perform same-position alignment processing on each device behavior chain corresponding to each channel respectively to obtain same-dimension device behavior chains.

[0199] In some embodiments of the present application, the behavior characterization extraction model includes an encoder and a decoder. The channel behavior quality evaluation device further includes: a sample acquisition module, configured to acquire a set of device behavior data sequence samples, where the set of device behavior data sequence samples contains multiple device behavior data sequence samples, and each device behavior data sequence sample is marked with a corresponding behavior characterization sequence; a sample encoding module, configured to input the device behavior data sequence sample into the encoder to obtain a behavior sequence characterization; a sample decoding module, configured to input the behavior sequence characterization into the decoder to obtain a behavior characterization vector; a loss function module, configured to determine a loss function based on the finally obtained behavior characterization vector and the marked behavior characterization sequence; a parameter update module, configured to update the parameters of the encoder and the decoder based on the loss function until the loss function converges.

[0200] In some embodiments of the present application, the cohesion and diversity determination module specifically includes: a channel cohesion value sub-module, configured to calculate the average value of the device behavior similarities corresponding to each channel respectively to obtain a channel cohesion value; a channel diversity value sub-module, configured to calculate the standard deviation of the device behavior similarities corresponding to each channel respectively to obtain a channel diversity value.

[0201] In some embodiments of the present application, the channel benchmark value includes a channel cohesion benchmark value and a channel diversity benchmark value. The behavior score determination module specifically includes: a behavior cohesion score sub-module, configured to determine the channel behavior cohesion score corresponding to each channel according to the channel cohesion benchmark value and the channel cohesion value corresponding to each channel, where the channel cohesion benchmark value is the cohesion value determined from the natural traffic; a behavior diversity score sub-module, configured to determine the channel behavior diversity score corresponding to each channel according to the channel diversity benchmark value and the channel diversity value corresponding to each channel, where the channel diversity benchmark value is the diversity value determined from the natural traffic; a channel behavior quality score sub-module, configured to obtain the channel behavior quality score corresponding to each channel according to the channel behavior cohesion score corresponding to each channel and the channel behavior diversity score corresponding to each channel.

[0202] In some embodiments of the present application, the sub-module for the cohesion of behaviors specifically includes: a cohesion benchmark comparison unit, configured to compare the magnitude of the cohesion benchmark value of the channels and the cohesion values of the corresponding channels of each channel; a first cohesion amplitude unit, configured to use the cohesion value of the channel as the cohesion behavior amplitude difference if the cohesion value of the corresponding channel is smaller than the cohesion benchmark value of the channel; a second cohesion amplitude unit, configured to use the difference between the unit value and the cohesion value of the channel as the cohesion behavior amplitude difference if the cohesion value of the corresponding channel is larger than the cohesion benchmark value of the channel; a cohesion amplitude extreme value unit, configured to determine the maximum cohesion behavior amplitude difference and the minimum cohesion behavior amplitude difference among the cohesion behavior amplitude differences corresponding to each channel; a cohesion score calculation unit, configured to obtain the channel behavior cohesion score according to the maximum cohesion behavior amplitude difference, the minimum cohesion behavior amplitude difference, and the cohesion values of the corresponding channels of each channel.

[0203] In some embodiments of the present application, the sub-module for the diversity of behaviors specifically includes: a diversity benchmark comparison unit, configured to compare the magnitude of the diversity benchmark value of the channels and the diversity values of the corresponding channels of each channel; a first diversity amplitude unit, configured to use the diversity value of the channel as the diversity behavior amplitude difference if the diversity value of the corresponding channel is smaller than the diversity benchmark value of the channel; a second diversity amplitude unit, configured to use the difference between the unit value and the diversity value of the channel as the diversity behavior amplitude difference if the diversity value of the corresponding channel is larger than the diversity benchmark value of the channel; a diversity amplitude extreme value unit, configured to determine the maximum diversity behavior amplitude difference and the minimum diversity behavior amplitude difference among the diversity behavior amplitude differences corresponding to each channel; a diversity score calculation unit, configured to obtain the channel behavior diversity score according to the maximum diversity behavior amplitude difference, the minimum diversity behavior amplitude difference, and the diversity values of the corresponding channels of each channel.

[0204] In some embodiments of the present application, the sub-module for the quality of channel behaviors specifically includes: a channel behavior quality score unit, configured to perform a weighted sum of the channel behavior cohesion scores and the channel behavior diversity scores corresponding to each channel to obtain the channel behavior quality scores corresponding to each channel.

[0205] In this embodiment, by analyzing the similarity of device behaviors, the consistency of traffic within channels is evaluated to help identify false traffic. At the same time, through two dimensions of behavior consistency and behavior difference, combined with a comparison with a benchmark value, a comprehensive quality score for each channel is obtained, which can more comprehensively analyze the quality of channel traffic, help make optimization decisions, effectively solve the technical problems that cannot be refined by traditional evaluation methods, enrich the quality evaluation system of the buying channels by introducing behavior-related factors such as behavior data and multi-dimensional evaluation indicators, maximize the accuracy of evaluation, optimize the channel selection process, and thus improve the scientific nature and efficiency of advertising placement and promotion effects.

[0206] Figure 7 FIG. shows a block diagram of a channel quality evaluation system according to an embodiment of the present application.

[0207] Refer to Figure 7 As shown, a channel quality evaluation system 100 according to an embodiment of the present application includes: a channel quality data acquisition device 110, a channel new user quality evaluation device 120, a channel retention quality evaluation device 130, a channel revenue quality evaluation device 140, a channel behavior quality evaluation device 150, and a channel quality evaluation score device 160.

[0208] Among them, the channel quality data acquisition device 110 is used to acquire channel evaluation data, and the channel evaluation data includes channel new user data, channel retention data, channel revenue data, and channel behavior data; the channel new user quality evaluation device 120 is used to determine a channel new user quality score according to the channel new user data; the channel retention quality evaluation device 130 is used to determine a channel retention quality score according to the channel retention data; the channel revenue quality evaluation device 140 is used to determine a channel revenue quality score according to the channel revenue data; the channel behavior quality evaluation device 150 is used to determine a channel behavior quality score according to the channel behavior data by using the above-mentioned channel behavior quality evaluation method, which is the Figure 6 shown channel behavior quality evaluation device 150; the channel quality evaluation score device 160 is used to determine a channel quality evaluation score according to the channel new user quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

[0209] In some embodiments of the present application, the channel quality data acquisition device specifically includes: a channel new user score module, which is used to respectively determine a new channel new user score and an existing channel new user score according to the channel new user data; a new user quality score module, which is used to determine a channel new user quality score according to the new channel new user score and the existing channel new user score.

[0210] In some embodiments of the present application, the channel retention quality evaluation device specifically includes: a retention parameter determination module, configured to determine the next-day retention rate, weekly retention rate, and long-term retention rate according to the channel retention data; and a retention quality score module, configured to determine the channel retention quality score according to the next-day retention rate, the weekly retention rate, and the long-term retention rate.

[0211] In some embodiments of the present application, the channel revenue quality evaluation device specifically includes: a revenue-cost determination module, configured to determine the active device revenue, channel revenue, and channel cost according to the channel revenue data; and a revenue quality score module, configured to determine the channel revenue quality score according to the active device revenue, the channel revenue, and the channel cost.

[0212] In some embodiments of the present application, the channel quality evaluation score device specifically includes: a channel quality evaluation score module, configured to perform a weighted sum of the channel new user quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score to obtain the channel behavior quality evaluation score.

[0213] In the embodiments of the present application, by combining data from multiple dimensions such as new user acquisition, retention, revenue, and behavior, the problem that a single-dimensional evaluation cannot comprehensively evaluate the channel quality is solved, making the channel quality evaluation more comprehensive and accurate. At the same time, by integrating the scores of different dimensions, the data of these dimensions are effectively combined, and a unified quality score is formed through a suitable weighting method, avoiding the complexity of calculation and data redundancy, and improving the evaluation efficiency and the practicality of the results. At the same time, the use of the weighted sum method can also flexibly adjust the weights of each dimension according to different business requirements, enabling the evaluation results to better adapt to specific business scenarios.

[0214] Figure 8 The structural schematic diagram of a computer system of an electronic device suitable for implementing the embodiments of the present application is shown.

[0215] It should be noted that Figure 8 The computer system of the electronic device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present application.

[0216] Such as Figure 8As shown, the computer system includes a Central Processing Unit (CPU) 1801, which can perform various appropriate actions and processes according to the program stored in the Read-Only Memory (ROM) 1802 or the program loaded from the storage section 1808 into the Random Access Memory (RAM) 1803, such as executing the method described in the above embodiments. In the RAM 1803, various programs and data required for system operation are also stored. The CPU 1801, ROM 1802, and RAM 1803 are connected to each other via a bus 1804. An Input / Output (I / O) interface 1805 is also connected to the bus 1804.

[0217] The following components are connected to the I / O interface 1805: an input section 1806 including a keyboard, a mouse, etc.; an output section 1807 including, for example, a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), etc. and a speaker, etc.; a storage section 1808 including a hard disk, etc.; and a communication section 1809 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1809 performs communication processing via a network such as the Internet. A drive 1810 is also connected to the I / O interface 1805 as needed. A removable medium 1811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1810 as needed so that a computer program read from it can be installed into the storage section 1808 as needed.

[0218] Specifically, according to the embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments of the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via the communication section 1809, and / or installed from the removable medium 1811. When the computer program is executed by the Central Processing Unit (CPU) 1801, various functions defined in the system of the present application are executed.

[0219] It should be noted that the computer-readable medium shown in the embodiments of the present application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0220] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. Among them, each block in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram or flowchart, as well as the combination of blocks in the block diagram or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0221] The units involved in the embodiments of this application can be implemented in software or in hardware, and the described units can also be provided in a processor. Among them, the names of these units do not constitute a limitation to the unit itself in certain cases.

[0222] On the other hand, this application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments; or may exist alone without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiments.

[0223] This specification also provides a computer program product, which stores at least one instruction, and the at least one instruction is loaded and executed by the processor to perform the method as described in the above Figures 1 to 5 illustrated embodiments. For the specific execution process, reference may be made to the specific description of the illustrated embodiments, which will not be elaborated here. Figures 1 to 5 illustrated embodiments. For the specific execution process, reference may be made to the specific description of the illustrated embodiments, which will not be elaborated here.

[0224] It should be noted that although several modules or units of a device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of the two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0225] Through the description of the above embodiments, those skilled in the art can easily understand that the exemplary embodiments described here can be implemented in software or in the form of software combined with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (such as a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiments of this application.

[0226] After considering the specification and practicing the disclosed embodiments here, those skilled in the art will readily think of other implementation schemes of this application. This application aims to cover any variations, uses, or adaptive changes of this application, which follow the general principles of this application and include common general knowledge or conventional technical means in the technical field not disclosed in this application.

[0227] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present application is only limited by the appended claims.

Claims

1. A channel behavior quality assessment method, characterized in that: The channel behavior quality assessment method comprises: According to the channel behavior data of each channel, the device behavior similarity between multiple devices corresponding to the channel is obtained; Determine the channel cohesion value and channel diversity value corresponding to each channel according to the similarity of the behaviors of each device corresponding to each channel; According to the channel benchmark value, channel cohesion value and channel diversity value corresponding to each channel, the channel behavior quality score corresponding to each channel is determined.

2. The channel behavior quality assessment method according to claim 1, characterized in that: The obtaining, based on the channel behavior data of each channel, the device behavior similarity between the multiple devices corresponding to the channel specifically includes: The device behavior representation is extracted from the channel behavior data of each channel through the behavior representation extraction model to obtain the device behavior chain; Perform same-dimension processing on each device behavior chain corresponding to each channel to obtain the corresponding same-dimension device behavior chain; The behavior chains of the devices of the same dimension corresponding to each channel are compared respectively to obtain the similarity of the device behaviors between the devices.

3. The channel behavior quality assessment method according to claim 2, characterized in that: The same-dimensional processing is performed on each device behavior chain corresponding to each channel to obtain the corresponding same-dimensional device behavior chain, specifically including: Complete each device behavior chain corresponding to each channel to obtain a completed device behavior chain; The device behavior chains corresponding to each channel are aligned at the same position to obtain the device behavior chains of the same dimension.

4. The channel behavior quality assessment method according to claim 2, characterized in that: The behavior representation extraction model includes an encoder and a decoder, and the method further includes: Acquire a device behavior data sequence sample set, wherein the device behavior data sequence sample set includes a plurality of device behavior data sequence samples, and each of the device behavior data sequence samples is marked with a corresponding behavior representation sequence; Inputting the device behavior data sequence samples into an encoder to obtain a behavior sequence representation; Inputting the behavior sequence representation into the decoder to obtain a behavior representation vector; Determine a loss function based on the finally obtained behavior representation vector and the marked behavior representation sequence; Parameters of the encoder and the decoder are updated based on the loss function until the loss function converges.

5. The channel behavior quality assessment method according to claim 1, characterized in that: Determining the channel cohesion value and channel diversity value corresponding to each channel according to the similarity of the behaviors of each device corresponding to each channel specifically includes: Taking the average of the device behavior similarities corresponding to each channel to obtain the channel cohesion value; The standard deviation of the device behavior similarity corresponding to each channel is taken to obtain the channel diversity value.

6. A channel quality assessment method, characterized in that: The channel behavior quality assessment method comprises: Obtaining channel evaluation data, wherein the channel evaluation data includes channel new data, channel retention data, channel revenue data, and channel behavior data; Determine a new quality score of the channel based on the new data of the channel; Determining a channel retention quality score based on the channel retention data; Determining a channel revenue quality score based on the channel revenue data; Determine a channel behavior quality score according to the channel behavior data using the channel behavior quality evaluation method according to any one of claims 1 to 5; A channel quality assessment score is determined based on the channel new addition quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

7. A channel behavior quality assessment device, characterized in that: The channel behavior quality assessment device comprises: A device similarity determination module, used to obtain device behavior similarities between multiple devices corresponding to each channel based on the channel behavior data of each channel; A cohesion and diversity determination module, used to determine a channel cohesion value and a channel diversity value corresponding to each channel according to the similarity of the behaviors of each device corresponding to each channel; The behavior score determination module is used to determine the channel behavior quality score corresponding to each channel according to the channel benchmark value, channel cohesion value and channel diversity value corresponding to each channel.

8. A channel quality evaluation system, characterized in that: The channel quality evaluation system comprises: A channel quality data acquisition device, used to acquire channel evaluation data, wherein the channel evaluation data includes channel new data, channel retention data, channel revenue data and channel behavior data; A channel addition quality assessment device, used to determine a channel addition quality score based on the channel addition data; A channel retention quality evaluation device, used to determine a channel retention quality score based on the channel retention data; A channel revenue quality evaluation device, used to determine a channel revenue quality score based on the channel revenue data; A channel behavior quality evaluation device, configured to determine a channel behavior quality score based on the channel behavior data using the channel behavior quality evaluation method according to any one of claims 1 to 5; A channel quality evaluation scoring device is used to determine a channel quality evaluation score based on the channel new addition quality score, the channel retention quality score, the channel revenue quality score, and the channel behavior quality score.

9. A computer readable medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the channel behavior quality evaluation method according to any one of claims 1 to 5 or the channel quality evaluation method according to claim 6 is implemented.

10. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the channel behavior quality assessment method as described in any one of claims 1 to 5 or the channel quality assessment method as described in claim 6.

Images