A method and system for enterprise information advertising push based on big data

Abstract

This application provides a method and system for enterprise information and advertising push based on big data. It sets multiple recommendation levels for online information and advertisements, divides each recommendation level into intervals, determines user behavior feature values ​​for each recommendation level interval, determines boundary data estimation factors for the corresponding recommendation level interval based on these feature values, collects user behavior datasets of target users for a type of online information and advertisement within a preset time period, estimates the membership degree of the user behavior dataset based on the boundary data estimation factors for each recommendation level interval to obtain a behavior membership matrix, determines the recommendation level of the target user for that type of online information and advertisement using the behavior membership matrix, and recommends online information and advertisements to the target user based on these recommendation levels. This achieves the segmentation of user behavior data at the boundary between two recommendation level intervals.

Description

Technical Field

[0001] This application relates to the field of network information and advertising recommendation technology. Specifically, this application relates to a method and system for pushing enterprise information advertisements based on big data. Background Technology

[0002] Online news and advertising recommendations are based on user interests and behavioral data, recommending content and ads that users are likely to be interested in. Common methods for pushing news and ads include collaborative filtering and content-based recommendation. These methods all collect user behavior datasets and divide them into different clusters to determine the recommendation level of online news and ads for the target user. The distribution of user behavior datasets within a recommendation level interval typically follows a normal distribution. However, according to normal distribution theory, a small number of data points will fall outside the predetermined interval at the extremes, making user behavior data at the boundary between two recommendation level intervals easily misclassified. Being misclassified into the wrong recommendation level range is particularly serious for companies with large amounts of data. This is because the larger the data volume, the greater the number of users falling into the incorrect range (as the absolute number of tails in a normal distribution increases with the total data volume; for example, in 1 million users, 5% of the boundary users means 50,000 people might be misclassified, while in 100,000 users, only 5,000 might be misclassified). This can lead to: users receiving irrelevant recommendations, resulting in a poor user experience (such as decreased ad click-through rates); and companies losing potential revenue (misclassification prevents high-value users from being accurately targeted). Therefore, how to classify user behavior data at the boundary between two recommendation level ranges has become a major challenge for big data companies. Summary of the Invention

[0003] This application provides a method and system for pushing enterprise information advertisements based on big data, which can divide user behavior data at the boundary between two recommendation level intervals.

[0004] Firstly, this application provides a method for pushing enterprise information advertisements based on big data, comprising the following steps:

[0005] Set multiple recommendation levels for online information and advertisements, and divide each recommendation level into recommendation level ranges;

[0006] Determine the user behavior feature value for each recommendation level interval, and determine the boundary data estimation factor for the corresponding recommendation level interval based on the user behavior feature value for each recommendation level interval;

[0007] Collect a dataset of user behavior data of target users regarding a type of online information and advertisements within a preset time period;

[0008] The membership degree of the user behavior dataset is estimated based on the boundary data estimation factor for each recommendation level interval, resulting in a behavior membership matrix;

[0009] The behavior membership matrix is ​​used to determine the recommendation level of the online information and advertisements for the target user, and the recommendation level is used to recommend online information and advertisements to the target user.

[0010] In some embodiments, determining the boundary data estimation factor for each recommendation level interval based on the user behavior feature value of each recommendation level interval specifically includes:

[0011] Select a recommendation level range and determine the distribution steepness based on the user behavior feature values ​​of that recommendation level range;

[0012] The boundary data estimation factor for each recommendation level interval is determined based on the steepness of the distribution.

[0013] Continue to determine the boundary data estimation factors for the remaining recommendation level intervals.

[0014] In some embodiments, determining the boundary data estimation factor for the recommendation level interval based on the distribution steepness specifically includes:

[0015] The behavioral data distribution entropy for the recommendation level interval is determined based on the user behavior feature values ​​and the distribution steepness of that interval.

[0016] The boundary data estimation factor for each recommendation level interval is determined based on the behavioral data distribution entropy.

[0017] In some embodiments, the membership degree estimation of the user behavior dataset is performed based on the boundary data estimation factor for each recommendation level interval to obtain the behavior membership matrix, specifically including:

[0018] Obtain user behavior feature values ​​for each recommendation level range;

[0019] Obtain the user behavior dataset;

[0020] Obtain the boundary data estimation factors for each recommendation level interval;

[0021] Obtain the distribution steepness for each recommendation level interval;

[0022] Select a user behavior data point from the user behavior dataset, and determine the user behavior membership degree between the user behavior data point and each recommendation level interval based on the user behavior feature value of each recommendation level interval, the distribution steepness of each recommendation level interval, and the boundary data estimation factor.

[0023] Continue to determine the remaining user behavior data in the user behavior dataset and the user behavior membership degree between each recommendation level interval;

[0024] Determine the behavior membership matrix based on the membership degrees of all user behaviors.

[0025] In some embodiments, determining the recommendation level of such online information and advertisements for target users through the behavioral membership matrix specifically includes:

[0026] The recommendation incentive value corresponding to each user behavior data is determined based on the user behavior membership matrix.

[0027] Determine the recommended incentive vector based on all recommended incentive values;

[0028] The recommendation level of this type of online information and advertisement for the target user is determined based on the recommendation incentive vector and the user behavior membership matrix.

[0029] In some embodiments, determining the recommendation incentive value corresponding to each user behavior data based on the user behavior membership matrix, and thus obtaining the recommendation incentive vector, specifically includes:

[0030] The user behavior membership matrix is ​​standardized to obtain a standard user behavior matrix;

[0031] The standard user behavior matrix is ​​transformed by deviation to obtain the behavior deviation matrix;

[0032] The recommendation incentive value corresponding to each user behavior data is determined based on the behavior deviation matrix.

[0033] The recommendation incentive vector is determined based on all the recommended incentive values.

[0034] In some embodiments, recommending online information and advertisements to target users based on the recommendation level specifically includes:

[0035] The intermediate level is determined based on all preset recommended levels;

[0036] When the recommendation level of such online information and advertisements for the target user is higher than the intermediate level, the recommendation volume of such online information and advertisements is increased;

[0037] When the recommendation level of such online information and advertisements for a target user is lower than the intermediate level, the recommendation volume of such online information and advertisements shall be reduced.

[0038] Secondly, this application provides a big data-based enterprise information advertising push system, comprising:

[0039] The settings module is used to set multiple recommendation levels for online information and advertisements, and to divide each recommendation level into recommendation level ranges;

[0040] The determination module is used to determine the user behavior feature value for each recommendation level interval, and to determine the boundary data estimation factor for the corresponding recommendation level interval based on the user behavior feature value for each recommendation level interval.

[0041] The data collection module is used to collect a dataset of user behavior of target users towards a type of online information and advertisement within a preset time period.

[0042] The estimation module is used to estimate the membership degree of the user behavior dataset based on the boundary data estimation factors of each recommendation level interval, so as to obtain the behavior membership matrix;

[0043] The execution module is used to determine the recommendation level of the network information and advertisement for the target user through the behavior membership matrix, and to recommend network information and advertisement to the target user based on the recommendation level.

[0044] Thirdly, this application provides a computer electronic device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described method for pushing enterprise information advertisements based on big data.

[0045] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for pushing enterprise information advertisements based on big data.

[0046] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0047] In this embodiment, multiple recommendation levels for online information and advertisements are set, and recommendation level intervals are divided for each level. A normal distribution within a conventional finite interval will have a small number of data points distributed into adjacent intervals at the interval boundaries. Therefore, when the collected user behavior data is near the interval boundaries, the recognition effect for the recommendation level of that behavior data is poor, resulting in incorrect online information and advertisement recommendations. This application determines a boundary data estimation factor for each recommendation level interval. This boundary data estimation factor is half the length of the corrected recommendation level interval. Since the two corrected recommendation level intervals have overlapping areas, user behavior data in these overlapping areas may be incorrect. The data is categorized, and the user behavior dataset of the target users is collected. By estimating the membership degree of the user behavior dataset through boundary data estimation factors, the membership degree corresponding to each user behavior data and each recommendation level interval can be obtained, resulting in a behavior membership matrix. In particular, when estimating the membership degree of data at the intersection of two recommendation level intervals, the membership degree of the data with the theoretically normal distribution within the two recommendation level intervals will be considered simultaneously, thereby determining which interval the data at the boundary belongs to. This enhances the identification effect of user behavior data at the boundary of recommendation level intervals, and then the recommendation level is determined through the behavior membership matrix, realizing the division of user behavior data at the intersection of two recommendation level intervals.

[0048] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0049] Figure 1 This is an exemplary flowchart illustrating a big data-based enterprise information advertising push method according to some embodiments of this application;

[0050] Figure 2 This is a distribution diagram of user behavior membership according to some embodiments of this application;

[0051] Figure 3 This is an exemplary flowchart illustrating the recommendation of online information and advertisements to target users based on recommendation levels, according to some embodiments of this application;

[0052] Figure 4 These are schematic diagrams of exemplary hardware and / or software for a big data-based enterprise information advertising push system, as shown in some embodiments of this application.

[0053] Figure 5 This is a schematic diagram of the structure of a computer device for implementing a big data-based enterprise information advertising push method, according to some embodiments of this application. Detailed Implementation

[0054] This application establishes multiple recommendation levels for online news and advertisements, divides each recommendation level into intervals, determines user behavior feature values ​​for each recommendation level interval, determines boundary data estimation factors for the corresponding recommendation level interval based on the user behavior feature values, collects user behavior datasets of target users for a type of online news and advertisement within a preset time period, estimates the membership degree of the user behavior dataset based on the boundary data estimation factors for each recommendation level interval to obtain a behavior membership matrix, determines the recommendation level of the target user for that type of online news and advertisement through the behavior membership matrix, and recommends online news and advertisements to the target user based on the recommendation level. This allows for the segmentation of user behavior data at the boundary between two recommendation level intervals.

[0055] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0056] In step 101, multiple recommendation levels for online information and advertisements are set, and recommendation level ranges are divided for each recommendation level.

[0057] It should be noted that the multiple recommendation levels mentioned in this application can be 5 levels, with the first level being the very low recommendation level and the fifth level being the very high recommendation level. The recommendation level range can be divided into [0, 0.2), [0.2, 0.4), [0.4, 0.6), [0.6, 0.8), [0.8, 1.0]. In other embodiments, it can also be divided into other different recommendation levels and other different recommendation level ranges, which are not limited here.

[0058] In step 102, the user behavior feature value for each recommendation level interval is determined, and the boundary data estimation factor for the corresponding recommendation level interval is determined based on the user behavior feature value for each recommendation level interval.

[0059] In some embodiments, determining the boundary data estimation factor for each recommendation level interval based on the user behavior feature value of each recommendation level interval can be achieved using the following steps:

[0060] Select a recommendation level range and determine the distribution steepness based on the user behavior feature values ​​of that recommendation level range;

[0061] The boundary data estimation factor for each recommendation level interval is determined based on the steepness of the distribution.

[0062] Continue to determine the boundary data estimation factors for the remaining recommendation level intervals.

[0063] In practice, the median value of each recommendation level interval can be used as the user behavior feature value for that recommendation level interval.

[0064] It should be noted that the user behavior feature value described in this application is a value used to represent the user behavior data that is most likely to belong to each recommendation level range.

[0065] In some embodiments, determining the distribution steepness based on the user behavior feature values ​​of the recommendation level interval can be achieved through the following steps:

[0066] Determine the right boundary of this recommendation level range;

[0067] Obtain user behavior feature values ​​for this recommendation level range;

[0068] The steepness of the recommendation level interval is determined based on the right boundary of the interval and the user behavior feature value. The steepness of the interval can be determined according to the following formula:

[0069]

[0070] Among them, K i Let B be the steepness of the distribution of the i-th recommendation level interval. i max B is the right boundary of the i-th recommendation level interval. i+1 max E represents the right boundary of the (i+1)th recommendation level interval. i Let be the user behavior feature value for the i-th recommendation level interval, and lg0.5 represent the logarithm of 0.5 to the base 10. Indicates base 10 The logarithm of .

[0071] It should be noted that the distribution steepness mentioned in this application is a value used to measure the probability distribution density of the normal distribution of user behavior data within the recommendation level interval. The greater the distribution steepness, the greater the probability distribution density of the normal distribution of user behavior data within the recommendation level interval, and the steeper the probability distribution curve of the normal distribution of user behavior data within the recommendation level interval.

[0072] In some embodiments, determining the boundary data estimation factor for the recommendation level interval based on the distribution steepness can be achieved using the following steps:

[0073] The behavioral data distribution entropy for the recommendation level interval is determined based on the user behavior feature values ​​and the distribution steepness of that interval.

[0074] The boundary data estimation factor for the recommendation level interval is determined based on the behavioral data distribution entropy, wherein the boundary data estimation factor can be determined according to the following formula:

[0075]

[0076] Wherein, normrnd(σ) i H i ) represents generating a σ i Let H be the expected value, H be the variance of the normally distributed random number, and H be the preset correction precision. i σ is the boundary data estimation factor for the i-th recommendation level interval. i K is the behavioral data distribution entropy for the i-th recommendation level interval. i It is the steepness of the distribution of the i-th recommendation level interval.

[0077] It should be noted that the boundary data estimation factor in this application is a parameter used to correct the classification of data at the boundary of each recommendation level interval. In specific implementation, for example, the boundary data estimation factor is half the length of the corrected recommendation level interval. Since there is an overlapping area between the two corrected recommendation level intervals, in subsequent steps, when determining the membership degree of user behavior data distributed in the overlapping area, the membership degree of that data with the theoretically normal distribution in the two recommendation level intervals will be considered at the same time.

[0078] Additionally, it should be noted that the correction precision mentioned in this application is a parameter used to control the accuracy of correcting the boundaries of the recommendation level range. The closer the correction precision value is to 0, the more accurate the recommendation results of online information and advertisements. In this application, the correction precision can be set to 0.001. In other embodiments, it can also be set to other correction precision values, which are not limited here.

[0079] In some embodiments, determining the behavioral data distribution entropy of a recommendation level interval based on the user behavior feature value and the distribution steepness can be achieved using the following steps:

[0080] Obtain the user behavior feature values ​​and distribution steepness for this recommendation level range;

[0081] Determine the left boundary of this recommendation level range;

[0082] The behavioral data distribution entropy of the recommendation level interval is determined based on the left boundary of the interval, the user behavior feature value of the interval, and the distribution steepness. The behavioral data distribution entropy can be determined as follows:

[0083]

[0084] Where, σ i E is the behavioral data distribution entropy for the i-th recommendation level interval. i Let B be the user behavior feature value for the i-th recommendation level interval. i-1 minK represents the left boundary of the (i-1)th recommendation level interval. i This represents the kurtosis of the i-th recommendation level interval. It should be noted that when calculating the behavioral data distribution entropy corresponding to the first recommendation level interval, B in the formula should be removed. i-1 min Replace with B i+1 max .

[0085] It should be noted that the behavioral data distribution entropy mentioned in this application is a parameter used to measure the distribution of user behavioral data within the recommendation level range. The smaller the behavioral data distribution entropy, the more concentrated the distribution of user behavioral data within the recommendation level range is near the user behavioral feature value of the recommendation level range.

[0086] Furthermore, it should be noted that in a conventional finite interval, a normal distribution will have a small number of data points distributed into adjacent intervals at the interval boundaries. Therefore, when the collected user behavior data is near the interval boundaries, the identification effect of the recommendation level for that behavior data is poor, resulting in incorrect online information and advertising recommendations. In this application, by modifying the boundaries of the recommendation level intervals corresponding to each recommendation level, the probability distribution of the normal distribution within each recommendation level interval at the interval boundaries is greatly improved. This enhances the segmentation or identification effect of user behavior data at the boundaries of recommendation level intervals, thereby improving the accuracy of online information and advertising recommendations.

[0087] In step 103, a dataset of user behavior of target users towards a type of online information and advertisements is collected within a preset time period.

[0088] In practice, the target user's behavior data on a type of online information and advertisement can be collected from the background of the network server within a preset time period, thereby obtaining the target user's user behavior dataset.

[0089] It should be noted that the user behavior data described in this application may include the number of times a user views a type of online information and advertisement, the browsing time, the number of times they share it, and the number of times they like it. In other embodiments, the user behavior data may also include other behaviors of the target user, which is not limited here.

[0090] In step 104, the membership degree of the user behavior dataset is estimated based on the boundary data estimation factor of each recommendation level interval to obtain the behavior membership matrix.

[0091] In some embodiments, the membership estimation of the user behavior dataset based on the boundary data estimation factor for each recommendation level interval, to obtain the behavior membership matrix, can be achieved through the following steps:

[0092] Obtain user behavior feature values ​​for each recommendation level range;

[0093] Obtain the user behavior dataset;

[0094] Obtain the boundary data estimation factors for each recommendation level interval;

[0095] Obtain the distribution steepness for each recommendation level interval;

[0096] Select a user behavior data point from the user behavior dataset, and determine the user behavior membership degree between the user behavior data point and each recommendation level interval based on the user behavior feature value of each recommendation level interval, the distribution steepness of each recommendation level interval, and the boundary data estimation factor.

[0097] Continue to determine the remaining user behavior data in the user behavior dataset and the user behavior membership degree between each recommendation level interval;

[0098] Determine the behavior membership matrix based on the membership degrees of all user behaviors.

[0099] The membership degree of the user behavior is determined according to the following formula:

[0100]

[0101] Among them, y ij Let l be the membership degree of user behavior between the j-th user behavior data and the ith recommendation level interval. i K is the boundary data estimation factor for the i-th recommendation level interval. i Let E be the steepness of the distribution of the i-th recommendation level interval. i Let x be the user behavior feature value for the i-th recommendation level interval, H be the preset correction accuracy, and x be the value for the i-th recommendation level interval. j Let j be the j-th user behavior data in the user behavior dataset.

[0102] In practice, the membership degree of user behavior between the first user behavior data and one recommendation level interval can be used as the data in the first row and first column of the behavior membership matrix, and the membership degree of user behavior between the first user behavior data and two recommendation level intervals can be used as the data in the first row and second column of the behavior membership matrix, and so on, to obtain the behavior membership matrix.

[0103] It should be noted that the user behavior membership matrix described in this application is a matrix that includes the user behavior data of the target user and the user behavior membership degree between each recommendation level interval. The user behavior membership degree is a parameter used to measure the probability that a certain type of user behavior data of the target user belongs to a recommendation level interval. The larger the user behavior membership degree, the more likely the recommendation level of the online information and advertisement is to be the recommendation level corresponding to that recommendation level interval for that type of user behavior data of the target user.

[0104] Additionally, it should be noted that by calculating the degree of user behavior affiliation between each type of user behavior data and the corresponding recommendation level interval at each recommendation level, it is possible to determine the probability that each type of user behavior data of the target user belongs to the recommendation level interval at each recommendation level.

[0105] In some embodiments, reference Figure 2 This figure is a distribution diagram of user behavior membership according to some embodiments of this application, and is explained in detail below:

[0106] The vertical axis of the graph represents the membership degree of user behavior, and the horizontal axis represents the number of times the target user viewed news and advertisements within the range of [0,1] after standardization. The colors are used to distinguish different recommendation levels. As can be seen from the graph, data points of two colors will appear at the boundary between the two recommendation levels. If the recommendation level is directly divided based on the number of clicks of the target user through the data range, incorrect classification will occur.

[0107] In step 105, the recommendation level of the network information and advertisement for the target user is determined by the behavior membership matrix, and the network information and advertisement are recommended to the target user based on the recommendation level.

[0108] In some embodiments, determining the recommendation level of such online information and advertisements for target users through the behavioral membership matrix can be achieved through the following steps:

[0109] The recommendation incentive value corresponding to each user behavior data is determined based on the user behavior membership matrix.

[0110] Determine the recommended incentive vector based on all recommended incentive values;

[0111] The recommendation level of this type of online information and advertisement for the target user is determined based on the recommendation incentive vector and the user behavior membership matrix.

[0112] In practice, the recommendation incentive value corresponding to the first type of user behavior data can be used as the first data in the recommendation incentive vector, and so on, to obtain the recommendation incentive vector.

[0113] In some embodiments, determining the recommendation incentive value corresponding to each user behavior data based on the user behavior membership matrix, and then obtaining the recommendation incentive vector, can be achieved through the following steps:

[0114] The user behavior membership matrix is ​​standardized to obtain a standard user behavior matrix;

[0115] The standard user behavior matrix is ​​transformed by deviation to obtain the behavior deviation matrix;

[0116] The recommendation incentive value corresponding to each user behavior data is determined based on the behavior deviation matrix.

[0117] The recommendation incentive vector is determined based on all the recommended incentive values.

[0118] The recommended incentive value can be determined according to the following formula:

[0119]

[0120] Among them, u j w is the recommendation incentive value corresponding to the j-th type of user behavior data. ij For the element in the i-th row and j-th column of the behavior deviation matrix, w it Let L be the element in the i-th row and t-th column of the behavior deviation matrix, L be the length of the user behavior dataset, and M be the number of preset recommendation level intervals.

[0121] It should be noted that the recommendation incentive vector in this application is a vector composed of multiple recommendation incentive values, which are parameters used to measure the weight of user behavior membership in the corresponding recommendation level range.

[0122] In practice, each user behavior membership degree in the user behavior membership matrix can be subtracted from the smallest user behavior membership degree in the column containing that user behavior membership degree. The difference is then divided by the difference between the largest and smallest user behavior membership degrees in the column containing that user behavior membership degree. The quotient is used as the standard user behavior data corresponding to that user behavior membership degree, thus obtaining the standard user behavior matrix.

[0123] In practice, each standard user behavior data in the standard user behavior matrix can be divided by the sum of all standard user behavior data in the column containing that standard user behavior data. The resulting value is used as the behavior deviation degree corresponding to that standard user behavior data, thus obtaining the behavior deviation degree matrix.

[0124] It should be noted that the deviation matrix in this application is a matrix used to measure the degree of deviation between each user behavior membership degree in the user behavior membership matrix and the true value.

[0125] In some embodiments, determining the recommendation level of such online information and advertisements for a target user based on the recommendation incentive vector and the user behavior membership matrix can be achieved through the following steps:

[0126] A comprehensive evaluation vector is determined based on the recommendation incentive vector and the user behavior membership matrix;

[0127] The recommendation level of this type of online information and advertisement for the target user is determined based on the comprehensive evaluation vector.

[0128] In practice, the comprehensive evaluation vector can be obtained by multiplying the recommendation incentive vector by the user behavior membership matrix.

[0129] In practice, the recommendation level corresponding to the largest comprehensive evaluation value in the comprehensive evaluation vector can be used as the recommendation level of the online information and advertisement for the target user. For example, if the third comprehensive evaluation value in the comprehensive evaluation vector is the largest, then the recommendation level of the online information and advertisement for the target user is the third level.

[0130] It should be noted that by determining the degree of user behavior membership of each user behavior data of the target user within the recommendation level range corresponding to each recommendation level, and by standardizing and transforming all user behavior memberships, the recommendation incentive value of each user behavior membership can be obtained. Thus, the recommendation level of the target user for this type of online information and advertisement can be accurately predicted through the target user's multiple behavioral data, thereby enabling accurate online information and advertisement recommendations.

[0131] In some embodiments, reference Figure 3 The diagram is an exemplary flowchart illustrating the recommendation of online information and advertisements to target users based on the recommendation level, according to some embodiments of this application. The details are as follows:

[0132] In step 1051, an intermediate level is determined based on all preset recommended levels;

[0133] In step 1052, when the recommendation level of the online information and advertisement for the target user is higher than the intermediate level, the recommendation volume of the online information and advertisement is increased;

[0134] In step 1053, when the recommendation level of such online information and advertisement for the target user is lower than the intermediate level, the recommendation volume of such online information and advertisement is reduced.

[0135] In practice, the average of all recommended levels can be used as the intermediate level.

[0136] Additionally, in some embodiments, references Figure 4The figure is a schematic diagram of exemplary hardware and / or software of a big data-based enterprise information advertising push system 400 according to some embodiments of this application. The big data-based enterprise information advertising push system 400 may include: a setting module 401, a determining module 402, a collection module 403, an estimation module 404, and an execution module 405, wherein:

[0137] The setting module 401 in this application is mainly used to set multiple recommendation levels for online information and advertisements, and to divide the recommendation level range for each recommendation level;

[0138] The determination module 402 in this application is mainly used to determine the user behavior feature value of each recommendation level interval, and to determine the boundary data estimation factor of the corresponding recommendation level interval based on the user behavior feature value of each recommendation level interval.

[0139] The data collection module 403 in this application is mainly used to collect a dataset of user behavior of target users towards a type of online information and advertisement within a preset time period.

[0140] Estimation module 404, in this application, is mainly used to estimate the membership degree of the user behavior dataset based on the boundary data estimation factor of each recommendation level interval, so as to obtain the behavior membership matrix;

[0141] The execution module 405 in this application is mainly used to determine the recommendation level of the network information and advertisement for the target user through the behavior membership matrix, and to recommend network information and advertisement to the target user based on the recommendation level.

[0142] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device applying a big data-based enterprise information advertising push method according to some embodiments of this application. The big data-based enterprise information advertising push method in the above embodiments can... Figure 5 The computer device shown is used to implement this, and the computer device 500 includes at least one processor 501, a communication bus 502, a memory 503, and at least one communication interface 504.

[0143] The processor 501 may be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more devices used to control the execution of the big data-based enterprise information advertising push method in this application.

[0144] The communication bus 502 may include a path for transmitting information between the aforementioned components.

[0145] Memory 503 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 503 may exist independently and be connected to processor 501 via communication bus 502. Memory 503 may also be integrated with processor 501.

[0146] The memory 503 stores program code for executing the solution of this application, and its execution is controlled by the processor 501. The processor 501 executes the program code stored in the memory 503. The program code may include one or more software modules. In the above embodiment, the enterprise information advertising push method based on big data can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.

[0147] Communication interface 504 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.

[0148] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0149] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0150] In addition, this application also discloses a computer-readable storage medium that stores a computer program, which, when executed by a processor, implements the above-mentioned big data-based enterprise information advertising push method.

[0151] The computer-readable or machine-readable medium of this application can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device to implement a big data-based enterprise information advertising push method. The computer-readable medium can be a machine-readable signal medium or a machine-readable storage medium. The computer-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory or flash memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0152] In summary, this application provides a method and system for enterprise information and advertising push based on big data. By setting multiple recommendation levels for online information and advertisements and dividing each recommendation level into intervals, it addresses the common problem that a normal distribution within a finite interval will have a few data points distributed into adjacent intervals at the interval boundaries. Therefore, when the collected user behavior data is near the interval boundaries, the recognition of the recommendation level for that behavior data is poor, leading to incorrect recommendations of online information and advertisements. This application addresses this by determining a boundary data estimation factor for each recommendation level interval, which is half the length of the corrected recommendation level interval. Since the two corrected recommendation level intervals have overlapping areas, data points within these overlapping areas are considered. User behavior data is data that may be misclassified. By collecting user behavior datasets of target users and estimating the membership degree of the user behavior dataset using boundary data estimation factors, we can obtain the membership degree of each user behavior data and each recommendation level interval, resulting in a behavior membership matrix. In particular, when estimating the membership degree of data at the boundary of two recommendation level intervals, we consider the membership degree of the data with the theoretical normal distribution within both recommendation level intervals, thereby determining which interval the data at the boundary belongs to. This enhances the recognition effect of user behavior data at the boundary of recommendation level intervals, and then determines the recommendation level through the behavior membership matrix, thus realizing the division of user behavior data at the boundary of two recommendation level intervals.

[0153] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0154] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of the invention. Therefore, if these modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A method for pushing enterprise information advertisements based on big data, characterized in that, Includes the following steps: Set multiple recommendation levels for online information and advertisements, and divide each recommendation level into recommendation level ranges; Determine the user behavior feature value for each recommendation level interval, and determine the boundary data estimation factor for the corresponding recommendation level interval based on the user behavior feature value for each recommendation level interval; Collect a dataset of user behavior of target users towards a type of online information and advertisement within a preset time period; The membership degree of the user behavior dataset is estimated based on the boundary data estimation factor for each recommendation level interval, resulting in a behavior membership matrix; The recommendation level of the online information and advertisements for the target user is determined by the behavioral membership matrix, and the online information and advertisements are recommended to the target user based on the recommendation level. Determining the distribution steepness based on user behavior characteristics within a specific recommendation level range is achieved through the following steps: Determine the right boundary of this recommendation level range; Obtain user behavior feature values ​​for this recommendation level range; The steepness of the distribution is determined according to the following formula: in, For the first The steepness of the distribution of each recommendation level interval, For the first The right boundary of each recommendation level range, For the first The right boundary of each recommendation level range, For the first User behavior feature values ​​for each recommendation level range This represents the logarithm to the base 10 of 0.

5. Indicates base 10 The logarithm of ; The boundary data estimation factor for each recommendation level interval is determined based on the steepness of the distribution using the following steps: The behavioral data distribution entropy for the recommendation level interval is determined based on the user behavior feature values ​​and the distribution steepness of that interval. The boundary data estimation factor for the recommendation level interval is determined based on the distribution entropy of the behavioral data, wherein the boundary data estimation factor is determined according to the following formula: in, Indicates the generation of a... For the expectation, A normally distributed random number with variance . For the preset correction accuracy, It is the first Estimated factors for boundary data of each recommendation level interval. It is the first The behavioral data distribution entropy for each recommendation level interval It is the first The steepness of the distribution of each recommendation level interval; The behavioral data distribution entropy for a given recommendation level interval is determined using the following steps, based on the user behavior feature values ​​and the distribution steepness: Obtain the user behavior feature values ​​and distribution steepness for this recommendation level range; Determine the left boundary of this recommendation level range; The behavioral data distribution entropy of the recommendation level interval is determined based on the left boundary of the interval, the user behavior feature value of the interval, and the distribution steepness, wherein the behavioral data distribution entropy is determined according to the following: in, It is the first The behavioral data distribution entropy for each recommendation level interval For the first User behavior feature values ​​for each recommendation level range For the first The left boundary of each recommendation level range, It is the first The steepness of the distribution of each recommendation level interval should be noted. It should be observed that when calculating the distribution entropy of the behavioral data corresponding to the first recommendation level interval, the formula should include... Replace with ; The membership degree of the user behavior dataset is estimated based on the boundary data estimation factors for each recommendation level interval, and the behavior membership matrix is ​​obtained by the following steps: Obtain user behavior feature values ​​for each recommendation level range; Obtain the user behavior dataset; Obtain the boundary data estimation factors for each recommendation level interval; Obtain the distribution steepness for each recommendation level interval; Select a user behavior data point from the user behavior dataset, and determine the user behavior membership degree between the user behavior data point and each recommendation level interval based on the user behavior feature value of each recommendation level interval, the distribution steepness of each recommendation level interval, and the boundary data estimation factor. Continue to determine the remaining user behavior data in the user behavior dataset and the user behavior membership degree between each recommendation level interval; Determine the behavior membership matrix based on the membership degrees of all user behaviors; The membership degree of the user behavior is determined according to the following formula: in, For the first User behavior data and the first User behavior membership degree between recommendation level ranges For the first Estimated factors for boundary data of each recommendation level interval. For the first The steepness of the distribution of each recommendation level interval, For the first User behavior feature values ​​for each recommendation level range For the preset correction accuracy, For the first in the user behavior dataset User behavior data.

2. The method as described in claim 1, characterized in that, The boundary data estimation factors for each recommendation level interval are determined based on the user behavior feature values ​​of that interval. Specifically, this includes: Select a recommendation level range and determine the distribution steepness based on the user behavior feature values ​​of that recommendation level range; The boundary data estimation factor for each recommendation level interval is determined based on the steepness of the distribution. Continue to determine the boundary data estimation factors for the remaining recommendation level intervals.

3. The method as described in claim 2, characterized in that, The boundary data estimation factors for determining the recommendation level interval based on the distribution steepness specifically include: The behavioral data distribution entropy for the recommendation level interval is determined based on the user behavior feature values ​​and the distribution steepness of that interval. The boundary data estimation factor for each recommendation level interval is determined based on the behavioral data distribution entropy.

4. The method as described in claim 1, characterized in that, The membership degree of the user behavior dataset is estimated based on the boundary data estimation factor for each recommendation level interval, resulting in a behavior membership matrix, which specifically includes: Obtain user behavior feature values ​​for each recommendation level range; Obtain the user behavior dataset; Obtain the boundary data estimation factors for each recommendation level interval; Obtain the distribution steepness for each recommendation level interval; Select a user behavior data point from the user behavior dataset, and determine the user behavior membership degree between the user behavior data point and each recommendation level interval based on the user behavior feature value of each recommendation level interval, the distribution steepness of each recommendation level interval, and the boundary data estimation factor. Continue to determine the remaining user behavior data in the user behavior dataset and the user behavior membership degree between each recommendation level interval; Determine the behavior membership matrix based on the membership degrees of all user behaviors.

5. The method as described in claim 1, characterized in that, Determining the recommendation level of this type of online information and advertising for the target user through the behavioral membership matrix specifically includes: The recommendation incentive value corresponding to each user behavior data is determined based on the user behavior membership matrix. Determine the recommended incentive vector based on all recommended incentive values; The recommendation level of this type of online information and advertisement for the target user is determined based on the recommendation incentive vector and the user behavior membership matrix.

6. The method as described in claim 5, characterized in that, Determining the recommendation incentive value corresponding to each user behavior data based on the user behavior membership matrix specifically includes: The user behavior membership matrix is ​​standardized to obtain a standard user behavior matrix; The standard user behavior matrix is ​​transformed by deviation to obtain the behavior deviation matrix; The recommended incentive value corresponding to each user behavior data is determined based on the behavior deviation matrix.

7. The method as described in claim 1, characterized in that, The specific methods for recommending online information and advertisements to target users based on the aforementioned recommendation levels include: The intermediate level is determined based on all preset recommended levels; When the recommendation level of such online information and advertisement for a target user is higher than the intermediate level, the recommendation volume of such online information and advertisement is increased; When the recommendation level of such online information and advertisement for a target user is lower than the intermediate level, the recommendation volume of such online information and advertisement shall be reduced.

8. A big data-based enterprise information advertising push system, characterized in that, Including: The settings module is used to set multiple recommendation levels for online information and advertisements, and to divide each recommendation level into recommendation level ranges; The determination module is used to determine the user behavior feature value for each recommendation level interval, and to determine the boundary data estimation factor for the corresponding recommendation level interval based on the user behavior feature value for each recommendation level interval. The data collection module is used to collect a dataset of user behavior of target users towards a type of online information and advertisement within a preset time period. The estimation module is used to estimate the membership degree of the user behavior dataset based on the boundary data estimation factors of each recommendation level interval, and obtain the behavior membership matrix. The execution module is used to determine the recommendation level of the network information and advertisement for the target user through the behavior membership matrix, and to recommend network information and advertisement to the target user based on the recommendation level; Determining the distribution steepness based on user behavior characteristics within a specific recommendation level range is achieved through the following steps: Determine the right boundary of this recommendation level range; Obtain user behavior feature values ​​for this recommendation level range; The steepness of the distribution is determined according to the following formula: in, For the first The steepness of the distribution of each recommendation level interval, For the first The right boundary of each recommendation level range, For the first The right boundary of each recommendation level range, For the first User behavior feature values ​​for each recommendation level range This represents the logarithm to the base 10 of 0.

5. Indicates base 10 The logarithm of ; The boundary data estimation factor for each recommendation level interval is determined based on the steepness of the distribution using the following steps: The behavioral data distribution entropy for the recommendation level interval is determined based on the user behavior feature values ​​and the distribution steepness of that interval. The boundary data estimation factor for the recommendation level interval is determined based on the distribution entropy of the behavioral data, wherein the boundary data estimation factor is determined according to the following formula: in, Indicates the generation of a... For the expectation, A normally distributed random number with variance . For the preset correction accuracy, It is the first Estimated factors for boundary data of each recommendation level interval. It is the first The behavioral data distribution entropy for each recommendation level interval It is the first The steepness of the distribution of each recommendation level interval; The behavioral data distribution entropy for a given recommendation level interval is determined using the following steps, based on the user behavior feature values ​​and the distribution steepness: Obtain the user behavior feature values ​​and distribution steepness for this recommendation level range; Determine the left boundary of this recommendation level range; The behavioral data distribution entropy of the recommendation level interval is determined based on the left boundary of the interval, the user behavior feature value of the interval, and the distribution steepness, wherein the behavioral data distribution entropy is determined according to the following: in, It is the first The behavioral data distribution entropy for each recommendation level interval For the first User behavior feature values ​​for each recommendation level range For the first The left boundary of each recommendation level range, It is the first The steepness of the distribution of each recommendation level interval should be noted. It should be observed that when calculating the distribution entropy of the behavioral data corresponding to the first recommendation level interval, the formula should include... Replace with ; The membership degree of the user behavior dataset is estimated based on the boundary data estimation factors for each recommendation level interval, and the behavior membership matrix is ​​obtained by the following steps: Obtain user behavior feature values ​​for each recommendation level range; Obtain the user behavior dataset; Obtain the boundary data estimation factors for each recommendation level interval; Obtain the distribution steepness for each recommendation level interval; Select a user behavior data point from the user behavior dataset, and determine the user behavior membership degree between the user behavior data point and each recommendation level interval based on the user behavior feature value of each recommendation level interval, the distribution steepness of each recommendation level interval, and the boundary data estimation factor. Continue to determine the remaining user behavior data in the user behavior dataset and the user behavior membership degree between each recommendation level interval; Determine the behavior membership matrix based on the membership degrees of all user behaviors; The membership degree of the user behavior is determined according to the following formula: in, For the first User behavior data and the first User behavior membership degree between recommendation level ranges For the first Estimated factors for boundary data of each recommendation level interval. For the first The steepness of the distribution of each recommendation level interval, For the first User behavior feature values ​​for each recommendation level range For the preset correction accuracy, For the first in the user behavior dataset User behavior data.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the big data-based enterprise information advertising push method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the enterprise information advertising push method based on big data as described in any one of claims 1 to 7.

Images