Behavior recognition method, device and equipment based on user behavior prediction model training

Abstract

The embodiments of the present disclosure disclose a behavior recognition method, device, and equipment based on user behavior prediction model training. A specific implementation of the method includes: obtaining a sample user browsing behavior information set, wherein the sample user browsing behavior information includes: a behavior response label, a behavior feedback label, and a user feature information set; training a behavior response label classification model and a behavior feedback label classification model based on the sample user browsing behavior information set; determining a user behavior causal feature set based on the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model; training an initial user behavior prediction model based on the user behavior causal feature set and the sample user browsing behavior information set to obtain a user behavior prediction model. This implementation avoids overfitting of the model and improves the accuracy of the recognition rate of the trained model.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of model training, and specifically to a behavior recognition method, apparatus, and device based on user behavior prediction model training. Background Art

[0002] At present, in order to determine the intention corresponding to the user's browsing behavior, the commonly used approach is to establish a user behavior feedback model, introduce stronger data processing methods, and introduce semi-supervised learning methods to use unlabeled data to enhance the performance of the model, thereby achieving causal inference of user browsing behavior.

[0003] However, when the above method is used to identify the intention corresponding to the user's web browsing behavior, the following technical problems often occur: the introduction of a stronger data processing method causes the model to be more complex, which easily leads to overfitting of the model, and the use of a semi-supervised learning method easily leads to label noise and data imbalance, resulting in a low recognition rate of the trained model.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art. Summary of the invention

[0005] The content of this disclosure is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this disclosure is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0006] Some embodiments of the present disclosure propose a behavior recognition method, device, electronic device and computer-readable medium based on user behavior prediction model training to solve one or more of the technical problems mentioned in the above background technology section.

[0007] In a first aspect, some embodiments of the present disclosure provide a behavior recognition method based on user behavior prediction model training, the method comprising: obtaining a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set comprises: a behavior response label, a behavior feedback label and a user feature information set; training a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set, wherein both the behavior response label classification model and the behavior feedback label classification model correspond to a user feature sequence; determining a user behavior causal feature set according to the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model; training an initial user behavior prediction model according to the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model; obtaining the browsing behavior information of each target user for the target web page to obtain a browsing behavior information set; inputting the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set, wherein the browsing behavior information in the browsing behavior information set corresponds to the user behavior prediction result in the user behavior prediction result set.

[0008] In a second aspect, some embodiments of the present disclosure provide a behavior recognition device based on user behavior prediction model training, the device comprising: a first acquisition unit, configured to acquire a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set comprises: a behavior response label, a behavior feedback label and a user feature information set; a first training unit, configured to train a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set, wherein the behavior response label classification model and the behavior feedback label classification model both correspond to a user feature sequence; a determination unit, configured to determine the user feature sequence according to the user feature sequence corresponding to the behavior response label classification model; The feature sequence is combined with the user feature sequence corresponding to the above-mentioned behavior feedback label classification model to determine the user behavior causal feature set; the second training unit is configured to train the initial user behavior prediction model according to the above-mentioned user behavior causal feature set and the above-mentioned sample user browsing behavior information set to obtain the trained user behavior prediction model; the second acquisition unit is configured to obtain the browsing behavior information of each target user for the target web page to obtain the browsing behavior information set; the input unit is configured to input the above-mentioned browsing behavior information set into the above-mentioned user behavior prediction model to obtain the user behavior prediction result set, wherein the browsing behavior information in the above-mentioned browsing behavior information set corresponds to the user behavior prediction result in the above-mentioned user behavior prediction result set.

[0009] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner of the above-mentioned first aspect.

[0010] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner of the above-mentioned first aspect is implemented.

[0011] The above-mentioned embodiments of the present disclosure have the following beneficial effects: through the behavior recognition method based on user behavior prediction model training of some embodiments of the present disclosure, the overfitting of the model is avoided, and the accuracy of the recognition rate of the trained model is improved. Specifically, the reason for the low recognition rate of the trained model is that the introduction of a stronger data processing method causes the model complexity to be higher, which easily makes the model overfitting, and the use of a semi-supervised learning method easily leads to label noise and data imbalance. Based on this, the behavior recognition method based on user behavior prediction model training of some embodiments of the present disclosure, first, obtains a sample user browsing behavior information set. Among them, the sample user browsing behavior information in the above sample user browsing behavior information set includes: behavior response label, behavior feedback label and user feature information set. Thus, the user behavior information set that can be obtained can be used as a real data set for training the user behavior prediction model. Secondly, according to the above sample user browsing behavior information set, the behavior response label classification model and the behavior feedback label classification model are trained. Among them, the above behavior response label classification model and the above behavior feedback label classification model both correspond to user feature sequences. Thus, the classification model of the behavior response label and the classification model of the behavior feedback label can be trained respectively using the behavior response label and the behavior feedback label, and the user feature sequence can represent each feature adopted by the classification model. Then, the user behavior causal feature set is determined according to the user feature sequence corresponding to the above-mentioned behavior response label classification model and the user feature sequence corresponding to the above-mentioned behavior feedback label classification model. Thus, the user behavior causal feature set can represent the important user features selected according to the user feature sequence corresponding to the behavior response label classification model and the behavior feedback label classification model, so that the scope of the subsequent representation learning reinforcement feature representation can be pre-selected. Afterwards, the initial user behavior prediction model is trained according to the above-mentioned user behavior causal feature set and the above-mentioned sample user browsing behavior information set to obtain a trained user behavior prediction model. Thus, the user behavior causal feature set can enable the subsequent representation learning to pay more attention to the feature expression after screening in the modeling process, reducing the risk of overfitting caused by the complexity of network calculation. Then, the browsing behavior information of each target user for the target web page is obtained to obtain the browsing behavior information set. Finally, the above browsing behavior information set is input into the above user behavior prediction model to obtain a user behavior prediction result set, wherein the browsing behavior information in the above browsing behavior information set corresponds to the user behavior prediction result in the above user behavior prediction result set. Also, because the user behavior information set contains behavior response labels, there is no need to use semi-supervised or weakly supervised methods, which reduces label noise and data imbalance caused by semi-supervised or weakly supervised methods. It avoids overfitting of the model and improves the accuracy of the recognition rate of the trained model. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar figure labels represent the same or similar elements. It should be understood that the drawings are schematic and that the components and elements are not necessarily drawn to scale.

[0013] Figure 1 is a flow chart of some embodiments of a behavior recognition method based on user behavior prediction model training according to the present disclosure;

[0014] Figure 2 is a flow chart of some embodiments of a behavior recognition device based on user behavior prediction model training according to the present disclosure;

[0015] Figure 3 It is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0016] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0017] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other.

[0018] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0019] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0020] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0021] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0022] Figure 11 is a flowchart of some embodiments of the behavior recognition method based on user behavior prediction model training according to the present disclosure. 100 of some embodiments of the behavior recognition method based on user behavior prediction model training according to the present disclosure is shown. The behavior recognition method based on user behavior prediction model training includes the following steps:

[0023] Step 101, obtaining a sample user browsing behavior information set.

[0024] In some embodiments, the execution subject (e.g., a computing device) of the behavior recognition method based on user behavior prediction model training can obtain a sample user browsing behavior information set through a wired connection or a wireless connection method. Among them, the sample user browsing behavior information in the above-mentioned sample user browsing behavior information set includes: a behavior response label, a behavior feedback label, and a user feature information set. The sample user browsing behavior information can be information related to user attributes and user browsing behavior. The above-mentioned behavior response label can characterize the type of response made by the user to the feedback items of browsing web page behavior. The behavior feedback label can characterize the feedback items of browsing web page behavior, and can include behavior feedback variables and behavior response variables. The behavior feedback variable can be whether it is a push information for the user, and the behavior response variable can be whether the user will browse the push information. The user feature information set can be various attributes and / or behavior feature information of the user (e.g., age, gender, preferred web page type, activity, number of web page views).

[0025] Step 102: training a behavior response label classification model and a behavior feedback label classification model based on the above sample user browsing behavior information set.

[0026] In some embodiments, the execution subject may train a behavior response label classification model and a behavior feedback label classification model based on the sample user browsing behavior information set. The behavior response label classification model and the behavior feedback label classification model both correspond to user feature sequences. For example, the behavior response label classification model may be LightGBM. The behavior feedback label classification model may be LightGBM.

[0027] In practice, the above execution entity can train the behavior response label classification model and the behavior feedback label classification model through the following steps:

[0028] The first step is to train the behavior response label classification model according to the training set corresponding to the above sample user browsing behavior information set and the various behavior response labels included in the above training set. In practice, the above execution subject can use the behavior response labels included in the browsing behavior information of each sample user in the training set as sample labels, and use the behavior feedback labels and user feature information sets included in the browsing behavior information of each sample user in the training set as input features, and obtain the behavior response label classification model according to the above training set training. Then, the various user features used for judgment in the classification process of the behavior response label classification model can be combined in sequence to obtain the user feature sequence corresponding to the above behavior response label classification model.

[0029] The second step is to train the behavior feedback label classification model based on the training set corresponding to the above sample user browsing behavior information set and the various behavior feedback labels included in the above training set. In practice, the above execution entity can use the behavior feedback labels included in the browsing behavior information of each sample user in the training set as sample labels, and the behavior response labels and user feature information sets included in the browsing behavior information of each sample user in the training set as input features, and obtain the behavior feedback label classification model based on the above training set training. Then, the various user features used by the behavior feedback label classification model for judgment during the classification process can be combined in sequence to obtain a user feature sequence corresponding to the above behavior feedback label classification model.

[0030] Step 103: determining a user behavior causal feature set according to the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model.

[0031] In some embodiments, the execution entity may determine a user behavior causal feature set based on a user feature sequence corresponding to the behavior response label classification model and a user feature sequence corresponding to the behavior feedback label classification model.

[0032] Furthermore, when the behavior recognition method based on user behavior prediction model training of the present application is used for user behavior recognition, the following problems often occur: the causal characteristics between the user behavior response label and the behavior feedback label are not considered. For these problems, the conventional solution is to manually annotate the causal characteristics between the behavior response label and the behavior feedback label in each sample.

[0033] However, the above solutions usually have the following technical problem three: manual labeling efficiency is low and there are deviations in the labeling, which reduces the robustness of the model.

[0034] In response to the above technical problem three, it was decided to adopt the following solution.

[0035] In practice, the above execution entity can determine the causal feature set of user behavior through the following steps:

[0036] In the first step, each user feature that meets a preset feature condition in the user feature sequence corresponding to the behavior response label classification model is determined as a first user feature group. The preset feature condition may be that the user feature is the first M user features in the user feature sequence.

[0037] In the second step, each user feature that meets a preset feature condition in the user feature sequence corresponding to the behavior feedback label classification model is determined as a second user feature group.

[0038] In the third step, the intersection of the first user feature group and the second user feature group is determined as the user behavior causal feature set.

[0039] The above-mentioned related content, as an inventive point of the present disclosure, solves the third technical problem "reduced robustness of the model.". The factors that reduce the robustness of the model are often as follows: the efficiency of manual labeling is low, and there are deviations in the labeling, which reduces the robustness of the model. If the above factors are solved, the effect of improving the robustness of the model can be achieved. In order to achieve this effect, first, each user feature that meets the preset feature conditions in the user feature sequence corresponding to the above-mentioned behavior response label classification model is determined as a first user feature group. Then, each user feature that meets the preset feature conditions in the user feature sequence corresponding to the above-mentioned behavior feedback label classification model is determined as a second user feature group. Finally, the intersection of the above-mentioned first user feature group and the above-mentioned second user feature group is determined as a user behavior causal feature set. Therefore, the user behavior causal characteristics can be judged by the feature intersection of the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model. Thereby, the recognition efficiency and accuracy of the causal characteristics are improved, thereby improving the robustness of the subsequent training model.

[0040] Step 104 , training the initial user behavior prediction model based on the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model.

[0041] In some embodiments, the execution subject may train the initial user behavior prediction model according to the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model. The initial user behavior prediction model may be an untrained user behavior prediction model. For example, the initial user behavior prediction model may be a DCN (Deep & Cross Network) or a transformer model.

[0042] Furthermore, when the behavior recognition method based on user behavior prediction model training of the present application is used to predict and recognize user behavior, the following problems often occur: the correlation between the causal characteristics of user behavior and user browsing behavior is not considered, resulting in a low recognition rate of the trained user behavior prediction model. For these problems, the conventional solution is to train the user behavior prediction model through semi-supervision or weak supervision.

[0043] However, the above solutions usually have the following technical problem 2: when training user behavior prediction models through semi-supervision or weak supervision, there are many cases of label noise and data imbalance, which leads to large recognition errors in the trained models.

[0044] In response to the above-mentioned technical problem 2, it was decided to adopt the following solution.

[0045] In practice, the above execution entity can train the initial user behavior prediction model through the following steps:

[0046] In the first step, the behavior feedback feature is added to the above user behavior causal feature set to obtain an updated user behavior causal feature set. The behavior feedback feature can be a feature corresponding to the behavior feedback label.

[0047] The second step is to generate a browsing behavior information training sample set based on the training set corresponding to the above sample user browsing behavior information set and the updated user behavior causal feature set. Each browsing behavior information training sample in the above browsing behavior information training sample set includes a behavior response label and each user feature information corresponding to the user behavior causal feature set, and the user feature information corresponding to the behavior feedback feature is the behavior feedback label. In practice, for each sample user browsing behavior information in the training set, the above execution entity can remove each user feature information except each user feature information corresponding to the updated user behavior causal feature set from the above sample user browsing behavior information to obtain a browsing behavior information training sample.

[0048] The third step is to select a target browsing behavior information training sample from the above browsing behavior information training sample set. The above execution subject can randomly select a browsing behavior information training sample from the above browsing behavior information training sample set as the target browsing behavior information training sample.

[0049] In the fourth step, the target browsing behavior information training sample is input into the initial user behavior prediction model including the embedding layer to obtain the word vectors of the target browsing behavior information training sample corresponding to the user behavior causal feature set. The embedding layer can be used to generate the word vector of the user feature information. One user behavior causal feature corresponds to one word vector.

[0050] The fifth step is to generate corresponding predicted behavior response labels based on the above word vectors and the above initial user behavior prediction model.

[0051] The fifth step may include the following sub-steps:

[0052] In the first sub-step, each of the above word vectors is input into the first feature interaction layer included in the initial user behavior prediction model to obtain the corresponding concatenated word vector. For example, the above first feature interaction layer can be an encoder of DCN or transformer.

[0053] The second sub-step is to input the concatenated word vector into the first prediction layer included in the initial response behavior prediction model to obtain the corresponding predicted behavior response label. For example, the prediction layer can be an output layer or a decoder.

[0054] Step 6: compare the predicted behavior response label with the behavior response label corresponding to the target browsing behavior information training sample to obtain a comparison result. In practice, the execution subject can determine whether the predicted behavior response label and the behavior response label are the same to generate a comparison result.

[0055] In step 7, in response to determining that the comparison result meets the target condition, the initial user behavior prediction model is determined as the trained user behavior prediction model. The target condition may be that the predicted behavior response label and the behavior response label are the same.

[0056] The above-mentioned related contents, as an inventive point of the present disclosure, solve the second technical problem "causing large recognition errors in the trained model." The factors that lead to large recognition errors in the trained model are often as follows: the user behavior prediction model is trained in a semi-supervised or weakly supervised manner, and there are many cases of label noise and data imbalance. If the above-mentioned factors are solved, the robustness of the model can be improved and the model recognition error can be reduced. In order to achieve this effect, first, the behavior feedback feature is added to the above-mentioned user behavior causal feature set to obtain an updated user behavior causal feature set. Secondly, a browsing behavior information training sample set is generated according to the training set corresponding to the above-mentioned sample user browsing behavior information set and the updated user behavior causal feature set. Among them, each browsing behavior information training sample in the above-mentioned browsing behavior information training sample set includes a behavior response label and each user feature information of the corresponding user behavior causal feature set, and the user feature information of the corresponding behavior feedback feature is a behavior feedback label. Then, the target browsing behavior information training sample is selected from the above-mentioned browsing behavior information training sample set. Afterwards, the target browsing behavior information training sample is input into the initial user behavior prediction model including the embedding layer, and each word vector of the target browsing behavior information training sample corresponding to the user behavior causal feature set is obtained. Then, based on the above-mentioned word vectors and the above-mentioned initial user behavior prediction model, the corresponding predicted behavior response label is generated. The predicted behavior response label is compared with the behavior response label corresponding to the target browsing behavior information training sample to obtain a comparison result. Finally, in response to determining that the above-mentioned comparison result meets the target condition, the initial user behavior prediction model is determined as the trained user behavior prediction model. As a result, the user behavior causal feature set can make subsequent representation learning pay more attention to the feature expression after screening in the modeling process, reducing the risk of overfitting caused by the complexity of network calculation. Also, because no enhanced or expanded samples are used, but the existing real data set is used, the error brought to the causal inference result and the complexity of modeling are reduced, and there is no need to collect real control experimental data without extending the collection time. Also, because the user behavior information set contains behavior response labels, it is not necessary to use semi-supervised or weakly supervised methods, which reduces the label noise and data imbalance caused by semi-supervised or weakly supervised methods. In this way, existing real data can be utilized, reducing the model complexity and the risk of overfitting, and improving the accuracy of model identification and causal inference.

[0057] In step 8, in response to determining that the comparison result does not meet the target condition, the network parameters of the initial user behavior prediction model are adjusted, and the target browsing behavior information training samples are reselected from the browsing behavior information training samples that have not been used, and the adjusted initial response behavior prediction model is trained again. For example, the execution subject can determine the loss value between the predicted response label and the behavior response label. Then, the gradient descent method can be used to adjust the network parameters of the initial user behavior prediction model according to the loss value.

[0058] The execution subject can adjust the network parameters of the initial user behavior prediction model through the following sub-steps:

[0059] The first sub-step is to generate a first loss value according to the predicted response label and the behavioral response label. In practice, the execution subject may use a preset loss function to determine the loss value between the predicted response label and the behavioral response label. For example, the loss function may refer to a hinge loss function or a cross entropy loss function.

[0060] The second sub-step is to generate a second loss value according to each word vector corresponding to the above target browsing behavior information training sample. First, determine the feature domain of each user behavior causal feature corresponding to the above target browsing behavior information training sample. The feature domain may refer to the category to which the user feature belongs. Then, for each feature domain, the Euclidean distance between each two word vectors corresponding to the above feature domain may be determined. Finally, the average value of each Euclidean distance may be determined as the second loss value.

[0061] The third sub-step is to generate a model loss value according to the first loss value and the second loss value. The sum of the first loss value and the second loss value can be determined as the model loss value.

[0062] The fourth sub-step is to adjust the network parameters of the initial user behavior prediction model according to the model loss value. The execution subject may use a gradient descent method to adjust the network parameters of the initial user behavior prediction model according to the model loss value.

[0063] Step 105: Obtain browsing behavior information of each target user for the target web page to obtain a browsing behavior information set.

[0064] In some embodiments, the execution entity may obtain the browsing behavior information of each target user for the target webpage to obtain a browsing behavior information set. The browsing behavior information may include: the number of page views, the number of clicks, and the behavior feedback information for the target webpage. The behavior feedback information may represent the recommendation information (e.g., advertisement) of the target webpage pushed to the target user.

[0065] Step 106: input the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set.

[0066] In some embodiments, the execution entity may input the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set. The browsing behavior information in the browsing behavior information set corresponds to the user behavior prediction result in the user behavior prediction result set. The user behavior prediction result may characterize the predicted browsing behavior of the target user for the target web page. For example, the user behavior prediction result may be "User 001, 1". 1 may represent that after the feedback information (advertisement) of the target web page is pushed to user 001, user 001 will click on the advertisement.

[0067] Optionally, the pending action feedback information of the target webpage is obtained.

[0068] In some embodiments, the execution subject may obtain the pending action feedback information of the target webpage. The pending feedback information may be the feedback information of the target webpage to be pushed to the user.

[0069] Optionally, for each target user, in response to determining that the user behavior prediction result corresponding to the target user satisfies a preset behavior feedback condition, the behavior feedback information to be sent to a user terminal corresponding to the target user.

[0070] In some embodiments, the execution subject may send the pending behavior feedback information to the user terminal corresponding to each target user in response to determining that the user behavior prediction result corresponding to the target user satisfies the preset behavior feedback condition. The preset behavior feedback condition may be that the user behavior prediction result indicates that the target user will push the pending behavior feedback information.

[0071] Further references Figure 2 As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a behavior recognition device based on user behavior prediction model training. These behavior recognition device embodiments based on user behavior prediction model training are Figure 1 Corresponding to the method embodiments shown, the behavior recognition device based on user behavior prediction model training can be specifically applied to various electronic devices.

[0072] like Figure 2As shown, in some embodiments, the behavior recognition device 200 based on user behavior prediction model training includes: a first acquisition unit 201, a first training unit 202, a determination unit 203, a second training unit 204, a second acquisition unit 205 and an input unit 206. The first acquisition unit 201 is configured to acquire a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set includes: a behavior response label, a behavior feedback label and a user feature information set; the first training unit 202 is configured to train a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set, wherein the behavior response label classification model and the behavior feedback label classification model both correspond to a user feature sequence; the determination unit 203 is configured to classify the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model feature sequence, determining the user behavior causal feature set; the second training unit 204 is configured to train the initial user behavior prediction model according to the above user behavior causal feature set and the above sample user browsing behavior information set, to obtain the trained user behavior prediction model; the second acquisition unit 205 is configured to obtain the browsing behavior information of each target user for the target web page, to obtain the browsing behavior information set; the input unit 206 is configured to input the above browsing behavior information set into the above user behavior prediction model, to obtain the user behavior prediction result set, wherein the browsing behavior information in the above browsing behavior information set corresponds to the user behavior prediction result in the above user behavior prediction result set.

[0073] It can be understood that the units recorded in the behavior recognition device 200 based on user behavior prediction model training are similar to those in the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the behavior recognition device 200 based on user behavior prediction model training and the units contained therein, and will not be repeated here.

[0074] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device (e.g., a computing device) 300 suitable for implementing some embodiments of the present disclosure. The electronic devices in some embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (e.g., vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0075] like Figure 3 As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0076] Typically, the following devices may be connected to the I / O interface 305: input devices 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 308 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 309. The communication devices 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0077] In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

[0078] It should be noted that the computer-readable medium recorded in some embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may 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 or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0079] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0080] The computer-readable medium may be included in the electronic device, or may exist independently without being installed in the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set includes: a behavior response label, a behavior feedback label and a user feature information set; trains a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set, wherein both the behavior response label classification model and the behavior feedback label classification model correspond to a user feature sequence; determines a user behavior causal feature set according to the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model; trains an initial user behavior prediction model according to the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model; obtains browsing behavior information of each target user for a target web page to obtain a browsing behavior information set; inputs the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set, wherein the browsing behavior information in the browsing behavior information set corresponds to the user behavior prediction result in the user behavior prediction result set.

[0081] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0082] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0083] The units described in some embodiments of the present disclosure may be implemented by software or by hardware. The described units may also be set in a processor, for example, may be described as: a processor comprising: a first acquisition unit, a first training unit, a determination unit, a second training unit, a second acquisition unit and an input unit. The names of these units do not constitute limitations on the units themselves in certain circumstances. For example, the second acquisition unit may also be described as "a unit for obtaining browsing behavior information of each target user for a target web page and obtaining a browsing behavior information set".

[0084] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0085] The above descriptions are only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the embodiments of the present disclosure (but not limited to) and the technical solutions formed.

Claims

1. A behavior recognition method based on user behavior prediction model training, comprising: Obtaining a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set includes: a behavior response tag, a behavior feedback tag, and a user characteristic information set, wherein the behavior response tag represents the type of response made by the user to the feedback item of the web browsing behavior, the behavior feedback tag represents the feedback item of the web browsing behavior, including a behavior feedback variable and a behavior response variable, the behavior feedback variable is whether it is the user's push information, the behavior response variable is whether the user will browse the push information, and the user characteristic information set is each attribute and / or behavior characteristic information of the user; According to the sample user browsing behavior information set, a behavior response label classification model and a behavior feedback label classification model are trained, wherein the behavior response label classification model and the behavior feedback label classification model both correspond to user feature sequences; Determining a user behavior causal feature set according to a user feature sequence corresponding to the behavior response label classification model and a user feature sequence corresponding to the behavior feedback label classification model; Training the initial user behavior prediction model according to the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model; Obtaining browsing behavior information of each target user on the target webpage to obtain a browsing behavior information set; Inputting the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set, wherein the browsing behavior information in the browsing behavior information set corresponds to the user behavior prediction result in the user behavior prediction result set; The step of determining the user behavior causal feature set according to the user feature sequence corresponding to the behavior response label classification model and the user feature sequence corresponding to the behavior feedback label classification model includes: Determine each user feature that meets a preset feature condition in the user feature sequence corresponding to the behavior response label classification model as a first user feature group, wherein the preset feature condition is that the user feature is the user feature that is arranged in the first M in the user feature sequence; Determine each user feature that meets a preset feature condition in the user feature sequence corresponding to the behavior feedback label classification model as a second user feature group; An intersection of the first user feature group and the second user feature group is determined as a user behavior causal feature set.

2. The method according to claim 1, wherein: The method further comprises: Obtaining pending action feedback information of the target webpage; For each target user, in response to determining that the user behavior prediction result corresponding to the target user meets the preset behavior feedback condition, the behavior feedback information to be sent to the user terminal corresponding to the target user.

3. The method according to claim 1, wherein: The step of training a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set includes: Training a behavior response label classification model according to a training set corresponding to the sample user browsing behavior information set and each behavior response label included in the training set; The behavior feedback label classification model is trained according to the training set corresponding to the sample user browsing behavior information set and each behavior feedback label included in the training set.

4. A behavior recognition device based on user behavior prediction model training, applied to the method according to claim 1, comprising: A first acquisition unit is configured to acquire a sample user browsing behavior information set, wherein the sample user browsing behavior information in the sample user browsing behavior information set includes: a behavior response label, a behavior feedback label and a user feature information set; A first training unit is configured to train a behavior response label classification model and a behavior feedback label classification model according to the sample user browsing behavior information set, wherein the behavior response label classification model and the behavior feedback label classification model both correspond to user feature sequences; a determination unit configured to determine a user behavior causal feature set according to a user feature sequence corresponding to the behavior response label classification model and a user feature sequence corresponding to the behavior feedback label classification model; A second training unit is configured to train the initial user behavior prediction model according to the user behavior causal feature set and the sample user browsing behavior information set to obtain a trained user behavior prediction model; A second acquisition unit is configured to acquire browsing behavior information of each target user on the target webpage to obtain a browsing behavior information set; The input unit is configured to input the browsing behavior information set into the user behavior prediction model to obtain a user behavior prediction result set, wherein the browsing behavior information in the browsing behavior information set corresponds to the user behavior prediction result in the user behavior prediction result set.

5. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 3.

6. A computer readable medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images