Information processing methods, information processing model training methods and devices

Abstract

This disclosure relates to an information processing method, an information processing model training method, and an apparatus. The information processing method includes: acquiring object attribute information of a target object and second historical operation information of the target object on previously pushed information; the previously pushed information includes push information in at least one push format; predicting the push format information of the target object based on the object attribute information, the second historical operation information, and the information processing model to obtain push format prediction information; determining the target push format of the information to be pushed based on the push format prediction information; and pushing the push information in the target push format to the target object. This disclosure can expand the push format and improve the conversion rate of push information.

Description

Technical Field

[0001] This disclosure relates to the field of machine learning technology, and in particular to an information processing method, an information processing model training method, and an apparatus. Background Technology

[0002] In information push systems, new push information is constantly uploaded to the system and awaits push notification. In related technologies, when pushing information to the recipient, it is generally based on the content of the push information. This may involve pushing push information containing content that the recipient is highly interested in to the recipient. This information push method is based on the content dimension of the push information and cannot be based on other dimensions of the push information, thus the push form is relatively simple. Summary of the Invention

[0003] This disclosure provides an information processing method, an information processing model training method, and an apparatus to at least solve the problems in related technologies where information cannot be pushed based on dimensions other than the content being pushed, and where the push format is too limited. The technical solution of this disclosure is as follows:

[0004] According to a first aspect of the present disclosure, an information processing method is provided, comprising:

[0005] Obtain object attribute information of the target object, and second historical operation information of the target object on the pushed information; the pushed information includes push information in at least one push form;

[0006] Based on the object attribute information of the target object, the second historical operation information, and the information processing model, the push format information of the target object is predicted to obtain push format prediction information; the information processing model is trained on a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push information, and training labels; the sample push information includes push information in multiple push formats, and the push format represents the display form of the sample push information when a display operation is triggered on the sample push information; the training labels are determined based on the information analysis results of the first historical operation information; the training labels represent the sample object's attention information to the multiple push formats;

[0007] The target push format of the information to be pushed is determined based on the push format prediction information.

[0008] Push the push information in the target push format to the target object.

[0009] In an exemplary embodiment, determining the target push format of the information to be pushed based on the push format prediction information includes:

[0010] Obtain preset object attribute information and preset operation information corresponding to various push formats;

[0011] Based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information, target correction information for the push pattern prediction information is determined;

[0012] Based on the target correction information, the push pattern prediction information is corrected to obtain the target prediction information;

[0013] The target delivery pattern is determined based on the target prediction information.

[0014] In an exemplary embodiment, determining the target correction information for the push pattern prediction information based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information includes:

[0015] The object attribute information of the target object is matched with the preset object attribute information corresponding to each of the multiple push forms to obtain the first matching similarity between the target object and each of the multiple preset object attribute information.

[0016] The second historical operation information of the target object is matched with the preset operation information corresponding to each of the multiple push forms to obtain the second matching similarity between the target object and each of the multiple preset operation information.

[0017] Based on the first matching similarity and the second matching similarity corresponding to the target object and each of the various push formats, the matching push format and the non-matching push format corresponding to the target object are determined.

[0018] Determine the first weight corresponding to the matching push form and the second weight corresponding to the non-matching push form;

[0019] The target correction information is determined based on the first weight and the second weight.

[0020] In one exemplary embodiment, the method further includes:

[0021] In response to a push notification display request triggered by the target object, the historical operation information of the target object within a preset historical time period is obtained; the preset historical time period is a time period with the current time as the end point and a preset duration.

[0022] The target object's historical operation information on the pushed information within the preset historical time period is determined as the second historical operation information.

[0023] According to a second aspect of the present disclosure, an information processing model training method is provided, comprising:

[0024] Obtain the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information;

[0025] The first historical operation information is analyzed, and training labels corresponding to the sample object are determined based on the analysis results; the training labels represent the sample object's attention information to the various push formats.

[0026] Based on the object attribute information of the sample object, the first historical operation information, and the training label, a preset machine learning model is trained to obtain an information processing model.

[0027] In an exemplary embodiment, before obtaining the first historical operation information of the sample object on the sample push information, the method further includes:

[0028] Obtain push form templates corresponding to each of the various push forms; each push form template includes at least one target field;

[0029] Obtain sample push material information;

[0030] The sample push material information is matched with the target fields in the push form templates corresponding to the various push forms, and the matching push form template is determined from the push form templates corresponding to the various push forms based on the matching results.

[0031] Based on the sample push material information, the matching push pattern template is populated with information to obtain the sample push information.

[0032] In one exemplary embodiment, the sample push information includes exposure display information and a shape identifier of the push form;

[0033] Before obtaining the first historical operation information of the sample object on the sample push information, the method further includes:

[0034] The sample push information is sent to the terminal corresponding to the sample object; when the sample push information is exposed on the terminal, the terminal displays the exposure display information and the shape identifier.

[0035] In one exemplary embodiment, the training labels corresponding to the sample objects include sub-labels corresponding to each of the various push formats;

[0036] The step of performing information analysis on the first historical operation information and determining the training label corresponding to the sample object based on the information analysis results includes:

[0037] The target push information corresponding to each of the various push formats is determined from the sample push information;

[0038] Based on the first historical operation information, determine the number of target push information being operated on in the target push information corresponding to each push form;

[0039] Based on the relationship between the number of pushed information to the target corresponding to each push form and the total number of pushed information in the sample, the sub-tag corresponding to each push form is determined;

[0040] The training labels are determined based on the sub-labels corresponding to each of the various push formats.

[0041] In an exemplary embodiment, the first historical operation information includes historical operation information corresponding to each of the multiple push formats;

[0042] The step of training a preset machine learning model based on the object attribute information of the sample object, the first historical operation information, and the training labels to obtain an information processing model includes:

[0043] Based on the object attribute information and the historical operation information corresponding to each of the various push formats, a joint input feature is constructed.

[0044] The joint input features are input into the preset machine learning model for information prediction to obtain the predicted label;

[0045] Loss information is determined based on the predicted labels and the training labels;

[0046] The information processing model is obtained by training the preset machine learning model based on the loss information.

[0047] According to a third aspect of the present disclosure, an information processing apparatus is provided, comprising:

[0048] The second information acquisition unit is configured to acquire object attribute information of the target object and second historical operation information of the target object on the pushed information; the pushed information includes push information in at least one type.

[0049] The second prediction unit is configured to predict the push notification pattern information of the target object based on the object attribute information of the target object, the second historical operation information, and the information processing model, to obtain push notification pattern prediction information. The information processing model is trained on a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push notification information, and training labels. The sample push notification information includes push notification information in multiple push notification patterns, and the push notification pattern represents the display format of the sample push notification information when a display operation is triggered on the sample push notification information. The training labels are determined based on the information analysis results of the first historical operation information. The training labels represent the sample object's attention information to the multiple push notification patterns.

[0050] The push form determination unit is configured to determine the target push form of the information to be pushed based on the push form prediction information.

[0051] The information push unit is configured to push push information in the target push format to the target object.

[0052] In an exemplary embodiment, the push pattern determination unit includes:

[0053] The fourth acquisition unit is configured to acquire preset object attribute information and preset operation information corresponding to each of the various push formats;

[0054] The target correction information determination unit is configured to determine the target correction information for the push pattern prediction information based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information.

[0055] The correction unit is configured to correct the push pattern prediction information based on the target correction information to obtain the target prediction information;

[0056] The fifth determining unit is configured to determine the target push pattern based on the target prediction information.

[0057] In an exemplary embodiment, the target correction information determination unit includes:

[0058] The first similarity determination unit is configured to perform matching of the object attribute information of the target object with the preset object attribute information corresponding to each of the multiple push forms, and obtain the first matching similarity between the target object and each of the multiple preset object attribute information.

[0059] The second similarity determination unit is configured to perform a match between the second historical operation information of the target object and the preset operation information corresponding to each of the multiple push forms, so as to obtain the second matching similarity between the target object and each of the multiple preset operation information.

[0060] The similarity processing unit is configured to perform a first matching similarity and a second matching similarity based on the target object and each of the multiple push formats to determine the matching push format and the non-matching push format corresponding to the target object.

[0061] The weight determination unit is configured to determine the first weight corresponding to the matching push form and the second weight corresponding to the non-matching push form.

[0062] The target correction information generation unit is configured to perform the task of determining the target correction information based on the first weight and the second weight.

[0063] In one exemplary embodiment, the apparatus further includes:

[0064] The fifth acquisition unit is configured to execute a push information display request triggered by the target object and acquire the historical operation information of the target object within a preset historical time period; the preset historical time period is a time period with the current time as the end point and a preset duration.

[0065] The sixth determining unit is configured to determine the target object's historical operation information on the pushed information within the preset historical time period as the second historical operation information.

[0066] According to a fourth aspect of the present disclosure, an information processing model training apparatus is provided, comprising:

[0067] The first acquisition unit is configured to acquire the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information.

[0068] The information analysis unit is configured to perform information analysis on the first historical operation information and determine training labels corresponding to the sample object based on the information analysis results; the training labels represent the sample object's attention information to the multiple push formats.

[0069] The first training unit is configured to train a preset machine learning model based on the object attribute information of the sample object, the first historical operation information, and the training label to obtain an information processing model.

[0070] In one exemplary embodiment, the apparatus further includes:

[0071] The second acquisition unit is configured to acquire a push form template corresponding to each of the multiple push forms; the push form template includes at least one target field.

[0072] The third acquisition unit is configured to acquire sample push material information;

[0073] The first matching unit is configured to perform the task of pushing the sample material information and matching it with the target fields in the push form templates corresponding to the various push forms, and to determine the matching push form template from the push form templates corresponding to the various push forms based on the matching results.

[0074] The information filling unit is configured to perform information filling on the matching push form template based on the sample push material information to obtain the sample push information.

[0075] In one exemplary embodiment, the sample push information includes exposure display information and a shape identifier of the push form;

[0076] The device further includes:

[0077] The information delivery unit is configured to send the sample push information to the terminal corresponding to the sample object; when the sample push information is exposed on the terminal, the terminal displays the exposure display information and the morphological identifier.

[0078] In one exemplary embodiment, the training labels corresponding to the sample objects include sub-labels corresponding to each of the various push formats;

[0079] The information analysis unit includes:

[0080] The first determining unit is configured to determine the target push information corresponding to each of the multiple push formats from the sample push information;

[0081] The second determining unit is configured to perform the following based on the first historical operation information: determining the number of target push information items being operated on in the target push information items corresponding to each push format;

[0082] The third determining unit is configured to determine the sub-tag corresponding to each push form based on the relationship between the number of target push information corresponding to each push form and the total number of sample push information.

[0083] The fourth determining unit is configured to determine the training label based on the sub-labels corresponding to each of the multiple push formats.

[0084] In an exemplary embodiment, the first historical operation information includes historical operation information corresponding to each of the multiple push formats;

[0085] The first training unit includes:

[0086] The joint input feature construction unit is configured to construct joint input features based on the object attribute information and the historical operation information corresponding to each of the multiple push forms.

[0087] The first prediction unit is configured to input the joint input features into the preset machine learning model to perform information prediction and obtain a predicted label.

[0088] The loss information determination unit is configured to determine loss information based on the predicted label and the training label;

[0089] The second training unit is configured to train the preset machine learning model based on the loss information to obtain the information processing model.

[0090] According to a fifth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the information processing model training method or information processing method as described above.

[0091] According to a sixth aspect of the present disclosure, a computer-readable storage medium is provided, wherein when instructions in the computer-readable storage medium are executed by a processor of a server, the server is able to perform the information processing model training method or the information processing method as described above.

[0092] According to a seventh aspect of the present disclosure, a computer program product is provided, the computer program product including a computer program stored in a readable storage medium, wherein at least one processor of a computer device reads from the readable storage medium and executes the computer program, causing the device to perform the above-described information processing model training method or information processing method.

[0093] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects:

[0094] This disclosure obtains the first historical operation information of sample objects to push information including various push formats, performs information analysis on the first historical operation information, and obtains training labels corresponding to the sample objects. The training labels can represent the sample objects' attention information to various push formats. Based on the object attribute information, the first historical operation information, and the training labels of the sample objects, a preset machine learning model is trained to obtain an information processing model. That is, this disclosure introduces a push format dimension for push information. The push format can represent the display form of the push information when a display operation is triggered. Therefore, the model can be trained based on the first historical operation information of sample objects to push information of different formats. This allows the trained information processing model to predict the degree of attention of the target object to different push formats based on the target object's historical operation information. Thus, information can be pushed based on the degree of attention to different push formats, expanding information push beyond the push content dimension to include push format dimensions, thereby achieving the expansion and diversity of push formats. Simultaneously, by pushing push information of push formats that the target object is interested in, the conversion rate of push information can be improved.

[0095] 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

[0096] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0097] Figure 1 This is a schematic diagram of an implementation environment according to an exemplary embodiment.

[0098] Figure 2 This is a flowchart illustrating an information processing model training method according to an exemplary embodiment.

[0099] Figure 3 This is a flowchart illustrating a sample information generation method according to an exemplary embodiment.

[0100] Figure 4 This is a flowchart illustrating a training label determination method according to an exemplary embodiment.

[0101] Figure 5 This is a flowchart illustrating a supervised model training method according to an exemplary embodiment.

[0102] Figure 6 This is a flowchart illustrating an information processing method according to an exemplary embodiment.

[0103] Figure 7 This is a flowchart illustrating a method for determining a target push pattern according to an exemplary embodiment.

[0104] Figure 8 This is a flowchart illustrating a method for determining historical operation information of a target object according to an exemplary embodiment.

[0105] Figure 9 This is a schematic diagram of an information processing model training device according to an exemplary embodiment.

[0106] Figure 10 This is a schematic diagram of an information processing apparatus according to an exemplary embodiment.

[0107] Figure 11 This is a schematic diagram of an electronic device structure according to an exemplary embodiment. Detailed Implementation

[0108] To enable those skilled in the art to better understand the technical solutions of this disclosure, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0109] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0110] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.

[0111] Please see Figure 1 The illustration shows an implementation environment provided by an embodiment of the present disclosure. The implementation environment may include at least one client 110 and an information push terminal 120, which can communicate with each other via a network.

[0112] Specifically, the target object can send a push information display request to the information push terminal 120 via the client 110. The information push terminal 120 can predict the push form information of the target object based on the object attribute information of the target object and the historical operation information of push information of at least one push form, and determine the target push form corresponding to the target object, and push the push information of the target push form to the target object. Furthermore, the information push terminal 120 can perform push form information prediction based on a preset information processing model.

[0113] Client 110 can communicate with information push terminal 120 based on browser / server (B / S) or client / server (C / S) mode. Client 110 may include physical devices such as smartphones, tablets, laptops, digital assistants, smart wearable devices, in-vehicle terminals, and servers, and may also include software running on the physical device, such as applications. The operating system running on client 110 in this embodiment may include, but is not limited to, Android, iOS, Linux, and Windows.

[0114] The information push terminal 120 and the client 110 can establish a communication connection via wired or wireless means. The information push terminal 120 may include an independently operating server, a distributed server, or a server cluster composed of multiple servers, wherein the server may be a cloud server.

[0115] To address the limitations of related technologies in terms of information delivery based on dimensions other than the content being pushed, and the limited range of push formats, this disclosure provides an information processing model training method, which may include:

[0116] S210. Obtain the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information.

[0117] Sample objects can be a subset of objects determined from the full set of pushed objects. When determining sample objects, they can be randomly selected from the full set of pushed objects. By randomly selecting sample objects, the features of the sample objects can be comprehensively represented by the full set of pushed objects, improving the generalization of sample selection and thus improving the generalization of the trained model. Alternatively, sample objects can be selected based on the historical operation information of the sample push information from the full set of pushed objects. Push objects that have operated on the sample push information more than or equal to a preset number of times can be identified as sample objects. For push objects with more than or equal to a preset number of operations, the corresponding amount of historical operation information can also be more than or equal to a preset number, thereby increasing the amount of sample information and thus improving the accuracy of model training.

[0118] In one optional embodiment, the operation on the sample push information may include exposure operation, click operation, conversion operation, etc. of the sample push information; correspondingly, the first historical operation information may include the sample object's exposure operation information, click operation information, conversion operation information, etc. of the sample push information; for example, for sample push information a, its corresponding first historical operation information may include exposure operation information, or include exposure operation information and click operation information, or include exposure operation information, click operation information and conversion operation information.

[0119] Furthermore, the sample push notifications include various push notification formats, each with its own format. The push notification format represents how the notification will be displayed when a display action is triggered. For example, a push notification format could be an image + text format, a video + text format, a live stream + text format, etc. One push notification can correspond to one push notification format, and for the same push material, multiple push notifications can be generated, each with a different push notification format.

[0120] S220. Perform information analysis on the first historical operation information, and determine the training label corresponding to the sample object based on the information analysis results; the training label represents the sample object's attention information to the multiple push formats.

[0121] In one optional embodiment, the first historical operation information may include exposure operation information, click operation information, conversion operation information, etc., of the sample object to the sample push information. That is, the first historical operation information is determined from the dimension of push information. Since each push information has a corresponding push format, the data of the push format dimension can be determined based on the data of the push information dimension. After determining the data of the push format dimension, the training labels corresponding to the sample object can be determined accordingly.

[0122] In one example, training labels can represent the sample object's attention information to various push notification formats. That is, training labels can be data items containing multiple data dimensions, each data dimension corresponding to a push notification format. The number of data dimensions is the same as the number of push notification formats. The data of each data dimension can represent the sample object's attention information or degree of attention to the push notification format corresponding to that data dimension. Generally, the larger the data of a certain data dimension, the higher the degree of attention of the sample object to the push notification format corresponding to that data dimension; conversely, the lower the degree of attention.

[0123] This allows us to determine the push notification format dimension based on data from the sample object's push notification information dimensions, such as exposure operation information, click operation information, and conversion operation information, and then determine the training label corresponding to the sample object based on the push notification format dimension data.

[0124] S230. Based on the object attribute information of the sample object, the first historical operation information, and the training label, a preset machine learning model is trained to obtain an information processing model.

[0125] In this embodiment, the object attribute information can be static object feature information, which can be object feature information that will not change within a preset time period. Therefore, during model training, the input feature information of the model can be generated based on the object attribute information of the sample object and the first historical operation information. This input feature information, along with the training labels, is used as input to a preset machine learning model to achieve supervised training of the preset machine learning model.

[0126] This disclosure introduces a push notification format dimension. The push notification format can characterize the display form of the push notification when a display operation is triggered. Therefore, a model can be trained based on the first historical operation information of push notifications with different formats from sample objects. This allows the trained information processing model to predict the target audience's level of attention to different push notification formats based on their historical operation information. This enables information push based on the level of attention to different push notification formats, expanding information push beyond the push notification content dimension. This achieves both the expansion and diversity of push notification formats. Furthermore, by pushing push notifications with formats that the target audience is interested in, the conversion rate of push notifications can be improved.

[0127] In one example, push notifications with corresponding push formats can be generated based on push format templates. Each push format can correspond to one push format template. Therefore, when push notifications with a specific push format are needed, the push material information can be filled into the push format template corresponding to that push format. For details, please refer to [link to relevant documentation]. Figure 3It illustrates a method for generating sample information, which may include:

[0128] S310. Obtain the push form template corresponding to each of the multiple push forms; the push form template includes at least one target field.

[0129] S320. Obtain sample push material information.

[0130] S330. The sample push material information is matched with the target fields in the push form templates corresponding to the various push forms, and the matching push form template is determined from the push form templates corresponding to the various push forms based on the matching results.

[0131] S340. Based on the sample push material information, fill the information of the matching push form template to obtain the sample push information.

[0132] Specifically, each type of push notification has a corresponding push notification template. Push notification formats can be image + text, video + text, live stream + text, etc. Correspondingly, the push notification template for image + text may include text description fields, image fields, etc.; the push notification template for video + text may include text description fields, video fields, etc.; and the push notification template for live stream + text may include text description fields, live stream ID fields, etc.

[0133] The sample push notification material information may include one or more of the following: text information, image information, video information, and live stream information. This allows for matching of the sample push notification material information with target fields in the corresponding push format templates for each of the various push formats, resulting in template matching results. For example, if the sample push notification material information includes text, image, and video information, during field matching, text and image information can be matched with the text description and image fields in the push format template corresponding to the image + text format, and text and video information can be matched with the text description and video fields in the push format template corresponding to the video + text format. The resulting template matching results are the sample push notification material information matched with the push format templates corresponding to the image + text format and the video + text format. Therefore, the push format templates corresponding to the image + text format and the video + text format can be identified as the matching push format templates.

[0134] Once a matching push format template is determined, relevant information from the sample push material information can be filled into the corresponding fields of the matching push format template to generate sample push information. Furthermore, when there are multiple matching push format templates, they can be filled sequentially, one at a time; or, based on the sample push material information, a matching push format template containing each field information can be determined, and then that field information can be filled into the corresponding matching push format template, thus achieving one-time filling of the same field information.

[0135] By pre-generating push form templates corresponding to each push form and setting the target fields contained in the push form templates corresponding to each push form, when sample push material information is obtained, the corresponding matching push form template can be determined based on the matching of the target fields, thereby improving the accuracy and convenience of determining the matching push form template. Furthermore, by filling the matching push form template with information based on the sample material push information to obtain the corresponding sample push information, the adaptability and convenience of the sample push information can be improved.

[0136] In one optional embodiment, the push notification format can be displayed when the push notification is shown to the recipient, allowing the recipient to determine the push notification format and whether further clicks are needed. Specifically, the sample push notification includes not only the main content information but also exposure display information and a format identifier for the push notification format. Correspondingly, when the sample push notification is sent to the client corresponding to the sample recipient, the exposure display information and the format identifier are also sent to the client. Thus, when the sample push notification is shown on the client, the client can display the exposure display information and the format identifier.

[0137] Exposure display information may include one or more of the following: image information, video information, text information, and live broadcast information. Specifically, exposure display information can be understood as the cover information of sample push information.

[0138] The form identifier of the push notification can be static identifier information displayed on the exposure display information. Specifically, the form identifier can be "image and text", "video text", "live text", etc. Thus, the sample object can determine the push notification form of the currently exposed sample push information based on the form identifier information on the exposure display information, and determine whether further click operations are needed on the currently exposed sample push information.

[0139] In another example, the push format identifier can be dynamic identifier information; for example, when the exposure display information includes video information or live broadcast information, the push format of the corresponding sample push information can be determined to be video or live broadcast, and the push format of the currently exposed sample push information can be determined based on the exposure display information. At this time, the push format identifier information can be the exposure display information.

[0140] Therefore, the sample push information includes exposure display information and push form identifier. When the client exposes the sample push information, the exposure display information and form identifier can be displayed, so that the sample object can quickly determine whether further click operation is needed on the currently exposed sample push information based on the exposure display information and form identifier, thereby improving the efficiency of the first historical operation information collection.

[0141] Since training labels can characterize the attention information of sample objects to various push notification formats, the training labels corresponding to sample objects can include sub-labels corresponding to each of the various push notification formats. Multiple sub-labels can be used to determine the corresponding training labels. Please refer to [link / reference] for details. Figure 4 It illustrates a training label determination method, which may include:

[0142] S410. Determine the target push information corresponding to each of the multiple push formats from the sample push information.

[0143] S420. Based on the first historical operation information, determine the number of target push information to be operated in the target push information corresponding to each push form.

[0144] S430. Based on the relationship between the number of target push information corresponding to each push form and the total number of sample push information, determine the sub-tag corresponding to each push form.

[0145] S440. Determine the training label based on the sub-labels corresponding to each of the various push formats.

[0146] The sample push information may include multiple sample push information. The target push information corresponding to each push form can be determined from the sample push information. Specifically, it can be classified by the form identifier carried by each sample push information, so as to obtain the target push information corresponding to each push form; or it can be that each sample push information is traversed and the target push information corresponding to each push form is determined based on the push form of each sample push information.

[0147] For each type of push notification, the first historical operation information can be used to determine which push notifications have corresponding historical operation information and which do not. This allows us to determine the number of target push notifications being operated on for each type of push notification. In this embodiment, the larger the number of target push notifications being operated on for a push notification, the higher the attention the sample object pays to that push notification; conversely, the lower the attention.

[0148] Based on the relationship between the number of target push messages corresponding to each push format and the total number of sample push messages, the sub-tags corresponding to each push format are determined. Specifically, the sub-tags corresponding to each push format can be determined based on the ratio of the number of target push messages corresponding to each push format to the total number of sample push messages; alternatively, the ratio of the number of target push messages corresponding to each push format to the total number of sample push messages can be weighted, and the sub-tags corresponding to each push format can be determined based on the weighted ratio. Here, the weight values ​​used for weighting can be pre-set based on the push feedback information of each push format.

[0149] In another optional embodiment, after determining the ratio of the number of target push information corresponding to each push form to the total number of sample push information, the ratio corresponding to each push form can be normalized, thereby determining the normalized ratio as the sub-tag corresponding to each push form.

[0150] By combining the sub-tags corresponding to each push format according to a preset format, the corresponding training tags can be obtained; for example, the sub-tags can be combined directly; or they can be represented in the form of key values ​​and then combined. This embodiment does not make specific limitations.

[0151] The first historical operation information can characterize the operation of the sample object on the sample push information. Based on the first historical operation information, the number of push information corresponding to each push form can be determined, which can improve the accuracy and convenience of determining the number of push information of the target. Furthermore, the training label information includes sub-labels corresponding to each of the various push forms, which can improve the comprehensiveness of the training label representation. Using the more comprehensive training labels as supervision labels for supervised training can improve the accuracy of information processing model training.

[0152] In this embodiment, the model can be trained using supervised training; for details, please refer to [link to relevant documentation]. Figure 5 It illustrates a supervised model training method, which may include:

[0153] S510. Based on the object attribute information and the historical operation information corresponding to each of the various push formats, construct joint input features.

[0154] S520. Input the joint input features into the preset machine learning model to perform information prediction and obtain the predicted label.

[0155] S530. Determine loss information based on the predicted label and the training label.

[0156] S540. The preset machine learning model is trained based on the loss information to obtain the information processing model.

[0157] In one example, the first historical operation information may also include historical operation information corresponding to each of the multiple push formats. The historical operation information corresponding to each push format may be the number of target push information that has been operated in the target push information corresponding to each push format; or it may be the ratio of the number of target push information that has been operated in the target push information corresponding to each push format to the total number of sample push information.

[0158] Thus, push form features can be constructed based on the historical operation information corresponding to various push forms. Push form features can be multi-dimensional features, with each dimension corresponding to a push form. Then, joint input features can be constructed based on the object attributes of the sample object and the push form features. The joint input features are then input into a preset machine learning model for information prediction to obtain predicted labels. Based on the loss information between the predicted labels and the training labels, the model parameters of the preset machine learning model are adjusted to obtain the information processing model.

[0159] Object attribute information and historical operation information corresponding to various push formats are information of different dimensions. By constructing joint input features, the information of different dimensions can be combined. On the one hand, this can improve the effective expression of input features; on the other hand, it can facilitate model reading and processing, that is, improve the readability of input features. Therefore, training the preset machine learning model based on the readable joint features can improve the efficiency of model training.

[0160] If an information processing model has already been trained, it can be used to process information; please refer to [link / reference needed]. Figure 6 It illustrates an information processing method that may include:

[0161] S610. Obtain object attribute information of the target object, and second historical operation information of the target object on the pushed information; the pushed information includes push information in at least one push form.

[0162] The target object can be the object to be pushed to. The object attribute information of the target object and the target object's second historical operation information regarding previously pushed information can be obtained. The second historical operation information is similar to the first historical operation information; that is, the second historical operation information may include the target object's operation information regarding previously pushed information, and may also include the target object's historical operation information regarding at least one push format. When the second historical operation information may include the target object's operation information regarding previously pushed information, further information analysis can be performed on the second historical operation information to obtain the target object's historical operation information regarding at least one push format. Since each pushed information item has a corresponding push format, the pushed information to the target object includes push information in at least one push format.

[0163] S620. Based on the object attribute information of the target object, the second historical operation information, and the information processing model, predict the push form information of the target object to obtain push form prediction information; the information processing model is obtained by training a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push information, and training labels; the sample push information includes push information in multiple push forms, and the push form represents the display format of the sample push information when a display operation is triggered on the sample push information; the training labels are determined based on the information analysis results of the first historical operation information; the training labels represent the sample object's attention information to the multiple push forms.

[0164] In an optional embodiment, before predicting the push form information based on the information processing model, corresponding feature encoding can be performed based on the object attribute information of the target object and the second historical operation information to obtain a feature information expression suitable for model input. Specifically, the object attribute information of the target object can be feature-encoded or mapped to an attribute feature space to obtain feature information corresponding to the object attribute information of the target object. Similarly, the second historical operation information can also be feature-encoded or mapped to an operation feature space to obtain feature information corresponding to the second historical operation information. Thus, the feature information corresponding to the object attribute information and the feature information corresponding to the second historical operation information can be used as the input feature information of the information processing model. By performing feature expression on the object attribute information or historical operation information to generate corresponding feature information, it is easier for the information processing model to read the information, improving the readability of the information and thus improving the prediction efficiency of the model.

[0165] In another optional embodiment, corresponding joint input features can be generated based on the feature information corresponding to the object attribute information and the feature information corresponding to the second historical operation information, and the joint input features can be used as the input features of the information processing model.

[0166] In one example, if the target object is a newly registered object and has no corresponding second historical operation information, the input features corresponding to the second historical operation information can be set to empty when determining the input features of the information processing model. This allows the input features of the information processing model to be determined based on the object attribute information of the target object, enabling the model to predict the corresponding push notification pattern. If the target object is not a newly registered user, it has corresponding second historical operation information. The corresponding input features can be determined based on the object attribute information and the second historical operation information, and then input into the information processing model to predict the corresponding push notification pattern.

[0167] For the training process of the information processing model, please refer to the above content of this embodiment, which will not be repeated here.

[0168] S630. Determine the target push format of the information to be pushed based on the push format prediction information.

[0169] Push format prediction information can characterize the target object's attention to at least one push format. Specifically, push format prediction information can include data in multiple dimensions, each dimension corresponding to a push format. The prediction data for each dimension characterizes the target object's predicted attention to the push format corresponding to that dimension. The larger the prediction data, the higher the target object's predicted attention to the push format corresponding to that dimension; conversely, the lower the target object's predicted attention to the push format corresponding to that dimension.

[0170] The target push format can be one or more push formats that the target audience pays close attention to, and its specifics can be determined by data from multiple dimensions in the push format prediction information.

[0171] S640. Push push information of the target push form to the target object.

[0172] Once the target push format is determined, the push material information can be filled into the push format template corresponding to the target push format by obtaining the push format template corresponding to the target push format, so as to obtain the push information of the target push format and push the push information of the target push format to the target object.

[0173] In this embodiment, the information processing model can predict push notification patterns based on the target object's attribute information and historical operation information, thereby obtaining the target object's attention information for each push notification pattern. Based on the target object's attention information for each push notification pattern, the target push notification pattern is determined, and push notification information of the target push notification pattern is pushed to the target object. In other words, this embodiment can push information to the target object based on the push notification pattern dimension, thus expanding the information push notification format and improving the diversity of push notification formats.

[0174] In one example, push notification pattern prediction information can include data from multiple dimensions. Once push notification pattern prediction information is obtained, the corresponding target push notification pattern can be determined directly based on the data from multiple dimensions. For example, the push notification pattern corresponding to the dimension with data greater than a preset value can be determined as the target push notification pattern; or, based on the data size of multiple dimensions, a preset number of dimensions can be selected from multiple dimensions, and the push notification pattern corresponding to this preset number of dimensions can be determined as the target push notification pattern.

[0175] In another example, the push notification pattern prediction information can be further processed to determine the corresponding target push notification pattern; please refer to [link / reference needed]. Figure 7 It illustrates a method for determining the target push pattern, which may include:

[0176] S710. Obtain the preset object attribute information and preset operation information corresponding to each of the various push formats.

[0177] S720. Based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information, determine the target correction information for the push pattern prediction information.

[0178] S730. Based on the target correction information, the push pattern prediction information is corrected to obtain the target prediction information.

[0179] S740. Determine the target delivery pattern based on the target prediction information.

[0180] In one optional embodiment, during the information push process, operation information of each object on the pushed information can be continuously collected, thereby obtaining operation information of each object on multiple push forms. For each push form, information analysis can be performed on the object operating the push information of that push form to obtain common object attribute information and common operation information of the push information operating that push form. Thus, common object attribute information and common operation information corresponding to each push form can be obtained. The common object attribute information can be determined as preset object attribute information, and the common operation information can be determined as preset operation information. Thus, preset object attribute information and preset operation information corresponding to each of the multiple push forms can be obtained.

[0181] The target object's attribute information is matched with preset object attribute information corresponding to various push formats to obtain a first matching similarity between the target object and each of the preset object attribute information. The target object's second historical operation information is matched with preset operation information corresponding to each of the various push formats to obtain a second matching similarity between the target object and each of the preset operation information. Then, based on the first and second matching similarities between the target object and each of the various push formats, the matching push format and the non-matching push format corresponding to the target object are determined. The first weight corresponding to the matching push format and the second weight corresponding to the non-matching push format are determined. The target correction information is determined based on the first and second weights.

[0182] When determining the matching push format, it can be based on the sum of the first matching similarity and the second matching similarity, or the weighted sum of the first matching similarity and the second matching similarity, or the product of the first matching similarity and the second matching similarity.

[0183] For example, there are multiple push notification formats, including push notification format 1, push notification format 2, and push notification format 3. The similarity between the target object's object attribute information and the preset object attribute information corresponding to push notification format 1 is s1, and the similarity between the target object's second historical operation information and the preset operation information corresponding to push notification format 1 is s2. The similarity between the target object's object attribute information and the preset object attribute information corresponding to push notification format 2 is s3, and the similarity between the target object's second historical operation information and the preset operation information corresponding to push notification format 2 is s4. The similarity between the target object's object attribute information and the preset object attribute information corresponding to push notification format 3 is s5, and the similarity between the target object's second historical operation information and the preset operation information corresponding to push notification format 3 is s6. The sum of s1 and s2, or a weighted sum, or a product, is calculated to obtain the similarity fusion value S1 corresponding to push notification format 1. Correspondingly, the similarity fusion value S2 corresponding to push notification format 2 and the similarity fusion value S3 corresponding to push notification format 3 can be obtained. The push notification format with the largest similarity fusion value is determined as the matching push notification format. Once the matching push notification format is determined, corresponding weights can be assigned to it, with the weight of the matching push notification format being greater than that of the non-matching push notification format. The weights for each push notification format can be obtained based on a preset weight allocation method or through random allocation. This allows the weights corresponding to each of the various push notification formats to be determined as target correction information.

[0184] By matching the target object's attribute information and historical operation information with the preset object attribute information and preset operation information corresponding to various push formats, the corresponding matching similarity is obtained. The matching push format and non-matching push format are determined by similarity fusion, thereby improving the rationality and accuracy of the matching push format determination. Furthermore, the target correction information is determined based on the determined matching push format and non-matching push format, thereby improving the accuracy of the target correction information determination.

[0185] In one example, the push notification pattern prediction information can include data from multiple dimensions, each corresponding to a push notification pattern. The prediction data for each dimension represents the target object's level of attention to the push notification pattern corresponding to that dimension. Thus, target correction information can be used to weight the data from multiple dimensions to obtain target prediction information, and the target push notification pattern can be determined based on the target prediction information.

[0186] Since the preset object attribute information and preset operation information corresponding to each of the various push forms can characterize the common information of the objects that operate on the corresponding push forms, the push form prediction information can be corrected based on the corresponding preset object attribute information and preset operation information, based on the prediction through the information processing model. This can supplement the information processing model and improve the accuracy of the target push form determination.

[0187] In one optional embodiment, during the information push process, the target object can perform operations on different push information, thus the second historical operation information is constantly changing; accordingly, please refer to Figure 8 This disclosure provides a method for determining the historical operation information of a target object, which may include:

[0188] S810. In response to the push information display request triggered by the target object, obtain the historical operation information of the target object within a preset historical time period; the preset historical time period is a time period with the current time as the end point and a preset duration.

[0189] S820. The historical operation information of the target object on the pushed information within the preset historical time period is determined as the second historical operation information.

[0190] Each time a target object triggers a push notification display request, its historical operation information within a preset historical time period can be obtained. This historical operation information of the target object to pushed information within the preset historical time period is defined as the second historical operation information. The preset historical time period can be a period ending at the current time with a preset duration. The preset historical time period can be the last 5 days, the last week, or the last month, etc. Because the target object's second historical operation information is constantly changing, when the target object initiates a push notification display request at time node a, the push notification to the target object may be of push notification type 1; when the target object initiates a push notification display request at time node b, the push notification to the target object may be of push notification type 2.

[0191] Therefore, it can be seen that the historical operation information obtained may be different when the push information display request is triggered at different times. That is, it is possible to obtain the corresponding second historical operation information in real time at different times when the push information display request is triggered. This enables the prediction of the corresponding push form based on the dynamically changing second historical operation information, as well as the determination of the target push form, realizing the dynamic determination of the target push form, and thus realizing the dynamic push of push information.

[0192] Figure 9 This disclosure provides an information processing model training device, comprising:

[0193] The first acquisition unit 910 is configured to acquire the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information.

[0194] The information analysis unit 920 is configured to perform information analysis on the first historical operation information and determine training labels corresponding to the sample object based on the information analysis results; the training labels represent the sample object's attention information to the multiple push formats.

[0195] The first training unit 930 is configured to train a preset machine learning model based on the object attribute information of the sample object, the first historical operation information, and the training label to obtain an information processing model.

[0196] In one exemplary embodiment, the apparatus further includes:

[0197] The second acquisition unit is configured to acquire a push form template corresponding to each of the multiple push forms; the push form template includes at least one target field.

[0198] The third acquisition unit is configured to acquire sample push material information;

[0199] The first matching unit is configured to perform the task of pushing the sample material information and matching it with the target fields in the push form templates corresponding to the various push forms, and to determine the matching push form template from the push form templates corresponding to the various push forms based on the matching results.

[0200] The information filling unit is configured to perform information filling on the matching push form template based on the sample push material information to obtain the sample push information.

[0201] In one exemplary embodiment, the sample push information includes exposure display information and a shape identifier of the push form;

[0202] The device further includes:

[0203] The information delivery unit is configured to send the sample push information to the terminal corresponding to the sample object; when the sample push information is exposed on the terminal, the terminal displays the exposure display information and the morphological identifier.

[0204] In one exemplary embodiment, the training labels corresponding to the sample objects include sub-labels corresponding to each of the various push formats;

[0205] The information analysis unit includes:

[0206] The first determining unit is configured to determine the target push information corresponding to each of the multiple push formats from the sample push information;

[0207] The second determining unit is configured to perform the following based on the first historical operation information: determining the number of target push information items being operated on in the target push information items corresponding to each push format;

[0208] The third determining unit is configured to determine the sub-tag corresponding to each push form based on the relationship between the number of target push information corresponding to each push form and the total number of sample push information.

[0209] The fourth determining unit is configured to determine the training label based on the sub-labels corresponding to each of the multiple push formats.

[0210] In an exemplary embodiment, the first historical operation information includes historical operation information corresponding to each of the multiple push formats;

[0211] The first training unit 930 includes:

[0212] The joint input feature construction unit is configured to construct joint input features based on the object attribute information and the historical operation information corresponding to each of the multiple push forms.

[0213] The first prediction unit is configured to input the joint input features into the preset machine learning model to perform information prediction and obtain a predicted label.

[0214] The loss information determination unit is configured to determine loss information based on the predicted label and the training label;

[0215] The second training unit is configured to train the preset machine learning model based on the loss information to obtain the information processing model.

[0216] Figure 10 This disclosure provides an information processing apparatus, comprising:

[0217] The second information acquisition unit 1010 is configured to acquire object attribute information of a target object and second historical operation information of the target object on pushed information; the pushed information includes push information in at least one type of push.

[0218] The second prediction unit 1020 is configured to predict the push format information of the target object based on the object attribute information of the target object, the second historical operation information, and the information processing model, to obtain push format prediction information; the information processing model is trained on a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push information, and training labels; the sample push information includes push information in multiple push formats, and the push format represents the display form of the sample push information when a display operation is triggered on the sample push information; the training labels are determined based on the information analysis results of the first historical operation information; the training labels represent the sample object's attention information to the multiple push formats;

[0219] The push form determination unit 1030 is configured to determine the target push form of the information to be pushed based on the push form prediction information.

[0220] The information push unit 1040 is configured to push push information of the target push form to the target object.

[0221] In an exemplary embodiment, the push pattern determination unit 1030 includes:

[0222] The fourth acquisition unit is configured to acquire preset object attribute information and preset operation information corresponding to each of the various push formats;

[0223] The target correction information determination unit is configured to determine the target correction information for the push pattern prediction information based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information.

[0224] The correction unit is configured to correct the push pattern prediction information based on the target correction information to obtain the target prediction information;

[0225] The fifth determining unit is configured to determine the target push pattern based on the target prediction information.

[0226] In an exemplary embodiment, the target correction information determination unit includes:

[0227] The first similarity determination unit is configured to perform matching of the object attribute information of the target object with the preset object attribute information corresponding to each of the multiple push forms, and obtain the first matching similarity between the target object and each of the multiple preset object attribute information.

[0228] The second similarity determination unit is configured to perform a match between the second historical operation information of the target object and the preset operation information corresponding to each of the multiple push forms, so as to obtain the second matching similarity between the target object and each of the multiple preset operation information.

[0229] The similarity processing unit is configured to perform a first matching similarity and a second matching similarity based on the target object and each of the multiple push formats to determine the matching push format and the non-matching push format corresponding to the target object.

[0230] The weight determination unit is configured to determine the first weight corresponding to the matching push form and the second weight corresponding to the non-matching push form.

[0231] The target correction information generation unit is configured to perform the task of determining the target correction information based on the first weight and the second weight.

[0232] In one exemplary embodiment, the apparatus further includes:

[0233] The fifth acquisition unit is configured to execute a push information display request triggered by the target object and acquire the historical operation information of the target object within a preset historical time period; the preset historical time period is a time period with the current time as the end point and a preset duration.

[0234] The sixth determining unit is configured to determine the target object's historical operation information on the pushed information within the preset historical time period as the second historical operation information.

[0235] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided. Optionally, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is able to perform any of the methods described above.

[0236] In an exemplary embodiment, a computer program product is also provided, the computer program product including a computer program stored in a readable storage medium, wherein at least one processor of a computer device reads from the readable storage medium and executes the computer program, causing the device to perform any of the methods described above.

[0237] Figure 11 This is a block diagram illustrating an electronic device for an information processing model training method or an information processing method according to an exemplary embodiment. The electronic device may be a server, and its internal structure diagram may be as follows: Figure 11As shown, the electronic device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements an information processing model training method or an information processing method.

[0238] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present disclosure and does not constitute a limitation on the electronic device to which the present disclosure is applied. A specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0239] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0240] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. An information processing method, characterized in that, include: Obtain object attribute information of the target object, and second historical operation information of the target object on the pushed information; the pushed information includes push information in at least one push form; Based on the object attribute information of the target object, the second historical operation information, and the information processing model, the push form information of the target object is predicted to obtain push form prediction information. The information processing model is obtained by training a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push information, and training labels; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when the display operation is triggered on the sample push information. The training labels are determined based on the information analysis results of the first historical operation information; the training labels represent the sample object's attention information to the multiple push formats. The target push format of the information to be pushed is determined based on the push format prediction information. Push the push information in the target push format to the target object.

2. The method according to claim 1, characterized in that, Determining the target push format of the information to be pushed based on the push format prediction information includes: Obtain preset object attribute information and preset operation information corresponding to various push formats; Based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information, target correction information for the push pattern prediction information is determined; Based on the target correction information, the push pattern prediction information is corrected to obtain the target prediction information; The target delivery pattern is determined based on the target prediction information.

3. The method according to claim 2, characterized in that, The step of determining the target correction information for the push pattern prediction information based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information includes: The object attribute information of the target object is matched with the preset object attribute information corresponding to each of the multiple push forms to obtain the first matching similarity between the target object and each of the multiple preset object attribute information. The second historical operation information of the target object is matched with the preset operation information corresponding to each of the multiple push forms to obtain the second matching similarity between the target object and each of the multiple preset operation information. Based on the first matching similarity and the second matching similarity corresponding to the target object and each of the various push formats, the matching push format and the non-matching push format corresponding to the target object are determined. Determine the first weight corresponding to the matching push form and the second weight corresponding to the non-matching push form; The target correction information is determined based on the first weight and the second weight.

4. The method according to claim 1, characterized in that, The method further includes: In response to a push notification display request triggered by the target object, the historical operation information of the target object within a preset historical time period is obtained; the preset historical time period is a time period with the current time as the end point and a preset duration. The target object's historical operation information on the pushed information within the preset historical time period is determined as the second historical operation information.

5. A method for training an information processing model, characterized in that, include: Obtain the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information; The first historical operation information is analyzed, and training labels corresponding to the sample object are determined based on the analysis results; the training labels represent the sample object's attention information to the various push formats. Based on the object attribute information of the sample object, the first historical operation information, and the training label, a preset machine learning model is trained to obtain an information processing model.

6. The method according to claim 5, characterized in that, Before obtaining the first historical operation information of the sample object on the sample push information, the method further includes: Obtain push form templates corresponding to each of the various push forms; each push form template includes at least one target field; Obtain sample push material information; The sample push material information is matched with the target fields in the push form templates corresponding to the various push forms, and the matching push form template is determined from the push form templates corresponding to the various push forms based on the matching results. Based on the sample push material information, the matching push pattern template is populated with information to obtain the sample push information.

7. The method according to claim 5, characterized in that, The sample push information includes exposure display information and push form identifier; Before obtaining the first historical operation information of the sample object on the sample push information, the method further includes: The sample push information is sent to the client corresponding to the sample object; when the sample push information is exposed on the client, the client displays the exposure display information and the shape identifier.

8. The method according to claim 5, characterized in that, The training labels corresponding to the sample objects include sub-labels corresponding to each of the various push formats; The step of performing information analysis on the first historical operation information and determining the training label corresponding to the sample object based on the information analysis results includes: The target push information corresponding to each of the various push formats is determined from the sample push information; Based on the first historical operation information, determine the number of target push information being operated on in the target push information corresponding to each push form; Based on the relationship between the number of pushed information to the target corresponding to each push form and the total number of pushed information in the sample, the sub-tag corresponding to each push form is determined; The training labels are determined based on the sub-labels corresponding to each of the various push formats.

9. The method according to claim 5, characterized in that, The first historical operation information includes historical operation information corresponding to each of the various push formats; The step of training a preset machine learning model based on the object attribute information of the sample object, the first historical operation information, and the training labels to obtain an information processing model includes: Based on the object attribute information and the historical operation information corresponding to each of the various push formats, a joint input feature is constructed. The joint input features are input into the preset machine learning model for information prediction to obtain the predicted label; Loss information is determined based on the predicted labels and the training labels; The information processing model is obtained by training the preset machine learning model based on the loss information.

10. An information processing device, characterized in that, include: The second information acquisition unit is configured to acquire object attribute information of the target object and second historical operation information of the target object on the pushed information; the pushed information includes push information in at least one type. The second prediction unit is configured to perform a prediction of the push form information of the target object based on the object attribute information of the target object, the second historical operation information and the information processing model, so as to obtain push form prediction information. The information processing model is obtained by training a preset machine learning model based on the object attribute information of the sample object, the first historical operation information of the sample object on the sample push information, and training labels; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when the display operation is triggered on the sample push information. The training labels are determined based on the information analysis results of the first historical operation information; the training labels represent the sample object's attention information to the multiple push formats. The push form determination unit is configured to determine the target push form of the information to be pushed based on the push form prediction information. The information push unit is configured to push push information in the target push format to the target object.

11. The apparatus according to claim 10, characterized in that, The push pattern determination unit includes: The fourth acquisition unit is configured to acquire preset object attribute information and preset operation information corresponding to each of the various push formats; The target correction information determination unit is configured to determine the target correction information for the push pattern prediction information based on the object attribute information of the target object, the second historical operation information, the preset object attribute information, and the matching result of the preset operation information. The correction unit is configured to correct the push pattern prediction information based on the target correction information to obtain the target prediction information; The fifth determining unit is configured to determine the target push pattern based on the target prediction information.

12. The apparatus according to claim 11, characterized in that, The target correction information determination unit includes: The first similarity determination unit is configured to perform matching of the object attribute information of the target object with the preset object attribute information corresponding to each of the multiple push forms, and obtain the first matching similarity between the target object and each of the multiple preset object attribute information. The second similarity determination unit is configured to perform a match between the second historical operation information of the target object and the preset operation information corresponding to each of the multiple push forms, so as to obtain the second matching similarity between the target object and each of the multiple preset operation information. The similarity processing unit is configured to perform a first matching similarity and a second matching similarity based on the target object and each of the multiple push formats to determine the matching push format and the non-matching push format corresponding to the target object. The weight determination unit is configured to determine the first weight corresponding to the matching push form and the second weight corresponding to the non-matching push form. The target correction information generation unit is configured to perform the task of determining the target correction information based on the first weight and the second weight.

13. The apparatus according to claim 10, characterized in that, The device further includes: The fifth acquisition unit is configured to execute a push information display request triggered by the target object and acquire the historical operation information of the target object within a preset historical time period; the preset historical time period is a time period with the current time as the end point and a preset duration. The sixth determining unit is configured to determine the target object's historical operation information on the pushed information within the preset historical time period as the second historical operation information.

14. An information processing model training device, characterized in that, include: The first acquisition unit is configured to acquire the first historical operation information of the sample object on the sample push information; the sample push information includes push information in various forms, and the push form represents the display form of the sample push information when a display operation is triggered on the sample push information. The information analysis unit is configured to perform information analysis on the first historical operation information and determine the training label corresponding to the sample object based on the information analysis results. The training labels represent the sample objects' attention information regarding the various push notification formats; The first training unit is configured to train a preset machine learning model based on the object attribute information of the sample object, the first historical operation information, and the training label to obtain an information processing model.

15. The apparatus according to claim 14, characterized in that, The device further includes: The second acquisition unit is configured to acquire a push form template corresponding to each of the multiple push forms; the push form template includes at least one target field. The third acquisition unit is configured to acquire sample push material information; The first matching unit is configured to perform the task of pushing the sample material information and matching it with the target fields in the push form templates corresponding to the various push forms, and to determine the matching push form template from the push form templates corresponding to the various push forms based on the matching results. The information filling unit is configured to perform information filling on the matching push form template based on the sample push material information to obtain the sample push information.

16. The apparatus according to claim 14, characterized in that, The sample push information includes exposure display information and push form identifier; The device further includes: The information delivery unit is configured to send the sample push information to the terminal corresponding to the sample object; when the sample push information is exposed on the terminal, the terminal displays the exposure display information and the morphological identifier.

17. The apparatus according to claim 14, characterized in that, The training labels corresponding to the sample objects include sub-labels corresponding to each of the various push formats; The information analysis unit includes: The first determining unit is configured to determine the target push information corresponding to each of the multiple push formats from the sample push information; The second determining unit is configured to perform the following based on the first historical operation information: determining the number of target push information items being operated on in the target push information items corresponding to each push format; The third determining unit is configured to determine the sub-tag corresponding to each push form based on the relationship between the number of target push information corresponding to each push form and the total number of sample push information. The fourth determining unit is configured to determine the training label based on the sub-labels corresponding to each of the multiple push formats.

18. The apparatus according to claim 14, characterized in that, The first historical operation information includes historical operation information corresponding to each of the various push formats; The first training unit includes: The joint input feature construction unit is configured to construct joint input features based on the object attribute information and the historical operation information corresponding to each of the multiple push forms. The first prediction unit is configured to input the joint input features into the preset machine learning model to perform information prediction and obtain a predicted label. The loss information determination unit is configured to determine loss information based on the predicted label and the training label; The second training unit is configured to train the preset machine learning model based on the loss information to obtain the information processing model.

19. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the information processing method as described in any one of claims 1 to 4, or the information processing model training method as described in any one of claims 5 to 9.

20. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is able to perform the information processing method as described in any one of claims 1 to 4, or the information processing model training method as described in any one of claims 5 to 9.

21. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the information processing method of any one of claims 1 to 4, or the information processing model training method of any one of claims 5 to 9.

Images