CN117640743A - Message pushing method, device, equipment and medium based on user behavior - Google Patents
Message pushing method, device, equipment and medium based on user behavior Download PDFInfo
- Publication number
- CN117640743A CN117640743A CN202311600826.7A CN202311600826A CN117640743A CN 117640743 A CN117640743 A CN 117640743A CN 202311600826 A CN202311600826 A CN 202311600826A CN 117640743 A CN117640743 A CN 117640743A
- Authority
- CN
- China
- Prior art keywords
- information
- target
- pushing
- message
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 230000006399 behavior Effects 0.000 claims abstract description 37
- 230000008520 organization Effects 0.000 claims description 24
- 238000004891 communication Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 12
- 238000012549 training Methods 0.000 description 9
- 238000012986 modification Methods 0.000 description 8
- 230000004048 modification Effects 0.000 description 8
- 230000007547 defect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a message pushing method, device, equipment and medium based on user behaviors. Extracting features of the target identification information based on the message pushing model to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as target pushing feature information; pushing the target message to the target user according to the target pushing characteristic information. Different new user identification information is different, different pushing characteristic information can be determined based on a message pushing model aiming at different new users, and message pushing is carried out according to the pushing characteristic information. Compared with the method for pushing the message by adopting a unified and unchanged mode for all new users, the self-adaption and the accuracy of the message pushing of the new users are improved.
Description
Technical Field
The present disclosure relates to the field of message pushing technologies, and in particular, to a method, an apparatus, a device, and a medium for pushing a message based on user behavior.
Background
After logging in the system, the user receives the real-time message notification push based on the ws protocol (namely websokcet protocol, which is used for the real-time message push of the service end after long link). Due to the wide variety of services and the user behavior preference characteristics, different users, different service types, different message channels and different time points can be notified of the reception preference. Therefore, the user self-defined configuration of each message notification service type, message notification channel, receiving time, whether to receive and the like is provided by the user message notification configuration page on the current service. When the service newly adds the message notification type and channel, the configuration modes of default opening, default non-workday receiving and the like of new service message notification are generally adopted.
According to the current scheme, big data statistics is carried out according to user, time period and message channel data, and the channel message push with the largest statistics number is inquired according to user information and the current time period. For the new user, no history preference data exists, so that the message pushing of the new user can only be performed by adopting default setting, the default setting is unified and unchanged for all the new users, and a matched message pushing mode can not be provided for different new users. Thus, the prior art message pushing method is less accurate for new users.
Disclosure of Invention
The application provides a message pushing method, device, equipment and medium based on user behaviors, which are used for solving the problem of poor message pushing accuracy in the prior art.
In a first aspect, the present application provides a message pushing method based on user behavior, where the method includes:
acquiring target identification information of a target user for message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information for pushing target messages to the target user based on the message pushing model; pushing the target message to the target user according to the target pushing characteristic information;
the message pushing model is used for extracting features of the target identification information to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as the target pushing feature information.
In a second aspect, the present application provides a message pushing device based on user behavior, the device comprising:
the determining module is used for acquiring target identification information of a target user of message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information of pushing target messages to the target user based on the message pushing model; the message pushing model is used for extracting characteristics of the target identification information to obtain identification characteristic information, determining pushing characteristic information matched with the target user according to the identification characteristic information, and taking the pushing characteristic information as the target pushing characteristic information;
and the pushing module is used for pushing the target message to the target user according to the target pushing characteristic information.
In a third aspect, the present application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;
a memory for storing a computer program;
and the processor is used for realizing the steps of the method when executing the program stored in the memory.
In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored therein, which when executed by a processor, implements the method steps.
The application provides a message pushing method, device, equipment and medium based on user behaviors, wherein the method comprises the following steps: acquiring target identification information of a target user for message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information for pushing target messages to the target user based on the message pushing model; pushing the target message to the target user according to the target pushing characteristic information; the message pushing model is used for extracting features of the target identification information to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as the target pushing feature information.
The technical scheme has the following advantages or beneficial effects:
in the application, after the electronic equipment acquires the target identification information of the target user of message pushing, the target identification information is input into the message pushing model. Extracting features of the target identification information based on the message pushing model to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as target pushing feature information; pushing the target message to the target user according to the target pushing characteristic information. The method and the device can determine the target pushing characteristic information according to the target identification information of the target user based on the message pushing model. Different new user identification information is different, different pushing characteristic information can be determined based on a message pushing model aiming at different new users, and message pushing is carried out according to the pushing characteristic information. Compared with the method for pushing the message by adopting a unified and unchanged mode for all new users, the self-adaption and the accuracy of the message pushing of the new users are improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of a message pushing process based on user behavior provided in the present application;
fig. 2 is a schematic diagram of a determining process of a message pushing model provided in the present application;
FIG. 3 is a schematic diagram of a process for determining push feature information matching a target user provided in the present application;
fig. 4 is a schematic process diagram of determining push feature information matched with a target user provided in the present application;
FIG. 5 is a schematic diagram of a process for determining similarity between identification feature information and each multidimensional feature point according to the present application;
fig. 6 is a schematic structural diagram of a message pushing device based on user behavior provided in the present application;
fig. 7 is a schematic structural diagram of an electronic device provided in the present application.
Detailed Description
For purposes of clarity and implementation of the present application, the following description will make clear and complete descriptions of exemplary implementations of the present application with reference to the accompanying drawings in which exemplary implementations of the present application are illustrated, it being apparent that the exemplary implementations described are only some, but not all, of the examples of the present application.
It should be noted that the brief description of the terms in the present application is only for convenience in understanding the embodiments described below, and is not intended to limit the embodiments of the present application. Unless otherwise indicated, these terms should be construed in their ordinary and customary meaning.
The terms "first," second, "" third and the like in the description and in the claims and in the above-described figures are used for distinguishing between similar or similar objects or entities and not necessarily for limiting a particular order or sequence, unless otherwise indicated. It is to be understood that the terms so used are interchangeable under appropriate circumstances.
The terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a product or apparatus that comprises a list of elements is not necessarily limited to all elements explicitly listed, but may include other elements not expressly listed or inherent to such product or apparatus.
The term "module" refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware or/and software code that is capable of performing the function associated with that element.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.
The foregoing description, for purposes of explanation, has been presented in conjunction with specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the embodiments to the precise forms disclosed above. Many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles and the practical application, to thereby enable others skilled in the art to best utilize the embodiments and various embodiments with various modifications as are suited to the particular use contemplated.
Fig. 1 is a schematic diagram of a message pushing process based on user behavior, which is provided in the present application, and the process includes the following steps:
s101: acquiring target identification information of a target user for message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information for pushing target messages to the target user based on the message pushing model; the message pushing model is used for extracting features of the target identification information to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as the target pushing feature information.
S102: pushing the target message to the target user according to the target pushing characteristic information.
The message pushing method based on the user behaviors is applied to electronic equipment, and the electronic equipment can be PC, tablet personal computer and other equipment, and can also be a server.
The electronic equipment acquires target identification information of a target user needing message pushing, wherein the target identification information comprises at least one of user ID information, user organization information and user geographic latitude information of the target user. The user ID information is, for example, information characterizing the user's identity, such as the user's name, job number, identification card number, cell phone number, etc. The user organization information is, for example, a work division or a work combination to which the user belongs. User organization information of the target user can be determined according to the user ID information and the preset inclusion relation between each organization and each user. The user geographic dimension information is, for example, spatial geographic location information in which the user is located. The corresponding relation between each user and the geographic dimension information can be stored in advance, and then the geographic dimension information of the user can be determined according to the user ID information.
And the electronic equipment stores a pre-trained message pushing model, the target identification information of the target user is input into the message pushing model, and the message pushing model is used for extracting the characteristics of the target identification information to obtain the identification characteristic information. If the target identification information comprises user ID information, user organization information and user geographic latitude information, the message pushing model respectively performs feature extraction on the user ID information, the user organization information and the user geographic latitude information to obtain sub-feature messages corresponding to the user ID information, sub-feature messages corresponding to the user organization information and sub-feature messages corresponding to the user geographic latitude information.
And determining push characteristic information matched with the target user according to the identification characteristic information. And determining the similarity of the identification feature information and each multidimensional feature point in the message push model, and determining push feature information matched with the target user according to the push feature information in the multidimensional feature point with the highest similarity. The multi-dimensional feature points comprise multi-dimensional feature information, and the similarity between the identification feature information and each multi-dimensional feature point in the message push model can be determined according to the identification feature information and the multi-dimensional feature information contained in each multi-dimensional feature point. And taking the pushing characteristic information matched with the target user as target pushing characteristic information, and pushing the target message to the target user according to the target pushing characteristic information. The push characteristic information comprises at least one of push characteristic information corresponding to push time information, push characteristic information corresponding to push channel information and push characteristic information corresponding to push terminal type information. And determining at least one of target push time information, target push channel information and target push terminal type information according to the target push characteristic information, and pushing the target message to a target user according to at least one of the target push time information, the target push channel information and the target push terminal type information. The target message refers to the message content of the user pushed by the service server in real time, and comprises information such as notification, waiting, alarm, announcement and the like.
In the application, after the electronic equipment acquires the target identification information of the target user of message pushing, the target identification information is input into the message pushing model. Extracting features of the target identification information based on the message pushing model to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as target pushing feature information; pushing the target message to the target user according to the target pushing characteristic information. The method and the device can determine the target pushing characteristic information according to the target identification information of the target user based on the message pushing model. Different new user identification information is different, different pushing characteristic information can be determined based on a message pushing model aiming at different new users, and message pushing is carried out according to the pushing characteristic information. Compared with the method for pushing the message by adopting a unified and unchanged mode for all new users, the self-adaption and the accuracy of the message pushing of the new users are improved.
Fig. 2 is a schematic diagram of a determining process of a message push model provided in the present application, including the following steps:
s201: receiving user behavior data reported by each sample user, wherein the user behavior data comprises multidimensional characteristic factors; the multidimensional feature factors comprise user ID information, user organization information, user geographic latitude information, push time information, push channel information and push terminal type information.
S202: and carrying out feature extraction and clustering processing on the user behavior data reported by each sample user based on a clustering algorithm to obtain a message pushing model of each class cluster containing multidimensional feature points.
When the message pushing model is trained, user behavior data reported by each sample user is received, and multidimensional feature factors carried in the user behavior data are obtained, namely user ID information, user organization information, user geographic latitude information, pushing time information, pushing channel information and pushing terminal type information carried in the user behavior data are obtained. And carrying out feature extraction on the multidimensional feature factor user ID information, the user organization information, the user geographic latitude information, the push time information, the push channel information and the push terminal type information based on a clustering algorithm to obtain feature information corresponding to the user ID information, the user organization information, the user geographic latitude information, the push time information, the push channel information and the push terminal type information. In the multidimensional space, user behavior data of each sample user is converted into a multidimensional feature point according to the feature information corresponding to each multidimensional feature factor. And clustering each multi-dimensional feature point based on a clustering algorithm to obtain a message pushing model of each class cluster containing the multi-dimensional feature points. The clustering algorithm may be a K-means weighted clustering algorithm. The K-Means weighted clustering algorithm is used for setting weight coefficients for different feature factors based on the K-Means algorithm and calculating Euclidean distances between feature points containing the feature factors.
Fig. 3 is a schematic process diagram of determining push feature information matched with a target user according to the present application, including the following steps:
s301: and respectively determining the first similarity between the identification characteristic information and each multidimensional characteristic point in the message pushing model.
S302: and selecting the multidimensional feature point with the highest first similarity with the identification feature information as a first target multidimensional feature point.
S303: and determining push characteristic information matched with the target user according to the push characteristic information in the first target characteristic point.
The process for respectively determining the first similarity between the identification characteristic information and each multidimensional characteristic point in the message pushing model specifically comprises the following steps: determining each piece of sub-feature information in the multi-dimensional feature points aiming at each multi-dimensional feature point, and calculating the distance between the identification feature information and the multi-dimensional feature point according to each piece of sub-feature information in the identification feature information and each piece of sub-feature information in the multi-dimensional feature point; determining a first similarity between the identification feature information and the multidimensional feature point according to the distance; wherein the smaller the distance, the higher the first similarity. Each piece of sub-feature information in the identification feature information comprises sub-feature information corresponding to user ID information, user organization information and user geographic latitude information; each piece of sub-feature information in the multi-dimensional feature point comprises sub-feature information corresponding to user ID information, user organization information, user geographic latitude information, push time information, push channel information and push terminal type information. Preferably, the distance between the identification feature information and the multi-dimensional feature point is determined according to each piece of sub-feature information in the identification feature information, each piece of sub-feature information in the multi-dimensional feature point and the weight value corresponding to each piece of sub-feature information. The weight value corresponding to each piece of sub-feature information can be set according to the priority of each piece of sub-feature information, for example, the weight value corresponding to the sub-feature information with higher priority is larger. Optionally, if the sub-feature information corresponding to the user ID information has the highest priority, the weight value corresponding to the sub-feature information corresponding to the user ID information is set to the largest weight value. In addition, the weight value corresponding to each piece of sub-feature information can be adjusted according to the determined maximum similarity, if the determined maximum similarity is smaller than the set similarity threshold, the defect in the weight value setting is indicated, and the weight value can be adjusted at the moment so that the determined maximum similarity is not smaller than the set similarity threshold.
And selecting the multidimensional feature point with the highest first similarity with the identification feature information as a first target multidimensional feature point. Preferably, the multidimensional feature point with the highest first similarity with the identification feature information and the highest first similarity larger than the set similarity threshold value is selected as the first target multidimensional feature point. If the highest similarity is not greater than the set similarity threshold, at this time, the weight value corresponding to each piece of sub-feature information can be adjusted to recalculate the first similarity, and when the first similarity is the highest and greater than the set similarity threshold, the multi-dimensional feature point with the highest similarity with the identification feature information is determined to be selected as the first target multi-dimensional feature point.
And finally, determining push characteristic information matched with the target user according to the push characteristic information in the first target multidimensional characteristic point. The push characteristic information comprises at least one of push time information, push channel information and push terminal type information. The push feature information in the first target multidimensional feature point can be used as push feature information matched with the target user. The push time information refers to time information of a push target message. The push channel information is, for example, mail channel information, short message channel information, etc. The push terminal type information is, for example, a mobile phone, a PC, a computer, etc.
Because the message push model contains a large number of multidimensional feature points, if the first similarity between the identification feature information and each multidimensional feature point in the message push model is determined, the process of determining the push feature information matched with the target user is low in efficiency and consumes a large amount of resources. In order to solve the above problem, fig. 4 is a schematic process of determining push feature information matched with a target user according to the present application, including the following steps:
s401: and aiming at each class cluster in the message pushing model, determining a central multidimensional feature point of the class cluster according to each multidimensional feature point in the class cluster.
S402: and respectively determining the second similarity between the identification characteristic information and each central multidimensional characteristic point, and selecting the central multidimensional characteristic point with the highest second similarity with the identification characteristic information as a candidate multidimensional characteristic point.
S403: and taking the multidimensional feature point closest to the candidate multidimensional feature point as a second target feature point, and determining push feature information matched with the target user according to the push feature information in the second target feature point.
In the application, aiming at each class cluster in the message push model, a central multidimensional feature point of the class cluster is determined according to each multidimensional feature point contained in the class cluster. The average value of the multi-dimensional characteristic information of each multi-dimensional characteristic point contained in the cluster is calculated, and the characteristic point corresponding to the obtained evaluation characteristic information of each dimension is taken as a central multi-dimensional characteristic point.
Determining each piece of sub-feature information in the central multi-dimensional feature point aiming at each central multi-dimensional feature point, and calculating the distance between the identification feature information and the central multi-dimensional feature point according to each piece of sub-feature information in the identification feature information and each piece of sub-feature information in the central multi-dimensional feature point; determining a second similarity of the identification feature information and the central multidimensional feature point according to the distance; wherein the smaller the distance, the higher the second similarity. Preferably, the distance between the identification feature information and the central multidimensional feature point is determined according to each piece of sub-feature information in the identification feature information, each piece of sub-feature information in the central multidimensional feature point and the weight value corresponding to each piece of sub-feature information. The weight value corresponding to each piece of sub-feature information can be set according to the priority of each piece of sub-feature information, for example, the weight value corresponding to the sub-feature information with higher priority is larger. Optionally, if the sub-feature information corresponding to the user ID information has the highest priority, the weight value corresponding to the sub-feature information corresponding to the user ID information is set to the largest weight value. In addition, the weight value corresponding to each piece of sub-feature information can be adjusted according to the determined maximum similarity, if the determined maximum similarity is smaller than the set similarity threshold, the defect in the weight value setting is indicated, and the weight value can be adjusted at the moment so that the determined maximum similarity is not smaller than the set similarity threshold.
And selecting the central multidimensional feature point with the highest second similarity with the identification feature information as a candidate multidimensional feature point. Preferably, a center multidimensional feature point with the highest second similarity with the identification feature information and the highest second similarity being larger than a set similarity threshold value is selected as the candidate multidimensional feature point. If the highest similarity is not greater than the set similarity threshold, at this time, the weight value corresponding to each piece of sub-feature information can be adjusted to recalculate the second similarity, and when the second similarity is the highest and greater than the set similarity threshold, the center multidimensional feature point with the highest similarity with the identification feature information is determined to be selected as the candidate multidimensional feature point.
And finally, determining the multidimensional feature point closest to the candidate multidimensional feature point as a second target feature point. And determining push characteristic information matched with the target user according to the push characteristic information in the second target characteristic point. The push feature information in the second target multidimensional feature point can be used as push feature information matched with the target user.
Fig. 5 is a schematic diagram of a process for determining similarity between identification feature information and each multidimensional feature point, where the process includes the following steps:
s501: and aiming at each multi-dimensional feature point, determining the distance between the identification feature information and the multi-dimensional feature point according to each piece of first sub-feature information in the identification feature information, each piece of second sub-feature information in the multi-dimensional feature point and the weight value corresponding to each piece of sub-feature information.
S502: determining the similarity between the identification feature information and each multidimensional feature point according to the distance between the identification feature information and each multidimensional feature point; wherein, the smaller the distance, the higher the similarity.
Each multidimensional feature point comprises each multidimensional feature point in the message push model or comprises each central multidimensional feature point. Each piece of sub-feature information in the identification feature information is referred to as each piece of first sub-feature information, and each piece of sub-feature information in the multi-dimensional feature point is referred to as each piece of second sub-feature information.
The obtaining the target identification information of the target user pushed by the message comprises the following steps: and acquiring at least one of user ID information, user organization information and user geographic latitude information of the target user pushed by the message.
Determining, based on the message pushing model, target pushing feature information for pushing a target message to the target user includes: and determining at least one of target pushing time information, target pushing channel information and target pushing terminal type information for pushing the target message to the target user based on the message pushing model.
The method and the device utilize user behavior operation events, based on multidimensional feature factors such as user ID information, user organization information, user geographic latitude information, push time information, push channel information, push terminal type information and the like, combine a K-Means weighted clustering algorithm to carry out vector quantization analysis on the multidimensional feature factors, set weights of the feature factors based on relevant priorities of users, and carry out message push model training. When the service system sends a user message notification, the service system calculates the distance by combining the current time information, the user ID information, the user geographic dimension information, the user organization information and other data, and selects the target push characteristic information such as push channel information, time information, push terminal type information and the like corresponding to the cluster with the closer distance. User configuration is not needed, and user information is sent according to the target push characteristic information.
According to the message pushing method, based on pushing time information and pushing channel information, multidimensional feature factors such as user geographic dimension information, pushing terminal type information, user organization information and user ID information and the like, machine learning model K-means weighted clustering training is utilized, and data information of a message pushing object user is combined, multi-target features such as a recommended channel, a time period and a terminal type are automatically searched for message notification and sending. The manual message receiving configuration page operation of a user is reduced, the user experience is improved, and the flexibility, the accuracy and the expansibility of sending different service types, different time periods and different message channels are improved. And after the service types are dynamically increased and the message channels are increased, the user does not need to configure pages, and the multi-target features such as the channels, the time points, the terminal types and the like are automatically searched according to the model training results. The user message informs the operation feedback that the feature factor weight can be optimized continuously, and the training model is optimized.
Considering that a plurality of service types exist in the current application system service, a notification channel needs to notify a user of real-time information, the user has various information notification receiving modes based on self preference or privacy convenience and the like, and meanwhile, the user needs to process notification information in time. After the information notification is received, the user operation behavior data is reported to the data analysis center, and the data analysis center carries out vectorization on multidimensional feature factors such as user ID information, user organization information, user geographic latitude information, push time information, push channel information, push terminal type information and the like, and carries out model training through feature factor weight clustering. When the message notifies the user object, the data feature factors contained in the target identification information such as the user ID information are combined to find out the multi-target feature message notification such as the target pushing time information, the target pushing channel information and the target pushing terminal type information of the target message pushed by the target user.
The data pushing system mainly comprises two processing units, namely a data pushing center and a data analysis center.
The processing procedure of the data pushing center is as follows:
after receiving the message notification, the user clicks on the message notification content. The user behavior data reporting data pushing center comprises multidimensional feature factors such as user ID information, user organization information, user geographic latitude information, pushing time information, pushing channel information, pushing terminal type information and the like.
And after receiving the user behavior data, sending the data analysis center to perform data model training.
After receiving the service message push notification, searching for multi-target characteristic message notifications such as recommended message push time information, push channel information, push terminal type information and the like from the data analysis center according to the characteristic factors such as user ID information, user organization information and the like.
After receiving the information notification, the user operates and processes the notification information, simultaneously provides user feedback no longer receives options, and finally reports user behavior data to the data pushing center.
The data analysis center processes as follows:
and receiving data pushing center user behavior data. And obtaining the configuration weight coefficient corresponding to the characteristic factor. And vectorizing the user behavior data by using a K-means algorithm, and carrying out model training by combining the characteristic factor weights. Selecting K points as initial centroids, calculating the distance from each sample to each centroid, simultaneously combining with characteristic factor weight coefficients, calculating Euclidean distance, dividing the samples into clusters corresponding to the centroids closest to the centroid, calculating the average value of all the samples in each cluster, updating the centroids of the clusters by using the average value, and repeating the calculation until the average value is smaller than a specified threshold value.
In which a is n And I is the weight coefficient of each characteristic factor.
When a message pushing center query request is received, multi-target features such as recommended pushing time information, pushing channel information, pushing terminal type information and the like are searched from the model according to data feature factors such as user ID and the like. When the user behavior data feedback is not received any more, the weight coefficient can be dynamically modified according to the current message, and model training is performed again.
The message pushing method based on the user behavior improves the accuracy of receiving the historical preference data message notification of the new user. Zero configuration of the user greatly improves the experience of the user system. The method and the device can recommend not only push channel information, but also multi-target characteristic sending information such as push time information, push terminal type information and the like. By configuring the weight coefficient and combining with user message processing feedback, the accurate receiving capability of the user can be improved.
Fig. 6 is a schematic structural diagram of a message pushing device based on user behavior provided in the present application, including:
the determining module 61 is configured to obtain target identification information of a target user for message pushing, input the target identification information into a message pushing model, and determine target pushing feature information of pushing a target message to the target user based on the message pushing model; the message pushing model is used for extracting characteristics of the target identification information to obtain identification characteristic information, determining pushing characteristic information matched with the target user according to the identification characteristic information, and taking the pushing characteristic information as the target pushing characteristic information;
and the pushing module 62 is configured to push the target message to the target user according to the target push feature information.
The determining module 61 is configured to determine first similarities between the identification feature information and each multidimensional feature point in the message pushing model, select a multidimensional feature point with the highest first similarity with the identification feature information as a first target multidimensional feature point, and determine pushing feature information matched with the target user according to the pushing feature information in the first target multidimensional feature point.
A determining module 61, configured to determine, for each class cluster in the message push model, a central multidimensional feature point of the class cluster according to each multidimensional feature point in the class cluster; respectively determining second similarity between the identification characteristic information and each central multidimensional characteristic point, and selecting the central multidimensional characteristic point with the highest second similarity with the identification characteristic information as a candidate multidimensional characteristic point; and taking the multidimensional feature point closest to the candidate multidimensional feature point as a second target feature point, and determining push feature information matched with the target user according to the push feature information in the second target feature point.
A determining module 61, configured to determine, for each multi-dimensional feature point, a distance between the identification feature information and the multi-dimensional feature point according to each piece of first sub-feature information in the identification feature information, each piece of second sub-feature information in the multi-dimensional feature point, and a weight value corresponding to each piece of sub-feature information; determining the similarity between the identification feature information and each multidimensional feature point according to the distance between the identification feature information and each multidimensional feature point; wherein, the smaller the distance, the higher the similarity.
The determining module 61 is configured to obtain at least one of user ID information, user organization information, and user geographical latitude information of a target user for message pushing.
A determining module 61, configured to determine at least one of target push time information, target push channel information and target push terminal type information for pushing a target message to the target user based on the message push model.
The apparatus further comprises:
the training module 63 is configured to receive user behavior data reported by each sample user, where the user behavior data includes multidimensional feature factors; the multidimensional feature factors comprise user ID information, user organization information, user geographic latitude information, push time information, push channel information and push terminal type information; and carrying out feature extraction and clustering processing on the user behavior data reported by each sample user based on a clustering algorithm to obtain a message pushing model of each class cluster containing multidimensional feature points.
The present application also provides an electronic device, as shown in fig. 7, including: the processor 71, the communication interface 72, the memory 73 and the communication bus 74, wherein the processor 71, the communication interface 72 and the memory 73 complete communication with each other through the communication bus 74;
the memory 73 has stored therein a computer program which, when executed by the processor 301, causes the processor 71 to perform any of the above method steps.
The communication bus mentioned above for the electronic devices may be a peripheral component interconnect standard (Peripheral Component Interconnect, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The communication bus may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, the figures are shown with only one bold line, but not with only one bus or one type of bus.
The communication interface 72 is used for communication between the above-described electronic device and other devices.
The Memory may include random access Memory (Random Access Memory, RAM) or may include Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the aforementioned processor.
The processor may be a general-purpose processor, including a central processing unit, a network processor (Network Processor, NP), etc.; but also digital signal processors (Digital Signal Processing, DSP), application specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
The present application also provides a computer-readable storage medium having stored thereon a computer program executable by an electronic device, which when run on the electronic device causes the electronic device to perform any of the above method steps.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the spirit or scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims and the equivalents thereof, the present application is intended to cover such modifications and variations.
Claims (10)
1. A message pushing method based on user behavior, the method comprising:
acquiring target identification information of a target user for message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information for pushing target messages to the target user based on the message pushing model; pushing the target message to the target user according to the target pushing characteristic information;
the message pushing model is used for extracting features of the target identification information to obtain identification feature information, determining pushing feature information matched with the target user according to the identification feature information, and taking the pushing feature information as the target pushing feature information.
2. The method of claim 1, wherein determining push feature information matching the target user based on the identifying feature information comprises:
and respectively determining the first similarity between the identification characteristic information and each multidimensional characteristic point in the message pushing model, selecting the multidimensional characteristic point with the highest similarity with the identification characteristic information as a first target multidimensional characteristic point, and determining pushing characteristic information matched with the target user according to the pushing characteristic information in the first target multidimensional characteristic point.
3. The method of claim 1, wherein determining push feature information matching the target user based on the identifying feature information comprises:
aiming at each class cluster in the message push model, determining a central multidimensional feature point of the class cluster according to each multidimensional feature point in the class cluster;
respectively determining second similarity between the identification characteristic information and each central multidimensional characteristic point, and selecting the central multidimensional characteristic point with the highest second similarity with the identification characteristic information as a candidate multidimensional characteristic point;
and taking the multidimensional feature point closest to the candidate multidimensional feature point as a second target feature point, and determining push feature information matched with the target user according to the push feature information in the second target feature point.
4. A method as claimed in claim 2 or 3, wherein the process of determining the similarity of the identifying feature information to each of the multi-dimensional feature points, respectively, comprises:
determining the distance between the identification feature information and the multi-dimensional feature points according to the first sub-feature information in the identification feature information, the second sub-feature information in the multi-dimensional feature points and the weight value corresponding to each sub-feature information;
determining the similarity between the identification feature information and each multidimensional feature point according to the distance between the identification feature information and each multidimensional feature point; wherein, the smaller the distance, the higher the similarity.
5. The method of claim 1, wherein the obtaining the target identification information of the target user of the message push comprises:
and acquiring at least one of user ID information, user organization information and user geographic latitude information of the target user pushed by the message.
6. The method of claim 1, wherein determining target push feature information for pushing a target message to the target user based on the message push model comprises:
and determining at least one of target pushing time information, target pushing channel information and target pushing terminal type information for pushing the target message to the target user based on the message pushing model.
7. The method of claim 1, wherein the determining of the message push model comprises:
receiving user behavior data reported by each sample user, wherein the user behavior data comprises multidimensional characteristic factors; the multidimensional feature factors comprise user ID information, user organization information, user geographic latitude information, push time information, push channel information and push terminal type information;
and carrying out feature extraction and clustering processing on the user behavior data reported by each sample user based on a clustering algorithm to obtain a message pushing model of each class cluster containing multidimensional feature points.
8. A message pushing device based on user behavior, the device comprising:
the determining module is used for acquiring target identification information of a target user of message pushing, inputting the target identification information into a message pushing model, and determining target pushing characteristic information of pushing target messages to the target user based on the message pushing model; the message pushing model is used for extracting characteristics of the target identification information to obtain identification characteristic information, determining pushing characteristic information matched with the target user according to the identification characteristic information, and taking the pushing characteristic information as the target pushing characteristic information;
and the pushing module is used for pushing the target message to the target user according to the target pushing characteristic information.
9. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;
a memory for storing a computer program;
a processor for implementing the method steps of any of claims 1-7 when executing a program stored on a memory.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein a computer program which, when executed by a processor, implements the method steps of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311600826.7A CN117640743A (en) | 2023-11-28 | 2023-11-28 | Message pushing method, device, equipment and medium based on user behavior |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311600826.7A CN117640743A (en) | 2023-11-28 | 2023-11-28 | Message pushing method, device, equipment and medium based on user behavior |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117640743A true CN117640743A (en) | 2024-03-01 |
Family
ID=90033268
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311600826.7A Pending CN117640743A (en) | 2023-11-28 | 2023-11-28 | Message pushing method, device, equipment and medium based on user behavior |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117640743A (en) |
-
2023
- 2023-11-28 CN CN202311600826.7A patent/CN117640743A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111226238B (en) | Prediction method, terminal and server | |
| CN109523237B (en) | Crowd-sourced task pushing method and related device based on user preference | |
| CN109816039B (en) | Cross-modal information retrieval method and device and storage medium | |
| CN106471851B (en) | Learning model based device positioning | |
| CN109344884B (en) | Media information classification method, method and device for training picture classification model | |
| CN107181776B (en) | Data processing method and related equipment and system | |
| CN105631707A (en) | Advertisement click rate estimation method based on decision tree, application recommendation method and device | |
| US20200219016A1 (en) | Method for Cleaning Up Background Application, Storage Medium, and Electronic Device | |
| CN107305611B (en) | Method and device for establishing model corresponding to malicious account and method and device for identifying malicious account | |
| CN108768695B (en) | KQI problem positioning method and device | |
| CN115017400B (en) | Application APP recommendation method and electronic equipment | |
| CN114372133A (en) | Query information question answering method, related device and computer program product | |
| US20230004776A1 (en) | Moderator for identifying deficient nodes in federated learning | |
| CN112883265A (en) | Information recommendation method and device, server and computer readable storage medium | |
| CN114140712A (en) | Automatic image recognition and distribution system and method | |
| CN107256231B (en) | Team member identification device, method and system | |
| CN117640743A (en) | Message pushing method, device, equipment and medium based on user behavior | |
| CN108810916B (en) | Wi-Fi hotspot recommendation method and device and storage medium | |
| CN113902132B (en) | Training method of negative feedback behavior prediction model, message pushing method and equipment | |
| CN111368858A (en) | User satisfaction evaluation method and device | |
| CN111814051B (en) | Resource type determining method and device | |
| CN114638308A (en) | Method and device for acquiring object relationship, electronic equipment and storage medium | |
| CN111612783B (en) | Data quality assessment method and system | |
| CN111898033A (en) | Content pushing method and device and electronic equipment | |
| KR20200036173A (en) | Method and apparatus for recommending user grouping and groupwise preferred machine learning analysis algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |