WO2020164269A1 - 基于用户群组的消息推送方法、装置及计算机设备 - Google Patents
基于用户群组的消息推送方法、装置及计算机设备 Download PDFInfo
- Publication number
- WO2020164269A1 WO2020164269A1 PCT/CN2019/117578 CN2019117578W WO2020164269A1 WO 2020164269 A1 WO2020164269 A1 WO 2020164269A1 CN 2019117578 W CN2019117578 W CN 2019117578W WO 2020164269 A1 WO2020164269 A1 WO 2020164269A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- information
- user
- behavior
- pushed
- grouping
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000003993 interaction Effects 0.000 claims abstract description 17
- 230000008569 process Effects 0.000 claims abstract description 15
- 230000006399 behavior Effects 0.000 claims description 180
- 239000000284 extract Substances 0.000 claims description 17
- 238000013145 classification model Methods 0.000 claims description 5
- 230000002452 interceptive effect Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 abstract description 6
- 238000006243 chemical reaction Methods 0.000 description 7
- 238000012797 qualification Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000012216 screening Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 208000028399 Critical Illness Diseases 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/02—Details
- H04L12/16—Arrangements for providing special services to substations
- H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
- H04L12/185—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast with management of multicast group membership
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/02—Details
- H04L12/16—Arrangements for providing special services to substations
- H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
- H04L12/1859—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast adapted to provide push services, e.g. data channels
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/55—Push-based network services
Definitions
- This application relates to the field of data processing technology. Specifically, this application relates to a message push method, device and computer equipment based on user groups.
- the existing technical solutions generally push full coverage information in batches to users who have mastered user information. For example, in some information push scenarios, it is necessary to obtain information push quotas from operators, and then push them to all users indiscriminately.
- obtaining the quota for pushing information from the operator requires a certain cost, and it also causes the user to be unable to correspond to the accurate information being pushed.
- Pushing information to users in batches in the prior art will result in low information push accuracy, low conversion rate, and poor effect, and further increase the cost of information push.
- the purpose of this application is to solve at least one of the technical defects mentioned above, especially the technical defects of low information push rate, low conversion rate, and poor push effect.
- This application provides a message push method based on user groups, including:
- the user behavior information is the behavior information in the user interaction process
- the user grouping information is the user grouping Information
- the information to be pushed in the corresponding classification in the classification information is matched, and the information to be pushed is pushed to the users in the user group.
- This application also provides a message grouping and pushing device based on user groups, including:
- the classification unit is used to classify the information to be pushed according to the content of the information to be pushed to obtain the classified information;
- the grouping unit is used to obtain user behavior information of users, and group the users into several user groups according to the user behavior information to obtain user grouping information; wherein, the user behavior information is behavior information in the user interaction process, and the user Grouping information is the information of user grouping;
- the pushing unit is configured to match the information to be pushed in the corresponding classification in the classification information according to the user grouping information, and push the information to be pushed to users in the user group.
- This application also provides an electronic device, which includes:
- a memory for storing processor executable instructions
- the processor is configured as the steps of the user group-based message pushing method described in any of the above embodiments.
- the present application also provides a non-transitory computer-readable storage medium, which is characterized in that, when the instructions in the storage medium are executed by the processor of the mobile terminal, the mobile terminal can execute any of the above-mentioned embodiments.
- the steps of the message push method based on user groups are described.
- the above-mentioned message push method based on user groups by dividing user groups for users, matching classification information according to the user group information corresponding to the user groups to obtain the information to be pushed, and finally pushing specific classifications to be pushed to users in the user group Information can improve the accuracy of the information to be pushed, thereby improving the efficiency of pushing the information to be pushed, and avoiding invalid push or negative push that causes reaction.
- FIG. 1 is a schematic diagram of an application scenario of a user group-based message pushing method in an embodiment
- FIG. 2 is a method flowchart of a message pushing method based on user groups in an embodiment
- Figure 3 is a flowchart of a method for obtaining classification information in an embodiment
- Fig. 5 is a flowchart of a method for obtaining and pushing corresponding classified information to be pushed in an embodiment
- Fig. 6 is a flowchart of a method for generating a push category model in an embodiment
- Figure 7 is a flowchart of a method for limiting the number of pushes in an embodiment
- FIG. 8 shows a schematic diagram of the device structure of the user group-based message pushing device in the embodiment.
- this embodiment provides a message push method based on user groups.
- the scene in FIG. 1 is used for exemplary display.
- Figure 1 shows an exemplary application scenario of the message push method based on user groups.
- the message push platform obtains the behavior information of user A through the network connection with the mobile phone 11, and the instructions of user A can also be fed back to the message push platform through the mobile phone 11 and via the network connection.
- the aforementioned mobile phone 11 may also be other terminals, such as electronic devices such as computers, tablets, and e-book readers.
- the message push platform also performs data transmission with the message database and user management system respectively.
- Figure 2 shows the method for pushing messages based on user groups in this embodiment, which includes the steps:
- Step S21 According to the content of the information to be pushed, the information to be pushed is classified to obtain the classified information.
- the classification information refers to the classification related to the content of the information to be pushed, for example, it can be classified as introduction information to be pushed, reminder information to be pushed, etc., and can also be classified according to business content, such as insured information to be pushed, deposits Categories such as information to be pushed and financial management information to be pushed.
- Step S22 Obtain user behavior information of the user, and group the users into several user groups according to the user behavior information to obtain user grouping information; wherein, the user behavior information is the behavior information in the user interaction process, and the user grouping information It is the information of user grouping.
- the user behavior information includes the user's click behavior, forwarding behavior, collection behavior, insurance purchase behavior, closing behavior, complaint behavior, and frequency and/or time node of the user within a specified time period.
- User grouping information is information that records the distribution of user groups. It can be information that a specific user belongs to a certain group, or information that a specific user is included in a certain group. In addition, the same user can also be in multiple groups.
- Step S23 Push the to-be-pushed information to the users in the user group according to the user grouping information matching the correspondingly classified information to be pushed in the classification information.
- the message push platform can obtain the information to be pushed through the message database.
- the classification of the information to be pushed can be performed through the message database, and at this time, the message pushing platform can directly obtain the classification information of the information to be pushed.
- the message push platform can obtain the information to be pushed in the message database and then classify it.
- the message push platform can obtain user behavior information of users through the user management system, and then group users into several user groups based on the user behavior information, and obtain user grouping information.
- the user management system can group the logged-in users in advance to obtain corresponding updated user groups and corresponding user grouping information.
- the user management system completes the division of corresponding user groups during user registration or user maintenance, and obtains user group information.
- the message push platform matches the corresponding classification information to be pushed according to the classification information. Different user groups match corresponding classification information according to corresponding user group characteristics. The message push platform then obtains the message to be pushed corresponding to the classification information from the message database according to the corresponding classification information. Finally, the message push platform pushes the messages to be pushed according to the classification information corresponding to different user groups.
- the above-mentioned technical solution of the message push method based on user groups is pushed through the user interaction behavior characteristics of the user behavior information and the association corresponding to the classification of the information to be pushed, which can improve the push accuracy of the message to be pushed, so that the message to be pushed can be pushed It is more targeted at the user behavior characteristics of users, has a higher conversion rate, and improves the efficiency of push.
- Step S22 obtains user behavior information of the user, and groups the users into several user groups according to the user behavior information.
- the method further includes :
- Step B1 Collect at least one user behavior of click behavior, forwarding behavior, collection behavior, insurance purchase behavior, closing behavior, and complaint behavior of each user within a specified time period.
- Step B2 Count the occurrence frequency of the user behavior within the specified time period and/or the time node corresponding to the user behavior to generate behavior time parameters.
- Step B3 Generate user behavior information according to the user behavior and behavior time parameters.
- step B1 can collect the user behavior of the user in a specific time period (for example, a natural day) through the mobile phone 11.
- User behaviors may include click behavior, forwarding behavior, collection behavior, insurance purchase behavior, closing behavior, and complaint behavior.
- the user A can use a handheld device such as the mobile phone 11 or a vertical counter to send an operation instruction by clicking a designated location or by clicking.
- user A can also use other operations such as sliding and heavy pressing as one of user A's operation behaviors.
- user A can confirm that user A has made the forwarding behavior by clicking a button, saving a forwarding link, and saving a picture with a specific pointing link.
- the user A closes a specific display page on the mobile phone 11, he can confirm that the behavior of the user A at this time is collected as the closing behavior.
- the user's closing behavior is recorded.
- the closing time can be the average reading time of the corresponding page or the preset reading time calculated according to the word count and reading speed.
- the normal closing behavior of the user can be excluded, and the abnormal closing behavior of the user can be collected.
- the mobile phone 11 may also continue to collect time information such as the corresponding time node of the user behavior and the frequency of occurrence in a specific time period (for example, the above-mentioned natural day) to generate time parameters.
- time information such as the corresponding time node of the user behavior and the frequency of occurrence in a specific time period (for example, the above-mentioned natural day) to generate time parameters.
- time information such as the corresponding time node of the user behavior and the frequency of occurrence in a specific time period (for example, the above-mentioned natural day) to generate time parameters.
- time information such as the corresponding time node of the user behavior and the frequency of occurrence in a specific time period (for example, the above-mentioned natural day) to generate time parameters.
- time information such as the corresponding time node of the user behavior and the frequency of occurrence in a specific time period (for example, the above-mentioned natural day) to generate time parameters.
- time information such as the corresponding time node
- the time parameter can be quantified as the most concentrated time period between 8:00 am and 9:00 am and between 7:00 pm and 8:00 pm.
- the frequency is 11 times/day.
- the above-mentioned time parameters can be expressed in other forms, such as coordinates, vectors, or time parameters obtained by quantization. Generate user behavior information based on the user behavior and behavior time parameters collected above.
- Step S21 classifies the to-be-pushed information according to the content of the to-be-pushed information to obtain the classified information, including:
- Step S31 Extract corresponding keywords from the content of the information to be pushed.
- Step S32 According to the keywords, the classification information is obtained by matching the classification in the classification list of the information to be pushed.
- the message push platform obtains the information to be pushed from the message database, and extracts keyword information correspondingly according to the content of the information to be pushed.
- the message push platform obtains the classification list, and uses the classification of the keyword on the classification list or the classification similar to the classification on the classification list as the classification information of the information to be pushed. For example, the keywords "balance, credit" extracted from the information to be pushed can be matched on the classification list to obtain the corresponding classification as reminder information to be pushed.
- the keywords "insurance, critical illness, insured amount” extracted from the information to be pushed are not matched on the classification list to obtain the corresponding classification.
- neural convolutional network operations can be performed on the keywords to obtain similar keywords, and then they can be matched on the classification list.
- the classification list is generated, the words or thesaurus related to the classification can be associated with the classification list.
- the classification information can also be matched by correlation calculation. For example, the correlation between "insurance” in the above classification list and the keyword "insurance, critical illness, insurance amount” is relatively high, and the classification information is "insurance". The error probability is low, and “insurance” can be used as the classification information of the information to be pushed.
- the classification list can also be updated within a certain period according to the content of the information to be pushed, so as to maintain the accuracy and timeliness of the classification list.
- step S22 to obtain user behavior information of users, and group users into groups according to user behavior information
- the steps to obtain user grouping information include:
- Step S41 Obtain user behavior information according to the user's interactive behavior.
- Step S42 Input user behavior information into a pre-trained user classification model to group users into several user groups.
- Step S43 Obtain user grouping information according to the user groups divided by the users.
- the message push platform can obtain the current interaction behavior of user A through the mobile phone 11 for processing.
- the message push platform generates user behavior information according to the type and frequency of interaction behavior.
- Input user behavior information into a pre-trained user classification model in the message push platform, and group users into user groups.
- the click-through rate as an example of user behavior information, if the user has a click-through rate of 90% for reminder messages and 30% for promotional messages, then the user is classified into a reminder user group; if The user has a click-through rate of 85% for reminder messages and 88% for promotion-type messages, then the user is divided into a full message user group.
- user groups can also be divided by parameters such as forwarding rate, reading time, secondary click rate, mobile client usage rate, and promotion conversion rate in the user behavior information.
- the aforementioned parameters may be combined with a certain proportional coefficient to form composite user behavior information.
- users in the above-mentioned user group all have similar interactive behaviors.
- users in the all-message user group have a feedback rate of 80% for all types of messages to be pushed; users in the reminder user group have a feedback rate of 85% for messages to be pushed in the reminder category.
- the feedback rate of messages to be pushed is 10%; the feedback rate of users in the promotion user group to the messages to be pushed in the promotion category is 75%, and the feedback rate of messages to be pushed in the reminder category is 30%.
- Step S23 matches the information to be pushed in the corresponding classification in the classification information according to the user grouping information, and pushes the information to be pushed
- the steps to users in the user group include:
- Step S51 Input the user grouping information into the push category model matching category information.
- Step S52 Obtain corresponding information to be pushed according to the classification information.
- Step S53 Push the information to be pushed to users in the user group according to the user grouping information.
- the message push platform inputs the user grouping information into the push category model in the message database, and obtains the classification information corresponding to the user grouping information by matching the push category model in the message database. According to the obtained classification information, the message push platform obtains the information to be pushed corresponding to the classification information from the message database. For example, for a reminder user group in the user grouping information, the message push platform inputs the user grouping information of the reminder user group into the push type model in the message database, and the corresponding classification is reminder information. Accordingly, the message push platform obtains the information to be pushed corresponding to the reminder information from the message database, such as balance status reminder information, annual fee payment reminder information, consumption record reminder information, and so on.
- the message push platform pushes the reminder information obtained from the above matching to users in the user group (for example, user A).
- the message push platform uses the above reminder information as a network connection.
- the information to be pushed is pushed to the mobile phone 11 of user A, and the mobile phone 11 displays or reminds the user A.
- the message push platform or message database undergoes effective screening in advance for different information to be pushed.
- Validity screening can be time condition, qualification condition and other screening conditions. Taking time conditions as an example, according to the comparison between the last effective time recorded in the information to be pushed and the current time, confirm that the current time is before the last effective time, otherwise the corresponding information to be pushed is expired information. For expired information, it is not necessary to participate in the message database or The classification information matching and message push steps of the message push platform. Take the qualification conditions as an example.
- the users who meet the qualification conditions are further screened in the user group to be pushed, and then the information to be pushed is pushed to the users who meet the qualification conditions to avoid disqualification.
- accuracy issues such as information to be pushed.
- the push category model in this embodiment can also be embedded in the message push platform to improve the speed at which the message push platform classifies the information to be pushed, which is beneficial to improve efficiency.
- Step S51 Input the user group information into the push category model matching classification information. Before, it also included:
- Step S61 Extract user characteristic information of each user group in the user grouping information.
- Step S62 Obtain the historical push information of each user group corresponding to the user grouping information.
- Step S63 calculate the feedback value according to the historical behavior information corresponding to the historical push information.
- Step S64 Determine whether the feedback value of the historical push information is greater than the feedback threshold. If yes, go to step S65.
- the feedback value is the ratio of the feedback behavior of each user group to the historical push information, for example, the total number of clicks on the historical push information of the users in the user group, the total browsing time and other user behavior data that have feedback on the historical push information.
- the above feedback threshold is a threshold for comparing the feedback value in the same type of user behavior data. For example, when the total number of clicks is selected as the feedback value, the feedback threshold is correspondingly the number of clicks threshold.
- Step S65 Extract historical push feature information from historical push information.
- Step S66 Generate a push category model according to the relationship between user characteristic information and historical push characteristic information.
- the above steps of generating the push category model can be input into the foregoing two after the message push platform, message database or external training is generated.
- the push category model is generated on the message push platform.
- the message push platform can extract user characteristic information of each user group in the user grouping information. These characteristics can be the user’s click-through rate or average browsing time of the information, request for credit card application, and submit Characteristic information such as consumer credit requests can also be characteristic information of the user's occupation, deposit status, loan status, insurance status, and other status.
- the message push platform obtains the historical push information of each user group corresponding to the user grouping information.
- the message push platform judges whether the feedback value of the multiple pieces of historical push information is greater than the feedback threshold. If yes, perform step S65, and the message push platform extracts historical push feature information corresponding to historical push information whose feedback value is greater than the feedback threshold.
- the historical push feature information may be the keyword, classification and other characteristic information of the historical push information.
- the message push platform may generate a push category model according to the mapping relationship between the extracted user characteristic information and the historical push characteristic information.
- step S21 before the step of classifying the information to be pushed according to the content of the information to be pushed to obtain classified information, the method further includes:
- Step S71 Extract the number threshold in the user behavior information.
- the number threshold can be understood as the upper limit of the number of messages that the user can accept to push, which can include the number of messages or the length of time for reading the messages.
- Step S72 Obtain the number of pushed information to be pushed to the user within a specified time period.
- Step S73 Determine whether the pushed quantity is less than the quantity threshold. If yes, go to step S74; if not, go to step S75.
- Step S74 Classify the information to be pushed according to the content of the information to be pushed to obtain the classified information.
- Step S75 Do not push the information to be pushed.
- the message push platform can obtain the number of pushed information to be pushed to user A within a specified time period corresponding to the number threshold. For example, if the period of the quantity threshold is a single day, the specified time period can be set as a single day or a single day average value within a period of time.
- the message push platform determines whether the number of pushes is less than the number threshold. For example, the number of messages to be pushed that the message push platform has pushed on the day is 8, which is less than 10 of the above-mentioned number threshold, and the push can continue at this time. Then, the message pushing platform may perform step S74 to classify the information to be pushed according to the content of the information to be pushed to obtain the classified information or perform step S21.
- the message push platform does not perform push, and the message push platform executes step S75. It can also be understood that the message push platform does not execute step S21 at this time.
- the push information restriction formed by the number threshold obtained by extracting the user behavior information can limit the push amount of the information to be pushed on the one hand, avoid causing the user's negative interaction behavior, and have the opposite effect.
- step S21 it is judged in advance whether the total amount of information to be pushed exceeds the quantity threshold, which can also reduce the amount of data processed by the message push platform, improve the processing speed and efficiency of the message push platform, and facilitate the rapid push of the information to be pushed .
- This application also provides a user group-based message grouping and pushing device corresponding to the above-mentioned user group-based message grouping and pushing method, including:
- the classification unit 81 is configured to classify the information to be pushed according to the content of the information to be pushed to obtain classified information;
- the pushing unit 83 is configured to match the information to be pushed in the corresponding classification in the classification information according to the user grouping information, and push the information to be pushed to users in the user group.
- a memory for storing processor executable instructions
- the processor is configured as the steps of the user group-based message pushing method described in any of the above embodiments.
- the embodiments of the present application also provide a non-transitory computer-readable storage medium, characterized in that, when the instructions in the storage medium are executed by the processor of the mobile terminal, the mobile terminal can perform any of the above-mentioned implementations.
- the example describes the steps of the message push method based on user groups.
- the above-mentioned message push method based on user groups obtains classification information by classifying the push information according to the content of the push information; obtains the user behavior information of the user, and divides the user into several user groups according to the user behavior information to obtain the user grouping Information; wherein the user behavior information is behavior information in the user interaction process, the user grouping information is user grouping information; according to the user grouping information matching the corresponding classification information to be pushed in the classification information, the information to be pushed is pushed to The technical solution of the users in the user group, the user group is divided for the users, the information to be pushed is obtained by matching the classification information according to the corresponding user group information of the user group, and finally the specific classification of the information to be pushed is pushed to the users in the user group , Which can improve the accuracy of pushing information to be pushed, thereby improving the pushing efficiency of the information to be pushed, and avoiding invalid pushing or negative pushing that causes a reaction.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
Abstract
本申请涉及数据处理技术领域，提供一种基于用户群组的消息推送方法，包括：根据待推送信息的内容对待推送信息进行分类得到分类信息；获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。本申请还相应提供一种基于用户群组的消息推送装置和计算机设备。上述基于用户群组的消息推送方法、装置及计算机设备，能提升待推送信息的推送准确性，进而提升待推送信息的推送效率，避免无效推送或者引起反作用的消极推送。
Description
本申请要求于2019年02月15日提交中国专利局、申请号为201910117573.5、申请名称为“基于用户群组的消息推送方法、装置及计算机设备”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
本申请涉及数据处理技术领域，具体而言，本申请涉及一种基于用户群组的消息推送方法、装置及计算机设备。
随着信息时代的信息数量的增多，在当前的场景中对用户进行一定数量的信息推送。
现有的技术方案一般是批量对已经掌握用户资料的用户进行全覆盖的信息推送。例如，在一些信息推送的场景中，需要向运营商获取信息推送的配额，再向用户全体进行无差别推送。
在上述例子中，向运营商获取信息推送的配额需要一定成本，而且还会导致用户无法对应被推送到准确的信息。现有技术中批量对用户推送信息将导致信息推送准确率低、转化率低、效果差，还进一步提升了信息推送的成本。
发明内容
本申请的目的旨在至少能解决上述的技术缺陷之一，特别是信息推送准确率低、转化率低、推送效果差的技术缺陷。
本申请提供一种基于用户群组的消息推送方法，包括：
根据待推送信息的内容对待推送信息进行分类得到分类信息；
获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；
根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
本申请还提供一种基于用户群组的消息分群推送装置，包括：
分类单元，用于根据待推送信息的内容对待推送信息进行分类得到分类信息；
分群单元，用于获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；
推送单元，用于根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
本申请还提供一种电子设备，所述电子设备包括：
处理器；
用于存储处理器可执行指令的存储器；
其中，所述处理器被配置为如上述任一项实施例所述基于用户群组的消息推送方法的步骤。
本申请的还提供一种非临时性计算机可读存储介质，其特征在于，当所述存储介质中的指令由移动终端的处理器执行时，使得移动终端能够执行如上述任一项实施例所述基于用户群组的消息推送方法的步骤。
上述基于用户群组的消息推送方法，通过针对用户进行划分用户群组，根据用户群组对应用户分群信息匹配分类信息得到待推送信息，最终向用户群组中的用户推送特定的分类的待推送信息，能够提升待推送信息的推送准确性，进而提升待推送信息的推送效率，避免无效推送或者引起反作用的消极推送。
本申请上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1为实施例中基于用户群组的消息推送方法的应用场景示意图；
图2为实施例中基于用户群组的消息推送方法的方法流程图；
图3为实施例中得到分类信息的方法流程图；
图4为本实施例的到用户分群信息的方法流程图；
图5为实施例的获取对应分类待推送信息并推送的方法流程图；
图6为实施例中生成推送类别模型的方法流程图；
图7为实施例中限制推送数量的方法流程图；
图8示出实施例中基于用户群组的消息推送装置的装置结构示意图。
下面详细描述本申请的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，仅用于解释本申请，而不能解释为对本申请的限制。
为了克服信息推送中准确率低、转化率低、推送效果差的技术缺陷，本实施例提供一种基于用户群组的消息推送方法。在本实施例中，以图1中的场景进行示例性的展示。
图1展示了基于用户群组的消息推送方法的一种示例性应用场景。其中，消息推送平台通过与手机11之间的网络连接获取A用户的行为信息，A用户的指令也可以通过手机11并借助网络连接反馈到消息推送平台。前述的手机11还可以是其他终端，例如是电脑、平板、电子书阅读器等电子设备。消息推送平台还分别与消息数据库和用户管理系统进行数据传输。
图2中展示了本实施例的基于用户群组的消息推送方法，包括步骤：
步骤S21：根据待推送信息的内容对待推送信息进行分类得到分类信息。
其中，分类信息是指待推送信息的内容相关的分类，例如可以是介绍类待推送信息、提醒类待推送信息等分类，还可以是根据业务内容进行分类，例如投保类待推送信息、存款类待推送信息、理财类待推送信息等分类。
步骤S22：获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息。
在上述步骤中用户行为信息包括用户在指定时间段内的点击行为、转发行为、收藏行为、投保行为、购买行为、关闭行为、投诉行为及其发生频次和/或时间节点。用户分群信息是记录用户群组的分布情况的信息，可以是具体用户属于某个群组的信息，也可以是某群组中包括具体用户的信息。此外，相同 的用户还可以在多个群组中。
步骤S23：根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
在上述实现的过程中，消息推送平台可以通过消息数据库获取待推送信息。待推送信息的分类可以通过消息数据库进行，此时消息推送平台可以直接获取待推送信息的分类信息。当然，消息推送平台可以获取消息数据库中的待推送信息再进行分类。
消息推送平台可以通过用户管理系统获取用户的用户行为信息，再根据用户行为信息将用户分群为若干用户群组并得到用户分群信息。此外，用户管理系统可以对登入的用户预先进行分群，得到相应更新的用户群组和对应的用户分群信息。用户管理系统在用户注册或者在用户维护的过程中完成相应用户群组的划分，得到用户分群信息。
消息推送平台根据用户分群信息中的不同用户群组，分别根据分类信息匹配对应分类的待推送信息。不同的用户群组根据相应的用户群组特征匹配相应的分类信息。消息推送平台再根据相应的分类信息从消息数据库中获取对应分类信息的待推送消息。最后，消息推送平台依据不同用户群组对应的分类信息对应推送待推送消息。
上述基于用户群组的消息推送方法的技术方案，通过用户行为信息的用户交互行为特征以及对应待推送信息的分类的关联进行推送，能够提升待推送消息的推送准确性，使待推送消息得推送更针对用户的用户行为特征，具有较高的转化率，提升了推送的效率。
本申请实施例中还提供一种生成用户行为信息的技术方案，步骤S22获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息的步骤之前，还包括：
步骤B1：采集各个用户在指定时间段内的点击行为、转发行为、收藏行为、投保行为、购买行为、关闭行为、投诉行为的至少一种用户行为。
步骤B2：统计在所述指定时间段内所述用户行为的发生频次和/或对应用户行为的时间节点生成行为时间参数。
步骤B3：根据所述用户行为和行为时间参数生成用户行为信息。
在上述过程中，步骤B1可以通过手机11采集用户在特定时间段(例如是一个自然日)内的用户行为。用户行为可以包括点击行为、转发行为、收藏行为、投保行为、购买行为、关闭行为、投诉行为。
其中，以点击行为为例，用户A可以借助手机11等手持设备或者立式柜台，通过点击指定的位置或者通过点击的方式发出操作指令。除此之外，用户A还可以采用滑动、重压等其他操作，作为用户A的操作行为之一。继续以转发行为为例，用户A可以通过点击按钮、保存转发链接、保存带有特定指向链接的图片时，可以确认用户A作出转发行为。再举一个例子，用户A通过关闭手机11上特定的展示页面时，可以确认采集此时用户A的行为作为关闭行为。更进一步，若用户在阅读该页面时的阅读时间小于设定的关闭时间时，记录该用户的关闭行为。关闭时间可以采用对应页面的平均阅读时长或者根据字数和阅读速度计算得到的预设阅读时长。通过阅读时长的参数，可以排除用户正常的关闭行为，采集得到用户不正常的关闭行为。当然，在另一些实施例中，还可以通过用户A停留的阅读位置判断用户是否属于正常的关闭行为。通过上述对关闭行为的采集，可以采集得到准确的用户行为，避免不准确的用户行为对后续数据处理产生影响。
在本实施例中，步骤B2中手机11还可以继续采集上述用户行为的对应发生的时间节点、在特定时间段(例如是上述的一个自然日)内发生的频次等时间信息，生成时间参数。例如，在一个自然日内，手机11采集到用户A在上午八点至九点共发生五次转发行为，在下午七点至八点之间共发生六次转发行为，其他时间没有采集到转发行为。此时根据前述的转发行为发生频次和时间节点，可以得到相应的时间参数，时间参数可以量化为转发行为在上午八点至九点和下午七点至八点之间的时间段最为集中，发生频次为11次/日。当然，上述的时间参数可以表述为其他的形式，例如坐标、向量或者量化得到的时间参数。根据上述采集得到的用户行为和行为时间参数生成用户行为信息。
为了更好地实现分类，本实施例中提供一种技术方案，请参考图3，步骤S21根据待推送信息的内容对待推送信息进行分类得到分类信息的步骤，包括：
步骤S31：从所述待推送信息的内容中提取相应的关键词。
步骤S32：根据所述关键词在待推送信息的分类列表中匹配分类得到分类 信息。
在上述步骤中，消息推送平台从消息数据库中获取待推送信息，根据待推送信息的内容，相应提取关键词信息。消息推送平台获取分类列表，并根据所述关键词在分类列表上的分类或者与分类列表上分类相近的分类，作为待推送信息的分类信息。例如，在待推送信息中提取得到关键词“余额、入账”在分类列表上可以匹配得到相应的分类为提醒类待推送信息。
再例如，在待推送信息中提取得到关键词“投保、重疾、保额”在分类列表上未匹配得到相应的分类。此时可以对关键词进行神经卷积网络运算得到相近的关键词，再在分类列表上匹配。此外，还可以在生成分类列表时，就将与分类相关的词或词库关联到分类列表中。再者，还可以通过相关度运算匹配分类信息，例如上述分类列表中的“保险”与关键词“投保、重疾、保额”的相关度较高，此时得到分类信息为“保险”的错误概率较低，可以将“保险”作为上述待推送信息的分类信息。分类列表还可以在一定的周期内，根据待推送信息的内容进行分类的更新，以保持分类列表的准确性和及时性。
为了更准确对用户划分群组，得到更准确的用户分群信息，本实施例中还提供一种技术方案，请参考图4，步骤S22获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息的步骤，包括：
步骤S41：根据用户的交互行为得到用户行为信息。
步骤S42：将用户行为信息输入预先经过训练的用户分类模型中将用户分群为若干用户群组。
步骤S43：根据用户所划分的用户群组得到用户分群信息。
在上述过程中，请结合图1的应用场景，消息推送平台可以通过手机11获取A用户当前的交互行为进行处理。消息推送平台根据交互行为的类型、发生频率生成用户行为信息。将用户行为信息输入消息推送平台中预先经过训练的用户分类模型，将用户分群为用户群组。以点击率作为用户行为信息的例子，若用户对提醒类的消息的点击率为90％，对推广类的消息的点击率为30％，那么将该用户划分到提醒类用户群组中；若用户对提醒类的消息的点击率为85％，对推广类的消息的点击率为88％，那么将该用户划分到全消息用户群组中。当然，除了点击率之外，还可以通过用户行为信息中的转发率、阅读时长、 二次点击率、手机客户端使用率、推广转化率等等参数进行用户群组的划分。更进一步，为了更精确地反映用户交互行为的特征，可以将前述参数以一定的比例系数组合在一起，形成复合的用户行为信息。
通过上述用户行为信息划分用户而得到的用户群组中，在后续待推送信息的推送后，上述用户群组内的用户都具有相似的交互行为。例如，全消息用户群组中的用户对全部类型的待推送消息的反馈率为80％；提醒类用户群组中的用户对提醒类的待推送消息的反馈率为85％，对推广类的待推送消息的反馈率为10％；推广类用户群组中的用户对推广类的待推送消息的反馈率为75％，对提醒类的待推送消息的反馈率为30％。
为了进一步提升待推送信息的准确性，本申请的实施例中还提供一种技术方案，请参考图5，步骤S23根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户的步骤，包括：
步骤S51：将所述用户分群信息输入推送类别模型匹配分类信息。
步骤S52：根据分类信息获取对应的待推送信息。
步骤S53：根据用户分群信息将待推送信息推送至用户群组内的用户。
在上述过程中，请结合图1的应用场景，消息推送平台将所述用户分群信息输入消息数据库中的推送类别模型，通过消息数据库中的推送类别模型匹配得到用户分群信息对应的分类信息。根据得到的分类信息，消息推送平台从消息数据库中获取对应分类信息的待推送信息。例如，对用户分群信息中提醒类用户群组，消息推送平台将提醒类用户群组的用户分群信息输入上述消息数据库中的推送类别模型，得到对应的分类是提醒类信息。据此，消息推送平台从消息数据库中获取对应提醒类信息的待推送信息，例如是余额状态提醒信息、年费缴纳提醒信息、消费记录提醒信息等。
消息推送平台根据用户分群信息所记载的用户群组信息，将上述匹配得到的提醒类信息推送至用户群组内的用户(例如是A用户)，消息推送平台通过网络连接将上述提醒类信息作为待推送信息推送到A用户的手机11上，由手机11向A用户进行展示或提醒。
更进一步，消息推送平台或消息数据库针对不同的待推送信息预先经过有 效性筛选。有效性筛选可以时间条件、资格条件等筛选条件。以时间条件为例，根据待推送信息中记载的最后有效时间和当前时间的比较，确认当前时间在最后有效时间之前，否则对应的待推送信息为过期信息，对于过期信息可以不参与消息数据库或者消息推送平台的分类信息匹配、消息推送等步骤。以资格条件为例，根据待推送信息对应目标用户的资格条件，在将要推送的用户群组中进一步筛选符合资格条件的用户，再向符合资格条件的用户推送待推送信息，避免不符合资格条件的用户接收到待推送信息等准确性问题。上述有效性筛选的技术方案，一方面提升了消息推送的有效性，另一方面有效降低无效信息对处理资源的占用，进一步提升消息推送的效率和速度。
除了上述应用场景的交互过程，本实施例中的推送类别模型也可以内嵌到消息推送平台中，提升消息推送平台对待推送信息进行分类的速度，有利于提升效率。
为了更进一步提升待推送信息的准确性，本申请的实施例中还提供一种生成推送类别模型技术方案，请参考图6，步骤S51将所述用户分群信息输入推送类别模型匹配分类信息的步骤之前，还包括：
步骤S61：提取用户分群信息中各用户群组的用户特征信息。
步骤S62：获取用户分群信息对应各用户群组的历史推送信息。
步骤S63：根据历史推送信息对应的历史行为信息计算得到反馈值。
步骤S64：判断历史推送信息的反馈值是否大于反馈阈值。若是，执行步骤S65。
其中，反馈值是各用户群组对历史推送信息的反馈行为的比率，例如用户群组中用户对历史推送信息的总点击次数、总浏览时长等对历史推送信息有反馈的用户行为数据。上述反馈阈值是针对反馈值在相同类型的用户行为数据进行比较的阈值。例如，当反馈值选用总点击次数时，那么反馈阈值相应是点击次数阈值。
步骤S65：从历史推送信息中提取得到历史推送特征信息。
步骤S66：根据用户特征信息和历史推送特征信息的关系生成推送类别模型。
上述生成推送类别模型的步骤，可以在消息推送平台、消息数据库或者在 外部训练生成后再输入前述两者中。在本实施例中，例如推送类别模型在消息推送平台生成。从用户管理系统中获取用户分群信息后，消息推送平台可以提取用户分群信息中各用户群组的用户特征信息，这些特征可以是用户对信息的点击率或者平均浏览时长、提出信用卡申请请求、提出消费信贷的请求等特征信息，还可以是用户的职业、存款状态、贷款状态、保险状态等状态的特征信息。从消息数据库中获取一段时间(例如是一个自然月)内对用户分群信息中各个用户群组推送的历史推送信息，消息推送平台获取用户分群信息对应各用户群组的历史推送信息。消息推送平台判断上述多条历史推送信息的反馈值是否大于反馈阈值。若是，执行步骤S65，消息推送平台提取反馈值是否大于反馈阈值的历史推送信息对应的历史推送特征信息。其中，历史推送特征信息可以是历史推送信息的关键词、分类等特征信息。其后，消息推送平台可以根据上述提取的用户特征信息和历史推送特征信息的映射关系，生成推送类别模型。
为了控制待推送信息的推送数量，请参考图7本申请所提供一种技术方案，步骤S21根据待推送信息的内容对待推送信息进行分类得到分类信息的步骤之前，还包括：
步骤S71：提取用户行为信息中的数量阈值。
其中，数量阈值可以理解为是用户可接受推送信息的数量上限，可以包括信息数量或者信息阅读时长等。
步骤S72：获取指定时间段内对用户推送待推送信息的已推送数量。
步骤S73：判断已推送数量是否小于数量阈值。若是，执行步骤S74；若否，执行步骤S75。
步骤S74：根据待推送信息的内容对待推送信息进行分类得到分类信息。
步骤S75：不推送待推送信息。
在上述过程中，请结合图1的应用场景，在步骤S21之前，消息推送平台可以根据用户行为信息提取得到数量阈值。用户行为信息可以是消极的用户交互行为，例如退出程序、小于指定浏览时间关闭页面、投诉请求关闭推送权限等交互行为。以具体的场景为例，当A用户单日接收的待推送信息达到10条时或者待推送信息的阅读总时长达到30min时，A用户退出程序，此时可以通过用户的退出行为，提取A用户的数量阈值可以是待推送信息达到10条或者 阅读总时长达到30min。
消息推送平台可以获取对应数量阈值的指定时间段内，对A用户推送待推送信息的已推送数量。例如，数量阈值的周期是单日，那么指定时间段相应可以设定为单日或者一段时间内的单日平均值。消息推送平台判断已推送数量是否小于数量阈值。例如，在当日消息推送平台已经推送的待推送信息的数量为8条，小于上述数量阈值的10条，此时可以继续推送。那么，消息推送平台可以执行步骤S74根据待推送信息的内容对待推送信息进行分类得到分类信息或者执行步骤S21。如果在当日消息推送平台已经推送的待推送信息的数量为11条，大于上述数量阈值的10条，此时继续推送会引发用户的消极交互行为。因此，消息推送平台不进行推送，消息推送平台执行步骤S75，也可以理解为此时消息推送平台不执行步骤S21。上述技术方案中，通过用户行为信息提取得到的数量阈值形成的推送信息限制，一方面能够限制待推送信息的推送量，避免引发用户的消极交互行为，起到相反的作用。另一方面，在步骤S21之前预先判断待推送信息总量是否超出数量阈值，还可以降低消息推送平台处理数据的数据量，提升消息推送平台的处理速度和效率，以便于待推送信息的快速推送。
请参考图8，本申请还提供一种对应上述基于用户群组的消息分群推送方法的基于用户群组的消息分群推送装置，包括：
分类单元81，用于根据待推送信息的内容对待推送信息进行分类得到分类信息；
分群单元82，用于获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；
推送单元83，用于根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
本申请的实施例还提供一种电子设备，所述电子设备包括：
处理器；
用于存储处理器可执行指令的存储器；
其中，所述处理器被配置为如上述任一项实施例所述基于用户群组的消息 推送方法的步骤。
本申请的实施例还提供一种非临时性计算机可读存储介质，其特征在于，当所述存储介质中的指令由移动终端的处理器执行时，使得移动终端能够执行如上述任一项实施例所述基于用户群组的消息推送方法的步骤。
上述基于用户群组的消息推送方法，通过根据待推送信息的内容对待推送信息进行分类得到分类信息；获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户的技术方案，针对用户进行划分用户群组，根据用户群组对应用户分群信息匹配分类信息得到待推送信息，最终向用户群组中的用户推送特定的分类的待推送信息，能够提升待推送信息的推送准确性，进而提升待推送信息的推送效率，避免无效推送或者引起反作用的消极推送。
以上所述仅是本申请的部分实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本申请原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本申请的保护范围。
Claims (20)
- 一种基于用户群组的消息推送方法，其特征在于，包括：根据待推送信息的内容对待推送信息进行分类得到分类信息；获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
- 根据权利要求1所述的基于用户群组的消息推送方法，其特征在于，所述获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息的步骤之前，还包括：采集各个用户在指定时间段内的点击行为、转发行为、收藏行为、投保行为、购买行为、关闭行为、投诉行为的至少一种用户行为；统计在所述指定时间段内所述用户行为的发生频次和/或对应用户行为的时间节点生成行为时间参数；根据所述用户行为和行为时间参数生成用户行为信息。
- 根据权利要求1所述的基于用户群组的消息推送方法，其特征在于，所述根据待推送信息的内容对待推送信息进行分类得到分类信息的步骤，包括：从所述待推送信息的内容中提取相应的关键词；根据所述关键词在待推送信息的分类列表中匹配分类得到分类信息。
- 根据权利要求1所述的基于用户群组的消息推送方法，其特征在于，所述获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息的步骤，包括：根据用户的交互行为得到用户行为信息；将用户行为信息输入预先经过训练的用户分类模型中将用户分群为若干用户群组；根据用户所划分的用户群组得到用户分群信息。
- 根据权利要求1所述的基于用户群组的消息推送方法，其特征在于，所述根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推 送至用户群组内的用户的步骤，包括：将所述用户分群信息输入推送类别模型匹配分类信息；根据分类信息获取对应的待推送信息；根据用户分群信息将待推送信息推送至用户群组内的用户。
- 根据权利要求5所述的基于用户群组的消息推送方法，其特征在于，将所述用户分群信息输入推送类别模型匹配分类信息的步骤之前，还包括：提取用户分群信息中各用户群组的用户特征信息；获取用户分群信息对应各用户群组的历史推送信息；根据历史推送信息对应的历史行为信息计算得到反馈值；筛选反馈值大于反馈阈值的历史推送信息并提取历史推送信息中的历史推送特征信息；根据用户特征信息和历史推送特征信息的关系生成推送类别模型。
- 根据权利要求1所述的基于用户群组的消息推送方法，其特征在于，所述根据待推送信息的内容对待推送信息进行分类得到分类信息的步骤之前，还包括：提取用户行为信息中的数量阈值；获取指定时间段内对用户推送待推送信息的已推送数量；当已推送数量小于数量阈值时，根据待推送信息的内容对待推送信息进行分类得到分类信息。
- 一种基于用户群组的消息分群推送装置，其特征在于，包括：分类单元，用于根据待推送信息的内容对待推送信息进行分类得到分类信息；分群单元，用于获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；推送单元，用于根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
- 根据权利要求8所述的装置，其特征在于，所述分群单元还用于：采集各个用户在指定时间段内的点击行为、转发行为、收藏行为、投保行 为、购买行为、关闭行为、投诉行为的至少一种用户行为；统计在所述指定时间段内所述用户行为的发生频次和/或对应用户行为的时间节点生成行为时间参数；根据所述用户行为和行为时间参数生成用户行为信息。
- 根据权利要求8所述的装置，其特征在于，所述分类单元具体用于：从所述待推送信息的内容中提取相应的关键词；根据所述关键词在待推送信息的分类列表中匹配分类得到分类信息。
- 根据权利要求8所述的装置，其特征在于，所述分群单元具体用于：根据用户的交互行为得到用户行为信息；将用户行为信息输入预先经过训练的用户分类模型中将用户分群为若干用户群组；根据用户所划分的用户群组得到用户分群信息。
- 根据权利要求8所述的装置，其特征在于，所述推送单元具体用于：将所述用户分群信息输入推送类别模型匹配分类信息；根据分类信息获取对应的待推送信息；根据用户分群信息将待推送信息推送至用户群组内的用户。
- 根据权利要求12所述的装置，其特征在于，所述推送单元将所述用户分群信息输入推送类别模型匹配分类信息之前，还用于：提取用户分群信息中各用户群组的用户特征信息；获取用户分群信息对应各用户群组的历史推送信息；根据历史推送信息对应的历史行为信息计算得到反馈值；筛选反馈值大于反馈阈值的历史推送信息并提取历史推送信息中的历史推送特征信息；根据用户特征信息和历史推送特征信息的关系生成推送类别模型。
- 根据权利要求8所述的装置，其特征在于，所述推送单元根据待推送信息的内容对待推送信息进行分类得到分类信息之前，还用于：提取用户行为信息中的数量阈值；获取指定时间段内对用户推送待推送信息的已推送数量；当已推送数量小于数量阈值时，根据待推送信息的内容对待推送信息进行 分类得到分类信息。
- 一种电子设备，其特征在于，所述电子设备包括：处理器；用于存储处理器可执行指令的存储器；其中，所述处理器被配置用于调用指令，执行以下步骤：根据待推送信息的内容对待推送信息进行分类得到分类信息；获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息；其中，所述用户行为信息是用户交互过程中的行为信息，所述用户分群信息是用户分群的信息；根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户。
- 根据权利要求15所述的电子设备，其特征在于，所述处理器在获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息之前，还执行以下步骤：采集各个用户在指定时间段内的点击行为、转发行为、收藏行为、投保行为、购买行为、关闭行为、投诉行为的至少一种用户行为；统计在所述指定时间段内所述用户行为的发生频次和/或对应用户行为的时间节点生成行为时间参数；根据所述用户行为和行为时间参数生成用户行为信息。
- 根据权利要求15所述的电子设备，其特征在于，所述处理器在根据待推送信息的内容对待推送信息进行分类得到分类信息时，具体执行以下步骤：从所述待推送信息的内容中提取相应的关键词；根据所述关键词在待推送信息的分类列表中匹配分类得到分类信息。
- 根据权利要求15所述的电子设备，其特征在于，所述处理器获取用户的用户行为信息，并根据用户行为信息将用户分群为若干用户群组，得到用户分群信息时，具体执行以下步骤：根据用户的交互行为得到用户行为信息；将用户行为信息输入预先经过训练的用户分类模型中将用户分群为若干用户群组；根据用户所划分的用户群组得到用户分群信息。
- 根据权利要求15所述的电子设备，其特征在于，所述处理器在根据用户分群信息匹配分类信息中对应分类的待推送信息，将待推送信息推送至用户群组内的用户时，具体执行以下步骤：将所述用户分群信息输入推送类别模型匹配分类信息；根据分类信息获取对应的待推送信息；根据用户分群信息将待推送信息推送至用户群组内的用户。
- 一种非临时性计算机可读存储介质，其特征在于，当所述存储介质中的指令由移动终端的处理器执行时，使得移动终端能够执行如权利要求1～7中任一项权利要求所述基于用户群组的消息推送方法的步骤。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910117573.5A CN109995847A (zh) | 2019-02-15 | 2019-02-15 | 基于用户群组的消息推送方法、装置及计算机设备 |
| CN201910117573.5 | 2019-02-15 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020164269A1 true WO2020164269A1 (zh) | 2020-08-20 |
Family
ID=67129452
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2019/117578 WO2020164269A1 (zh) | 2019-02-15 | 2019-11-12 | 基于用户群组的消息推送方法、装置及计算机设备 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109995847A (zh) |
| WO (1) | WO2020164269A1 (zh) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113076473A (zh) * | 2021-03-19 | 2021-07-06 | 土巴兔集团股份有限公司 | 用户数据处理方法、装置、计算机设备及存储介质 |
| CN113572841A (zh) * | 2021-07-23 | 2021-10-29 | 上海哔哩哔哩科技有限公司 | 信息推送方法及装置 |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109995847A (zh) * | 2019-02-15 | 2019-07-09 | 平安科技（深圳）有限公司 | 基于用户群组的消息推送方法、装置及计算机设备 |
| CN110765354B (zh) * | 2019-10-22 | 2022-11-22 | 北京字节跳动网络技术有限公司 | 信息的推送方法、装置、电子设备及存储介质 |
| CN111464641B (zh) * | 2020-03-31 | 2022-11-01 | 深圳前海微众银行股份有限公司 | 消息推送优化方法、设备及可读存储介质 |
| CN113300935B (zh) * | 2020-09-15 | 2023-05-26 | 阿里巴巴集团控股有限公司 | 群组处理方法、终端设备、服务端设备及存储介质 |
| CN112417284A (zh) * | 2020-11-23 | 2021-02-26 | 北京三快在线科技有限公司 | 推送展示信息的方法和装置 |
| CN113706153A (zh) * | 2021-08-04 | 2021-11-26 | 支付宝(杭州)信息技术有限公司 | 针对支付交易进行举报引导、举报处理的方法及装置 |
| CN114520826B (zh) * | 2021-12-31 | 2024-04-05 | 珠海华发金融科技研究院有限公司 | 跨平台信息匹配方法、装置及云端智能机器人 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6981040B1 (en) * | 1999-12-28 | 2005-12-27 | Utopy, Inc. | Automatic, personalized online information and product services |
| CN102855248A (zh) * | 2011-06-29 | 2013-01-02 | 中国移动通信集团广西有限公司 | 一种用户特征信息的确定方法、装置及系统 |
| CN109145280A (zh) * | 2017-06-15 | 2019-01-04 | 北京京东尚科信息技术有限公司 | 信息推送的方法和装置 |
| CN109325845A (zh) * | 2018-08-15 | 2019-02-12 | 深圳市和讯华谷信息技术有限公司 | 一种金融产品智能推荐方法及系统 |
| CN109995847A (zh) * | 2019-02-15 | 2019-07-09 | 平安科技（深圳）有限公司 | 基于用户群组的消息推送方法、装置及计算机设备 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102760138B (zh) * | 2011-04-26 | 2015-03-11 | 北京百度网讯科技有限公司 | 用户网络行为的分类方法和装置及对应的搜索方法和装置 |
| CN102916867B (zh) * | 2012-10-12 | 2019-05-28 | 北京百度网讯科技有限公司 | 一种消息推送方法及系统 |
| CN103294800B (zh) * | 2013-05-27 | 2016-12-28 | 华为技术有限公司 | 一种信息推送方法及装置 |
| CN103647800B (zh) * | 2013-11-19 | 2017-12-12 | 乐视致新电子科技（天津）有限公司 | 推荐应用资源的方法及系统 |
| CN105159910A (zh) * | 2015-07-03 | 2015-12-16 | 安一恒通（北京）科技有限公司 | 信息推荐方法和装置 |
| CN106777228A (zh) * | 2016-12-26 | 2017-05-31 | 北京金山安全软件有限公司 | 一种消息推送方法、装置及电子设备 |
-
2019
- 2019-02-15 CN CN201910117573.5A patent/CN109995847A/zh active Pending
- 2019-11-12 WO PCT/CN2019/117578 patent/WO2020164269A1/zh active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6981040B1 (en) * | 1999-12-28 | 2005-12-27 | Utopy, Inc. | Automatic, personalized online information and product services |
| CN102855248A (zh) * | 2011-06-29 | 2013-01-02 | 中国移动通信集团广西有限公司 | 一种用户特征信息的确定方法、装置及系统 |
| CN109145280A (zh) * | 2017-06-15 | 2019-01-04 | 北京京东尚科信息技术有限公司 | 信息推送的方法和装置 |
| CN109325845A (zh) * | 2018-08-15 | 2019-02-12 | 深圳市和讯华谷信息技术有限公司 | 一种金融产品智能推荐方法及系统 |
| CN109995847A (zh) * | 2019-02-15 | 2019-07-09 | 平安科技（深圳）有限公司 | 基于用户群组的消息推送方法、装置及计算机设备 |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113076473A (zh) * | 2021-03-19 | 2021-07-06 | 土巴兔集团股份有限公司 | 用户数据处理方法、装置、计算机设备及存储介质 |
| CN113076473B (zh) * | 2021-03-19 | 2024-05-07 | 土巴兔集团股份有限公司 | 用户数据处理方法、装置、计算机设备及存储介质 |
| CN113572841A (zh) * | 2021-07-23 | 2021-10-29 | 上海哔哩哔哩科技有限公司 | 信息推送方法及装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109995847A (zh) | 2019-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020164269A1 (zh) | 基于用户群组的消息推送方法、装置及计算机设备 | |
| US11436430B2 (en) | Feature information extraction method, apparatus, server cluster, and storage medium | |
| US10147037B1 (en) | Method and system for determining a level of popularity of submission content, prior to publicizing the submission content with a question and answer support system | |
| US10348550B2 (en) | Method and system for processing network media information | |
| US8874674B2 (en) | System for optimizing social networking | |
| US20140358694A1 (en) | Social media pricing engine | |
| WO2019128514A1 (zh) | 推送信息的方法和装置、确定输入默认值的方法和装置 | |
| CN108112038B (zh) | 一种控制访问流量的方法及装置 | |
| US11748452B2 (en) | Method for data processing by performing different non-linear combination processing | |
| US20230004979A1 (en) | Abnormal behavior detection method and apparatus, electronic device, and computer-readable storage medium | |
| CN106910054A (zh) | 一种支付业务数据处理的方法和装置 | |
| WO2019232821A1 (zh) | 风控数据处理方法、装置、计算机设备及存储介质 | |
| CN111061948B (zh) | 一种用户标签推荐方法、装置、计算机设备及存储介质 | |
| CN110910201A (zh) | 信息推荐的控制方法、装置、计算机设备及存储介质 | |
| US11550999B2 (en) | Data management using topic modeling | |
| Díaz Ferreyra et al. | Learning from online regrets: from deleted posts to risk awareness in social network sites | |
| CN115082203A (zh) | 生息方案推送方法、装置、电子设备及存储介质 | |
| CN110827155A (zh) | 信息处理方法、装置、电子设备及存储介质 | |
| US10394804B1 (en) | Method and system for increasing internet traffic to a question and answer customer support system | |
| CN117093762B (zh) | 一种舆情数据评估分析系统及方法 | |
| WO2018006726A1 (zh) | 识别用户潜在求助的知识点的方法及装置 | |
| CN114691894B (zh) | 电子信息全链路知识图谱的无纸化财务数据管理方法 | |
| CN112070564B (zh) | 广告拉取方法、装置、系统与电子设备 | |
| US20160342740A1 (en) | Software for emergency medical services | |
| CN115147117A (zh) | 资源使用异常的账户群组识别方法、装置及设备 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 19915119 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 19915119 Country of ref document: EP Kind code of ref document: A1 |