WO2020168758A1 - 信息推送方法、装置及计算机可读存储介质 - Google Patents
信息推送方法、装置及计算机可读存储介质 Download PDFInfo
- Publication number
- WO2020168758A1 WO2020168758A1 PCT/CN2019/118423 CN2019118423W WO2020168758A1 WO 2020168758 A1 WO2020168758 A1 WO 2020168758A1 CN 2019118423 W CN2019118423 W CN 2019118423W WO 2020168758 A1 WO2020168758 A1 WO 2020168758A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- weight
- message
- user
- day
- tag
- 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.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Recommending goods or services
Definitions
- This application relates to research and development management, and specifically to information push methods, devices and computer-readable storage media.
- every APP on the market has a user's message center entrance, and the content displayed is classified and displayed according to the message type.
- the bank APP message center is divided into announcements, exciting activities, debit card transactions, credit card services, wealth management services, and loan services , Welfare, other, etc.
- the inventor realized that after entering the message center, users need to click on each category to see the specific message list.
- the content of the message list may not be what the user wants to care about and needs.
- the message content that the user wants to see may be deep in the message list. Users cannot see.
- this application provides an information push method applied to an electronic device, including the following steps: by embedding message behavior events in the message center, obtaining user behavior data on messages on APP, and counting users Click behavior of all messages of the APP on that day; tag the messages clicked by users according to the message portrait with interest tags, and then count all the clicks of all the labels corresponding to all push messages, where the message portrait refers to the attribute information of the push message
- the abstract tagged message model establishes the mapping relationship between the user clicked on the message and the label through the message portrait, and converts the user’s click on the message into statistics on the click on the label; calculates the weight of the user’s label i on the day according to the following formula :
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- the tag weight of the day is calculated, and the message corresponding to the tag is pushed to the user according to the tag weight.
- This application also provides an information push device, including:
- the user behavior acquisition module by burying the message behavior event in the message center, acquires the user's behavior data on the message on the APP, and counts the user's click behavior on all messages of the APP on the day;
- the tagging module tags the messages clicked by the user with interest tags according to the message portrait, and then counts the total number of clicks of all the tags corresponding to all push messages.
- the message portrait refers to the tagged message abstracted based on the attribute information of the push message
- the model establishes the mapping relationship between the message clicked by the user and the label through the message portrait, and converts the user’s click on the message into statistics on the click on the label;
- the weight acquisition module calculates the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- the message recommendation module calculates the tag weight of the day for the tag corresponding to the push message of the APP, and decides to push the message corresponding to the tag to the user according to the tag weight.
- the present application also provides an electronic device, the electronic device comprising: a memory and a processor, and an information push program is stored in the memory.
- the information push program is executed by the processor, the following steps are implemented: Bury the message behavior event, obtain the user behavior data on the message on the APP, and count the user's click behavior on all the messages of the APP on the day; according to the message portrait, the user clicked on the message is marked with interest, and then all the push messages are counted
- the message profile refers to a tagged message model abstracted from the attribute information of the push message.
- the mapping relationship between the message clicked by the user and the tag is established through the message profile, and the user is The clicks are converted into statistics of clicks on the label;
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- the tag weight is calculated, and the message corresponding to the tag is pushed to the user according to the tag weight.
- the present application also provides a computer non-volatile readable storage medium, the computer non-volatile readable storage medium stores a computer program, the computer program includes program instructions, and when the program instructions are executed by a processor, Implement the information push method as described above.
- the information push method, device and computer non-volatile readable storage medium of the present application combine the current behavior and hobbies of the user to recommend the desired message content to the user in real time, thereby enhancing the retention and value of the message center.
- the weight can be obtained based on the clicks of the tags in the previous few days in the near future, so as to assist in judging the user's preference and to better understand the user's preference. And according to the user's operations such as clicking, forwarding, deleting, and staying time to correct the weight of the label, it can push messages to the user more accurately.
- FIG. 1 is a schematic flowchart of an information pushing method according to an embodiment of the present application
- FIG. 2 is a schematic diagram of the hardware architecture of an electronic device according to an embodiment of the present application.
- Fig. 3 is a block diagram of the information push program in an embodiment of the present application.
- FIG. 1 is a schematic flowchart of an information push method provided by an embodiment of the application. The method is applied to an electronic device and includes the following steps:
- Step S10 by burying the message behavior event in the message center, obtain the user behavior data of the message on the APP, and then report it.
- the message can be any change message such as fund transfer, wealth management products, loan products, financial consulting, advertisements, etc. For example, newly launched financial product promotion news, asset change notification news, insurance promotion information and so on.
- the behavior data may be clicks, deletions, forwarding, stay time, number of clicks, etc. of the message by the user.
- the APP refers to a kind of software installed on mobile terminals such as smart phones and tablet computers.
- Step S20 Tag the message clicked by the user with interest tags according to the message portrait, and then count the total number of clicks of all tags corresponding to all push messages, where the message portrait refers to a tagged message model abstracted based on the attribute information of the push message Establish the mapping relationship between the user clicked message and the label through the message portrait, and convert the user’s click on the message into statistics on the click on the label. All push messages have corresponding tags in the message portrait in advance.
- the label of the auto insurance push message can be auto insurance, and more specifically, it can also include the next-level label commercial insurance.
- the push message corresponding to the commercial insurance label may be, for example, scratch insurance, theft rescue, glass insurance, etc.
- the push message corresponding to the used car may include the label luxury car, and the push message may include the push message of luxury cars such as Porsche and Mercedes-Benz. It also includes mid-range cars, such as Volkswagen, Jetta, Volvo, etc. depending on the set price. There are also low-end car tags, such as push messages from Alto and Wuling. As mentioned above, in the message portrait, all push messages have corresponding tags.
- the push messages received by the user can be tagged, so that the messages clicked by the user are described with highly general and easy-to-understand tags, and the user's clicks on the tags can be counted.
- the label is a robust wealth management product.
- the user may click on multiple push messages a day, some of which correspond to the same label. This counts the number of clicks on the label by the user that day.
- the mapping relationship between the message clicked by the user and the label can be used to convert the user's click on the message into the statistics of the click on the label.
- Kafka a high-throughput distributed publish-subscribe messaging system
- spark streaming a streaming framework
- Step S30 for the tag corresponding to the push message of the APP, calculate the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day.
- S40 Determine to push a message corresponding to the tag to the user according to the tag weight.
- messages corresponding to tags are pushed in the order of tag weights from high to low.
- the user's label weights in the previous days are also counted, and the label weight summary is weighed by the impact factor.
- the formula is as follows:
- weight d, t (weight s, t * ⁇ +weight s, t-1 * ⁇ +weight s, t-2 * ⁇ ...+weight s, tm *k)
- weight d, t is the summary of the weight of user tag i on the day;
- weight s, t is the click weight of user tag i on the day
- weight s, t-1 is the weight summary of user tag i on the first day before the current day;
- weight s, t-2 is the summary of the weight of user tag i on the second day before that day;
- weight s, tm is the summary of the weight of user tag i on the mth day before that day;
- ⁇ , ⁇ , ⁇ ...k are influencing factors, taking values between 0 and 1, and the sum of ⁇ , ⁇ , ⁇ ...k is 1.
- the first day before the current day mentioned above refers to the day before the nearest day
- the second day before the current day refers to the day before the second nearest day
- the impact factor ⁇ can be artificially set. And, preferably, ⁇ , ⁇ , ⁇ ...k are set in the order of gradually decreasing numerical values. That is to say, based on the principle that the closer the date is, the larger the impact factor, and the earlier the date, the smaller the impact factor.
- a correction coefficient is also set, and its formula is as follows:
- weight d, t (weight s, t * ⁇ +h+weight s, t-1 * ⁇ +weight s, t-2 * ⁇ ...weight s, tm *k)
- C is the adjustment parameter, its initial value is 0,
- a correction coefficient is used to correct the weight.
- the adjustment parameter C+1 of the correction coefficient corresponding to the message For the staying time, if the staying time exceeds the average staying time, the adjustment parameter C+1 of the correction coefficient corresponding to the message, where: the average staying time is the average value of the user's staying time for all messages using the buried point statistics.
- the electronic device 2 is a device that can automatically perform numerical calculation and/or information processing according to pre-set or stored instructions.
- it can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack server, a blade server, a tower server or a rack server (including an independent server, or a server cluster composed of multiple servers).
- the electronic device 2 at least includes, but is not limited to, a memory 21, a processor 22, and a network interface 23 that can be communicatively connected to each other through a system bus.
- the memory 21 includes at least one type of computer non-volatile readable storage medium
- the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random Access memory (RAM), static random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, etc.
- the memory 21 may be an internal storage unit of the electronic device 2, such as a hard disk or a memory of the electronic device 2.
- the memory 21 may also be an external storage device of the electronic device 2, for example, a plug-in hard disk equipped on the electronic device 2, a smart media card (SMC), a secure digital (Secure Digital, SD) card, flash card (Flash Card), etc.
- the memory 21 may also include both the internal storage unit of the electronic device 2 and its external storage device.
- the memory 21 is generally used to store the operating system and various application software installed in the electronic device 2, such as the information push program code.
- the memory 21 can also be used to temporarily store various types of data that have been output or will be output.
- the processor 22 may be a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments.
- the processor 22 is generally used to control the overall operation of the electronic device 2, for example, to perform data interaction or communication-related control and processing with the electronic device 2.
- the processor 22 is used to run the program code or process data stored in the memory 21, for example, to run the information push program.
- the network interface 23 may include a wireless network interface or a wired network interface, and the network interface 23 is generally used to establish a communication connection between the electronic device 2 and other electronic devices.
- the network interface 23 is used to connect the electronic device 2 with a push platform through a network, and establish a data transmission channel and a communication connection between the electronic device 2 and the push platform.
- the network may be Intranet, Internet, Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network , Bluetooth (Bluetooth), Wi-Fi and other wireless or wired networks.
- the electronic device 2 may also include a display, and the display may also be called a display screen or a display unit.
- the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an organic light-emitting diode (OLED) display, etc.
- the display is used for displaying information processed in the electronic device 2 and for displaying a visualized user interface.
- FIG. 2 only shows the electronic device 2 with components 21-23, but it should be understood that it is not required to implement all of the illustrated components, and more or fewer components may be implemented instead.
- the memory 21 containing a readable storage medium may include an operating system, an information push program 50, and the like.
- the processor 22 implements the following steps when executing the information pushing program 50 in the memory 21:
- Step S10 by burying the message behavior event in the message center, obtain the user behavior data of the message on the APP, and then report it.
- the behavior data may be clicks, deletions, forwarding, stay time, number of clicks, etc. of the message by the user.
- Step S20 Tag the message clicked by the user with interest tags according to the message portrait, and then count the total number of clicks of all tags corresponding to all push messages, where the message portrait refers to a tagged message model abstracted based on the attribute information of the push message Establish the mapping relationship between the user clicked message and the label through the message portrait, and convert the user’s click on the message into statistics on the click on the label.
- Step S30 for the tag corresponding to the push message of the APP, calculate the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- Step S40 Determine to push the message corresponding to the tag to the user according to the tag weight.
- messages corresponding to tags are pushed in the order of tag weights from high to low.
- the information push program stored in the memory 21 may be divided into one or more program modules, and the one or more program modules are stored in the memory 21 and can be processed by one or more
- the processor (in this embodiment, the processor 22) is executed to complete the application.
- FIG. 3 shows a schematic diagram of the program modules of the information push program.
- the information push program 50 can be divided into a user behavior acquisition module 501, a tagging module 502, a weight acquisition module 503, and a message recommendation module. Module 504.
- the program module referred to in this application refers to a series of computer program instruction segments that can complete specific functions, and is more suitable than a program to describe the execution process of the information push program in the electronic device 2. The following description will specifically introduce the specific functions of the program modules.
- the user behavior obtaining module 501 is configured to obtain user behavior data of messages on the APP by embedding message behavior events in the message center.
- the message can be any change message such as fund transfer, wealth management products, loan products, financial consulting, advertisements, etc. For example, newly launched financial product promotion news, asset change notification news, insurance promotion information and so on.
- the behavior data may be clicks, deletions, forwarding, stay time, number of clicks, etc. of the message by the user.
- the user behavior acquisition module 501 counts the user's click behavior on all messages of the APP on that day. Through the method of burying points, it is possible to count the behavior data of the user's clicks, deletions, forwarding, stay time, and number of clicks on the message. In this step, the user's click behavior on the message is first counted. For example, China Merchants Securities' wealth management APP Zhiyuan Yihutong has stock recommendation news, wealth management recommendation news, and financial information. Count the user's clicks on all push messages on Zhiyuan One Account on the day.
- the tagging module 502 tags the messages clicked by the user with interest tags according to the message portrait, and then counts the total number of clicks of all the tags corresponding to all the push messages.
- the message portrait refers to the tagged message abstracted from the attribute information of the push message.
- the model establishes the mapping relationship between the message clicked by the user and the label through the message portrait, and converts the user’s click on the message into statistics on the click on the label. All push messages have corresponding tags in the message portrait in advance.
- the label of the auto insurance push message can be auto insurance, and more specifically, it can also include the next-level label commercial insurance.
- the push message corresponding to the commercial insurance label may be, for example, scratch insurance, theft rescue, glass insurance, etc.
- the push message corresponding to the used car may include the label luxury car, and the push message may include the push message of luxury cars such as Porsche and Mercedes-Benz. It also includes mid-range cars, such as Volkswagen, Jetta, Volvo, etc. depending on the set price. There are also tags for arriving cars, such as push messages from Alto and Wuling. As mentioned above, in the message portrait, all push messages have corresponding tags.
- the push messages received by the user can be tagged, so that the messages clicked by the user are described with highly general and easy-to-understand tags, and then the user's clicks on the tags can be counted.
- the label is a robust wealth management product.
- the user may click on multiple push messages a day, some of which correspond to the same label. This counts the number of clicks on the label by the user that day.
- the mapping relationship between the message clicked by the user and the label can be used to convert the user's click on the message into the statistics of the click on the label.
- Kafka a high-throughput distributed publish-subscribe messaging system
- spark streaming a streaming framework
- the weight obtaining module 503 is used to calculate the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- the tag weight is calculated, and the message recommendation module 504 decides to push the message corresponding to the tag to the user according to the tag weight.
- messages corresponding to tags are pushed in the order of tag weights from high to low.
- it further includes a previous tag weight acquisition module 505, which is used to count the tag weights of the user in the past days, and weigh the tag weight summary by the impact factor, the formula is as follows:
- weight d, t (weight s, t * ⁇ +weight s, t-1 * ⁇ +weight s, t-2 * ⁇ ...+weight s, tm *k)
- weight d, t is the summary of the weight of user tag i on the day;
- weight s, t is the click weight of user tag i on the day
- weight s, t-1 is the weight summary of user tag i on the first day before the current day;
- weight s, t-2 is the summary of the weight of user tag i on the second day before that day;
- weight s, tm is the summary of the weight of user tag i on the mth day before that day;
- ⁇ , ⁇ , ⁇ ...k are influencing factors, taking values between 0 and 1, and the sum of ⁇ , ⁇ , ⁇ ...k is 1.
- the impact factor ⁇ can be artificially set. And, preferably, ⁇ , ⁇ , ⁇ ... are set in a gradually decreasing order. That is to say, based on the principle that the closer the date is, the larger the impact factor, and the earlier the date, the smaller the impact factor.
- a weight correction module 506 is further included for setting the correction coefficient, and the formula is as follows:
- weight d, t (weight s, t * ⁇ +h+weight s, t-1 * ⁇ +weight s, t-2 * ⁇ ...+weight s, tm *k)
- C is the adjustment parameter, and its initial value is 0;
- a correction coefficient is used to correct the weight.
- the adjustment parameter C+1 of the correction coefficient corresponding to the message For the staying time, if the staying time exceeds the average staying time, the adjustment parameter C+1 of the correction coefficient corresponding to the message, where: the average staying time is the average value of the user's staying time for all messages using the buried point statistics.
- the present invention also provides an information pushing device, including:
- the user behavior acquisition module 501 by burying message behavior events in the message center, acquires user behavior data on messages on the APP, and counts the user's click behavior on all messages of the APP on the day;
- the tagging module 502 tags the messages clicked by the user with interest tags according to the message portrait, and then counts the total number of clicks of all tags corresponding to all push messages.
- the message portrait refers to the tagging abstracted based on the attribute information of the push message.
- the message model establishes the mapping relationship between the message clicked by the user and the label through the message portrait, and converts the user’s click on the message into statistics on the click on the label;
- the weight obtaining module 503 calculates the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of clicks on the label that day
- the message recommendation module 504 calculates the tag weight of the day for the tag corresponding to the push message of the APP, and decides to push the message corresponding to the tag to the user according to the tag weight.
- tag sorting module which is used to push messages corresponding to tags in the order of tag weight from high to low.
- the embodiment of the present application also proposes a computer non-volatile readable storage medium.
- the computer non-volatile readable storage medium may be a hard disk, a multimedia card, an SD card, a flash memory card, an SMC, a read-only memory (ROM), erasable programmable read-only memory (EPROM), portable compact disk read-only memory (CD-ROM), USB memory, etc., or any combination of several.
- the computer non-volatile readable storage medium includes an information push program, etc., and when the information push program 50 is executed by the processor 22, the following operations are implemented:
- Step S10 by embedding the message behavior event in the message center to obtain the behavior data of the user on the message on the APP.
- the message can be any change message such as fund transfer, wealth management products, loan products, financial consulting, advertisements, etc. For example, newly launched financial product promotion news, asset change notification news, insurance promotion information and so on.
- the behavior data may be clicks, deletions, forwarding, stay time, number of clicks, etc. of the message by the user.
- Step S20 Tag the message clicked by the user with interest tags according to the message portrait, and then count all the click times of all the tags corresponding to all push messages, where the message portrait refers to a tagged message model abstracted based on the attribute information of the push message Establish the mapping relationship between the user clicked message and the label through the message portrait, and convert the user’s click on the message into statistics on the click on the label. All push messages have corresponding tags in the message portrait in advance.
- Step S30 for the tag corresponding to the push message of the APP, calculate the weight of the user tag i of the day according to the following formula:
- weight s, t is the click weight of tag i on the day
- T i is the number of times the user clicks on the tag i on the day
- N is the number of tags clicked on that day
- Step S40 Determine to push the message corresponding to the tag to the user according to the tag weight.
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Information Transfer Between Computers (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
本方案涉及研发管理,提供信息推送方法、装置及计算机可读存储介质,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,然后上报;统计用户当日对APP的所有消息的点击行为;根据消息画像为用户点击的消息打上兴趣标签,统计出所有推送消息对应的所有标签的全部点击次数,获取当日用户标签i的权重:针对APP的推送消息对应的标签,都计算标签权重,依据标签权重决定对用户推送与标签对应的消息。本申请结合用户当前行为和兴趣爱好,给用户实时推荐想要的消息内容,提升消息中心的留存和价值。根据近期前几日的标签点击情况来获取权重,从而辅助判断用户的偏好,能够更好的了解用户偏好。
Description
本申请要求于2019年2月19日提交的中国专利申请号2019101218694的优先权益,上述案件全部内容以引用的方式并入本文中。
本申请涉及研发管理,具体地说,涉及信息推送方法、装置及计算机可读存储介质。
目前市场上各APP都有用户的消息中心入口,展示的内容都是按消息类型进行分类展示,比如银行APP消息中心分为公告、精彩活动、借记卡交易、信用卡服务、理财服务、贷款服务、福利、其他等。发明人意识到,用户进入消息中心后需要点击各分类后才能看到具体的消息列表,同时消息列表的内容可能不是用户想关心和需要的,用户想要看的消息内容可能在消息列表很深用户看不到。
发明内容
为解决以上技术问题,本申请提供一种信息推送方法,应用于电子装置,包括以下步骤:通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计;根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
针对该APP的推送消息对应的标签,都计算当日的标签权重,并依据标签权重决定对用户推送与标签对应的消息。
本申请还提供一种信息推送装置,包括:
用户行为获取模块,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;
打标签模块,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计;
权重获取模块,根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
消息推荐模块,针对该APP的推送消息对应的标签,都计算当日的标签权重,并依据标签权重决定对用户推送与标签对应的消息。
本申请还提供一种电子装置,该电子装置包括:存储器和处理器,所述存储器中存储有信息推送程序,所述信息推送程序被所述处理器执行时实现如下步骤:通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映 射关系,把用户对消息的点击转换为对标签的点击情况统计;
根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
针对该APP的推送消息对应的标签,都计算标签权重,并依据标签权重决定对用户推送与标签对应的消息。
本申请还提供一种计算机非易失性可读存储介质,所述计算机非易失性可读存储介质存储有计算机程序,所述计算机程序包括程序指令,所述程序指令被处理器执行时,实现如上所述的信息推送方法。
本申请的信息推送方法、装置及计算机非易失性可读存储介质结合用户当前行为和兴趣爱好,给用户实时推荐想要的消息内容,提升消息中心的留存和价值。可以根据近期前几日的标签点击情况来获取权重,从而辅助判断用户的偏好,能够更好的了解用户偏好。并根据用户对标签的例如点击、转发、删除、停留时间等操作来修正权重,可以更加准确的为用户推送消息。
通过结合下面附图对其实施例进行描述,本申请的上述特征和技术优点将会变得更加清楚和容易理解。
图1是本申请实施例的信息推送方法的流程示意图;
图2是本申请实施例的电子装置的硬件架构示意图;
图3是本申请实施例的信息推送程序的模块构成图。
下面将参考附图来描述本申请所述的信息推送方法、装置及计算机非易失性可读存储介质的实施例。本领域的普通技术人员可以认识到,在不偏离 本申请的精神和范围的情况下,可以用各种不同的方式或其组合对所描述的实施例进行修正。因此,附图和描述在本质上是说明性的,而不是用于限制权利要求的保护范围。此外,在本说明书中,附图未按比例画出,并且相同的附图标记表示相同的部分。
图1为本申请实施例提供的信息推送方法的流程示意图,该方法应用于电子装置,包括以下步骤:
步骤S10,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,然后上报。消息可以是例如资金动账、理财产品、贷款产品、金融咨询、广告等的任何变动消息。例如,新推出的理财产品推广消息,资产变动通知消息,保险推广信息等等。所述行为数据可以是用户对消息的点击、删除、转发、停留时间、点击次数等。所述APP是指一种安装在智能手机、平板电脑等移动终端上的软件。
统计用户当日对该APP的所有消息的点击行为。通过埋点方法,可以统计出用户对消息的点击、删除、转发、停留时间、点击次数等的行为数据。本步骤先将用户对消息的点击行为统计出来。例如,招商证券的理财APP致远一户通,上面有股票推荐消息、理财推荐消息和财经资讯等。统计用户当日对致远一户通上的所有推送消息的点击情况。
步骤S20,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。所有推送消息都事先在消息画像中有对应的标签对应。例如车险推送消息,其标签可以是车险,再具体,还可以包括下一级标签商业险。商业险标签对应的推送消息可以是例如划痕险、盗抢险、玻璃险等。对应二手车的推送消息,可以是包括标签豪车,推送消息中可以包括例如保时捷、奔驰等豪车的推送消息。还包括中档车,根据设定的价格不同包括例如大众、捷达、沃尔沃等。还有低档车标签,例如奥拓、五菱等车的推送消息。如上所述,在消息画像中,所有推送消息都有对应的标签。
通过消息画像就可以给用户接收到的推送消息打标签,从而将用户点击的消息利用高度概括、容易理解的标签来描述,进而统计出用户对于标签的 点击情况。例如,对于用户点击的年利率5%的理财产品的推送消息,其标签为稳健性理财产品。用户一天可能点击了多个推送消息,这其中部分推送消息对应同一个标签。由此统计出用户当日对该标签的点击次数。通过消息画像可以利用用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。
例如,获取了标签点击情况后,将标签点击数据存储在移动终端,并定时通过http上报,使用flume(分布式的海量日志采集、聚合和传输的系统)收集客户端发送的消息点击情况数据,并发送到kafka(一种高吞吐量的分布式发布订阅消息系统),kafka接收消息点击情况数据,然后使用spark streaming(一种流式处理框架)进行清洗和打标签。
步骤S30,针对该APP的推送消息对应的标签,根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量。
S40,依据标签权重决定对用户推送与标签对应的消息。优选地,按照标签权重从高到低的顺序推送与标签对应的消息。
在一个可选实施例中,还统计用户之前多日的标签权重,并以影响因子来权衡标签权重汇总,其公式如下:
weight
d,t=(weight
s,t*α+weight
s,t-1*β+weight
s,t-2*γ…+weight
s,t-m*k)
其中,weight
d,t是当日用户标签i权重汇总;
weight
s,t是当日用户标签i点击权重;
weight
s,t-1是当日之前第1日用户标签i权重汇总;
weight
s,t-2是当日之前第2日用户标签i权重汇总;
.
.
.
weight
s,t-m是当日之前第m日用户标签i权重汇总;
α、β、γ…k为影响因子,在0至1之间取值,且α、β、γ…k之和为1。
以上所说的当日之前第1日是指距离当日最近的之前的一天,当日之前的第2日是指距离当日第二近的之前的一天,以此类推。
由于引入了前几日的标签i的标签权重汇总,能够更准确的获知用户对于该标签的行为所代表的含义。其中,影响因子α可以是人为的去设定。并且,优选地,α、β、γ…k按照数值大小逐渐递减的顺序设置。也就是说,以越接近当日的影响因子越大为原则,之前越久的日期的影响因子越小。
在一个可选实施例中,还设置有修正系数,其公式如下:
weight
d,t=(weight
s,t*α+h+weight
s,t-1*β+weight
s,t-2*γ…weight
s,t-m*k)
C为调整参数,其初始值为0,
根据用户对该消息点击后的操作情况,使用修正系数来修正权重。
进一步地,对于点击次数,每点击一次,则消息对应的修正系数的调整参数C+1
对于转发消息,每转发一次消息,则消息对应的修正系数的调整参数C+1;
对于删除消息,每删除一次消息,则消息对应的修正系数的调整参数C-1;
对于停留时间,停留时间超过平均停留时间,则消息对应的修正系数的调整参数C+1,其中:平均停留时间采用埋点统计用户对于所有消息停留时间的平均值。
参阅图2所示,是本申请电子装置的实施例的硬件架构示意图。本实施例中,所述电子装置2是一种能够按照事先设定或者存储的指令,自动进行数值计算和/或信息处理的设备。例如,可以是智能手机、平板电脑、笔记本电脑、台式计算机、机架式服务器、刀片式服务器、塔式服务器或机柜式服 务器(包括独立的服务器,或者多个服务器所组成的服务器集群)等。如图2所示,所述电子装置2至少包括,但不限于,可通过系统总线相互通信连接的存储器21、处理器22、网络接口23。其中:所述存储器21至少包括一种类型的计算机非易失性可读存储介质,所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中,所述存储器21可以是所述电子装置2的内部存储单元,例如该电子装置2的硬盘或内存。在另一些实施例中,所述存储器21也可以是所述电子装置2的外部存储设备,例如该电子装置2上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。当然,所述存储器21还可以既包括所述电子装置2的内部存储单元也包括其外部存储设备。本实施例中,所述存储器21通常用于存储安装于所述电子装置2的操作系统和各类应用软件,例如所述信息推送程序代码等。此外,所述存储器21还可以用于暂时地存储已经输出或者将要输出的各类数据。
所述处理器22在一些实施例中可以是中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器22通常用于控制所述电子装置2的总体操作,例如执行与所述电子装置2进行数据交互或者通信相关的控制和处理等。本实施例中,所述处理器22用于运行所述存储器21中存储的程序代码或者处理数据,例如运行所述的信息推送程序等。
所述网络接口23可包括无线网络接口或有线网络接口,该网络接口23通常用于在所述电子装置2与其他电子装置之间建立通信连接。例如,所述网络接口23用于通过网络将所述电子装置2与推送平台相连,在所述电子装置2与推送平台之间建立数据传输通道和通信连接等。所述网络可以是企业内部网(Intranet)、互联网(Internet)、全球移动通讯系统(Global System of Mobile communication,GSM)、宽带码分多址(Wideband CodeDivision Multiple Access,WCDMA)、4G网络、5G网络、蓝牙(Bluetooth)、Wi-Fi等无线或有线网络。
可选地,该电子装置2还可以包括显示器,显示器也可以称为显示屏或 显示单元。在一些实施例中可以是LED显示器、液晶显示器、触控式液晶显示器以及有机发光二极管(Organic Light-Emitting Diode,OLED)显示器等。显示器用于显示在电子装置2中处理的信息以及用于显示可视化的用户界面。
需要指出的是,图2仅示出了具有组件21-23的电子装置2,但是应理解的是,并不要求实施所有示出的组件,可以替代的实施更多或者更少的组件。
包含可读存储介质的存储器21中可以包括操作系统、信息推送程序50等。处理器22执行存储器21中信息推送程序50时实现如下步骤:
步骤S10,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,然后上报。所述行为数据可以是用户对消息的点击、删除、转发、停留时间、点击次数等。
统计用户当日对该APP的所有消息的点击行为。通过埋点方法,可以统计出用户对消息的点击、删除、转发、停留时间、点击次数等的行为数据。本步骤先将用户对消息的点击行为统计出来。
步骤S20,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。
步骤S30,针对该APP的推送消息对应的标签,根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
步骤S40,依据标签权重决定对用户推送与标签对应的消息。优选地,按照标签权重从高到低的顺序推送与标签对应的消息。
在本实施例中,存储于存储器21中的所述信息推送程序可以被分割为一 个或者多个程序模块,所述一个或者多个程序模块被存储于存储器21中,并可由一个或多个处理器(本实施例为处理器22)所执行,以完成本申请。例如,图3示出了所述信息推送程序的程序模块示意图,该实施例中,所述信息推送程序50可以被分割为用户行为获取模块501、打标签模块502、权重获取模块503、消息推荐模块504。其中,本申请所称的程序模块是指能够完成特定功能的一系列计算机程序指令段,比程序更适合于描述所述信息推送程序在所述电子装置2中的执行过程。以下描述将具体介绍所述程序模块的具体功能。
其中,用户行为获取模块501用于通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据。消息可以是例如资金动账、理财产品、贷款产品、金融咨询、广告等的任何变动消息。例如,新推出的理财产品推广消息,资产变动通知消息,保险推广信息等等。所述行为数据可以是用户对消息的点击、删除、转发、停留时间、点击次数等。
用户行为获取模块501统计用户当日对该APP的所有消息的点击行为。通过埋点方法,可以统计出用户对消息的点击、删除、转发、停留时间、点击次数等的行为数据。本步骤先将用户对消息的点击行为统计出来。例如,招商证券的理财APP致远一户通,上面有股票推荐消息、理财推荐消息和财经资讯等。统计用户当日对致远一户通上的所有推送消息的点击情况。
打标签模块502根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。所有推送消息都事先在消息画像中有对应的标签对应。例如车险推送消息,其标签可以是车险,再具体,还可以包括下一级标签商业险。商业险标签对应的推送消息可以是例如划痕险、盗抢险、玻璃险等。对应二手车的推送消息,可以是包括标签豪车,推送消息中可以包括例如保时捷、奔驰等豪车的推送消息。还包括中档车,根据设定的价格不同包括例如大众、捷达、沃尔沃等。还有抵档车标签,例如奥拓、五菱等车的推送消息。如上所述,在消息画像中,所有推送消息都有对应的标签。
通过消息画像就可以给用户接收到的推送消息打标签,从而将用户点击 的消息利用高度概括、容易理解的标签来描述,进而统计出用户对于标签的点击情况。例如,对于用户点击的年利率5%的理财产品的推送消息,其标签为稳健性理财产品。用户一天可能点击了多个推送消息,这其中部分推送消息对应同一个标签。由此统计出用户当日对该标签的点击次数。通过消息画像可以利用用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。
例如,获取了标签点击情况后,将标签点击数据存储在移动终端,并定时通过http上报,使用flume(分布式的海量日志采集、聚合和传输的系统)收集客户端发送的消息点击情况数据,并发送到kafka(一种高吞吐量的分布式发布订阅消息系统),kafka接收消息点击情况数据,然后使用spark streaming(一种流式处理框架)进行清洗和打标签。
权重获取模块503用于根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
针对该APP的推送消息对应的标签,都计算标签权重,消息推荐模块504依据标签权重决定对用户推送与标签对应的消息。优选地,按照标签权重从高到低的顺序推送与标签对应的消息。
在一个可选实施例中,还包括之前标签权重获取模块505,用于统计用户之前多日的标签权重,并以影响因子来权衡标签权重汇总,其公式如下:
weight
d,t=(weight
s,t*α+weight
s,t-1*β+weight
s,t-2*γ…+weight
s,t-m*k)
其中,weight
d,t是当日用户标签i权重汇总;
weight
s,t是当日用户标签i点击权重;
weight
s,t-1是当日之前第1日用户标签i权重汇总;
weight
s,t-2是当日之前第2日用户标签i权重汇总;
.
.
.
weight
s,t-m是当日之前第m日用户标签i权重汇总;
α、β、γ…k为影响因子,在0至1之间取值,且α、β、γ…k之和为1。
由于引入了前几日的标签i的标签权重汇总,能够更准确的获知用户对于该标签的行为所代表的含义。其中,影响因子α可以是人为的去设定。并且,优选地,α、β、γ…按照逐渐递减的顺序设置。也就是说,以越接近当日的影响因子越大为原则,之前越久的日期的影响因子越小。
在一个可选实施例中,还包括权重修正模块506,用于对修正系数进行设置,其公式如下:
weight
d,t=(weight
s,t*α+h+weight
s,t-1*β+weight
s,t-2*γ…+weight
s,t-m*k)
C为调整参数,其初始值为0;
根据用户对该消息点击后的操作情况,使用修正系数来修正权重。
进一步地,对于点击次数,每点击一次,则消息对应的修正系数的调整参数C+1
对于转发消息,每转发一次消息,则消息对应的修正系数的调整参数C+1;
对于删除消息,每删除一次消息,则消息对应的修正系数的调整参数C-1;
对于停留时间,停留时间超过平均停留时间,则消息对应的修正系数的调整参数C+1,其中:平均停留时间采用埋点统计用户对于所有消息停留时间的平均值。
本发明还提供一种信息推送装置,包括:
用户行为获取模块501,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;
打标签模块502,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的 点击情况统计;
权重获取模块503,根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击标签的数量;
消息推荐模块504,针对该APP的推送消息对应的标签,都计算当日的标签权重,并依据标签权重决定对用户推送与标签对应的消息。
另外,还包括标签排序模块,用于按照标签权重从高到低的顺序推送与标签对应的消息。
此外,本申请实施例还提出一种计算机非易失性可读存储介质,所述计算机非易失性可读存储介质可以是硬盘、多媒体卡、SD卡、闪存卡、SMC、只读存储器(ROM)、可擦除可编程只读存储器(EPROM)、便携式紧致盘只读存储器(CD-ROM)、USB存储器等等中的任意一种或者几种的任意组合。所述计算机非易失性可读存储介质中包括信息推送程序等,所述信息推送程序50被处理器22执行时实现如下操作:
步骤S10,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据。消息可以是例如资金动账、理财产品、贷款产品、金融咨询、广告等的任何变动消息。例如,新推出的理财产品推广消息,资产变动通知消息,保险推广信息等等。所述行为数据可以是用户对消息的点击、删除、转发、停留时间、点击次数等。
统计用户当日对该APP的所有消息的点击行为。通过埋点方法,可以统计出用户对消息的点击、删除、转发、停留时间、点击次数等的行为数据。本步骤先将用户对消息的点击行为统计出来。例如,招商证券的理财APP致远一户通,上面有股票推荐消息、理财推荐消息和财经资讯等。统计用户当日对致远一户通上的所有推送消息的点击情况。
步骤S20,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送 消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计。所有推送消息都事先在消息画像中有对应的标签对应。
步骤S30,针对该APP的推送消息对应的标签,根据下式计算当日用户标签i的权重:
其中,weight
s,t是当日对标签i的点击权重;
T
i是用户当日点击标签i的次数;
N是当日点击的标签数量;
步骤S40,依据标签权重决定对用户推送与标签对应的消息。
本申请之计算机非易失性可读存储介质的具体实施方式与上述信息推送方法以及电子装置2的具体实施方式大致相同,在此不再赘述。
以上所述仅为本申请的优选实施例,并不用于限制本申请,对于本领域的技术人员来说,本申请可以有各种更改和变化。凡在本申请的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。
Claims (20)
- 一种信息推送方法,应用于电子装置,其特征在于,包括以下步骤:通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计;根据下式计算当日用户标签i的权重:其中,weight s,t是当日对标签i的点击权重;T i是用户当日点击标签i的次数;N是当日点击标签的数量;针对该APP的推送消息对应的标签,都计算当日的标签权重,并依据标签权重决定对用户推送与标签对应的消息。
- 根据权利要求1所述的信息推送方法,其特征在于:按照标签权重从高到低的顺序推送与标签对应的消息。
- 根据权利要求1所述的信息推送方法,其特征在于:还统计用户当日之前m日的标签权重,并以影响因子来权衡标签权重汇总,其公式如下:weight d,t=(weight s,t*α+weight s,t-1*β+weight s,t-2*γ…+weight s,t-m*k)其中,weight d,t是当日用户标签i权重汇总;weight s,t是当日对标签i的点击权重;weight s,t-1是当日之前第1日用户标签i权重汇总;weight s,t-2是当日之前第2日用户标签i权重汇总;...weight s,t-m是当日之前第m日用户标签i权重汇总;α、β、γ…k为影响因子,在0至1之间取值,且α、β、γ…k之和为1。
- 根据权利要求1所述的信息推送方法,其特征在于:α、β、γ…k按照数值大小逐渐递减的顺序设置。
- 根据权利要求1所述的信息推送方法,其特征在于:所述行为事件至少包括对消息的点击、删除、转发、停留时间、点击次数。
- 根据权利要求5所述的信息推送方法,其特征在于:对于点击次数,每点击一次,则消息对应的修正系数的调整参数C+1;对于转发消息,每转发一次消息,则消息对应的修正系数的调整参数C+1;对于删除消息,每删除一次消息,则消息对应的修正系数的调整参数C-1;对于停留时间,停留时间超过平均停留时间,则消息对应的修正系数的调整参数C+1,其中:平均停留时间采用埋点统计用户对于所有消息停留时间的平均值。
- 根据权利要求1所述的信息推送方法,其特征在于:获取了标签点击情况后,将标签点击数据存储在移动终端,并定时通过http上报,使用flume收集客户端发送的消息点击情况数据,并发送到kafka,kafka接收消息点击情况数据,然后使用spark streaming进行清洗和打标签。
- 一种信息推送装置,其特征在于,包括:用户行为获取模块,通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;打标签模块,根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计;权重获取模块,根据下式计算当日用户标签i的权重:其中,weight s,t是当日对标签i的点击权重;T i是用户当日点击标签i的次数;N是当日点击标签的数量;消息推荐模块,针对该APP的推送消息对应的标签,都计算当日的标签权重,并依据标签权重决定对用户推送与标签对应的消息。
- 根据权利要求9所述的信息推送装置,其特征在于:还包括标签排序模块,用于按照标签权重从高到低的顺序推送与标签对应的消息。
- 根据权利要求9所述的信息推送装置,其特征在于:还设置有之前标签权重获取模块,用于统计用户当日之前m日的标签权重,并以影响因子来权衡标签权重汇总,其公式如下:weight d,t=(weight s,t*α+weight s,t-1*β+weight s,t-2*γ…+weight s,t-m*k)其中,weight d,t是当日用户标签i权重汇总;weight s,t是当日对标签i的点击权重;weight s,t-1是当日之前第1日用户标签i权重汇总;weight s,t-2是当日之前第2日用户标签i权重汇总;...weight s,t-m是当日之前第m日用户标签i权重汇总;α、β、γ…k为影响因子,在0至1之间取值,且α、β、γ…k之和为1。
- 根据权利要求11所述的信息推送装置,其特征在于:α、β、γ…k按照数值大小逐渐递减的顺序设置。
- 一种电子装置,其特征在于,该电子装置包括:存储器和处理器,所述存储器中存储有信息推送程序,所述信息推送程序被所述处理器执行时实现如下步骤:通过在消息中心中对消息行为事件进行埋点,获取用户对APP上的消息的行为数据,统计用户当日对该APP的所有消息的点击行为;根据消息画像为用户点击的消息打上兴趣标签,进而统计出所有推送消息对应的所有标签的全部点击次数,其中,消息画像是指根据推送消息的属性信息而抽象出来的标签化消息模型,通过消息画像建立用户点击的消息与标签之间的映射关系,把用户对消息的点击转换为对标签的点击情况统计;根据下式计算当日用户标签i的权重:其中,weight s,t是当日对标签i的点击权重;T i是用户当日点击标签i的次数;N是当日点击标签的数量;针对该APP的推送消息对应的标签,都计算标签权重,并依据标签权重决定对用户推送与标签对应的消息。
- 根据权利要求14所述的电子装置,其特征在于,所述信息推送程序被所述处理器执行时还实现:所述行为事件至少包括对消息的点击、删除、转发、停留时间、点击次 数。
- 根据权利要求14所述的电子装置,其特征在于,所述信息推送程序被所述处理器执行时还实现:获取了标签点击情况后,将标签点击数据存储在移动终端,并定时通过http上报,使用flume收集客户端发送的消息点击情况数据,并发送到kafka,kafka接收消息点击情况数据,然后使用spark streaming进行清洗和打标签。
- 一种计算机非易失性可读存储介质,其特征在于,所述计算机非易失性可读存储介质存储有计算机程序,所述计算机程序包括程序指令,所述程序指令被处理器执行时,实现权利要求1所述的信息推送方法。
- 根据权利要求17所述的计算机非易失性可读存储介质,其特征在于,所述程序指令被处理器执行时,还实现:按照标签权重从高到低的顺序推送与标签对应的消息。
- 根据权利要求17所述的计算机非易失性可读存储介质,其特征在于,所述程序指令被处理器执行时,还实现:所述行为事件至少包括对消息的点击、删除、转发、停留时间、点击次数。
- 根据权利要求17所述的计算机非易失性可读存储介质,其特征在于,所述程序指令被处理器执行时,还实现:获取了标签点击情况后,将标签点击数据存储在移动终端,并定时通过http上报,使用flume收集客户端发送的消息点击情况数据,并发送到kafka,kafka接收消息点击情况数据,然后使用spark streaming进行清洗和打标签。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910121869.4 | 2019-02-19 | ||
| CN201910121869.4A CN109977321A (zh) | 2019-02-19 | 2019-02-19 | 消息中心消息内容推荐方法、装置及计算机可读存储介质 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020168758A1 true WO2020168758A1 (zh) | 2020-08-27 |
Family
ID=67077096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2019/118423 Ceased WO2020168758A1 (zh) | 2019-02-19 | 2019-11-14 | 信息推送方法、装置及计算机可读存储介质 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109977321A (zh) |
| WO (1) | WO2020168758A1 (zh) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112583915A (zh) * | 2020-12-09 | 2021-03-30 | 深圳前海微众银行股份有限公司 | 一种报表数据推送方法及装置 |
| CN113590441A (zh) * | 2021-06-29 | 2021-11-02 | 中孚安全技术有限公司 | 一种基于权重计算的内网用户行为分析方法及系统 |
| CN113689259A (zh) * | 2021-08-09 | 2021-11-23 | 青岛海尔科技有限公司 | 基于用户行为的商品个性化推荐方法及系统 |
| CN113689257A (zh) * | 2021-08-09 | 2021-11-23 | 青岛海尔科技有限公司 | 一种用户行为埋点构建方法及系统 |
| CN113821596A (zh) * | 2021-08-17 | 2021-12-21 | 深圳市东信时代信息技术有限公司 | 信息推荐方法、装置、计算机设备及存储介质 |
| CN114004630A (zh) * | 2021-10-13 | 2022-02-01 | 福建福诺移动通信技术有限公司 | 基于预测模型的app智能推送引擎及方法 |
| CN114238799A (zh) * | 2021-12-22 | 2022-03-25 | 金现代信息产业股份有限公司 | 基于计算机软件菜单分析的智能关联推送方法及系统 |
| CN114417101A (zh) * | 2021-12-22 | 2022-04-29 | 浙江力石科技股份有限公司 | 一种基于数据埋点的用户识别方法及系统 |
| CN114519604A (zh) * | 2022-01-26 | 2022-05-20 | 北京金堤科技有限公司 | 确定用户推送首活比例的方法和装置、存储介质和电子设备 |
| CN115982226A (zh) * | 2021-10-15 | 2023-04-18 | 腾讯科技(深圳)有限公司 | 操作数据处理方法、装置、计算机设备和存储介质 |
| CN118013129A (zh) * | 2024-03-27 | 2024-05-10 | 广东科技学院 | 基于大数据的财经资讯推送方法、系统、设备及存储介质 |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109977321A (zh) * | 2019-02-19 | 2019-07-05 | 平安科技(深圳)有限公司 | 消息中心消息内容推荐方法、装置及计算机可读存储介质 |
| CN111090810A (zh) * | 2019-12-20 | 2020-05-01 | 北京奇艺世纪科技有限公司 | 一种应用消息的推送方法、装置及电子设备 |
| CN111161076A (zh) * | 2019-12-31 | 2020-05-15 | 中国银行股份有限公司 | 金融市场产品的推送方法及装置 |
| CN111581452B (zh) * | 2020-03-26 | 2023-10-17 | 浙江口碑网络技术有限公司 | 推荐对象数据的获得方法、装置及电子设备 |
| CN112650925B (zh) * | 2020-12-25 | 2025-02-14 | 通号信息产业有限公司 | 一种用于一卡通的app信息推送系统、方法及介质 |
| CN113259446B (zh) * | 2021-05-25 | 2022-10-14 | 平安科技(深圳)有限公司 | 基于用户忠诚度的app消息发送方法、装置、设备及介质 |
| CN113364670B (zh) * | 2021-06-16 | 2022-07-26 | 北京轻松筹信息技术有限公司 | 数据推送方法、装置及电子设备 |
| CN115187410A (zh) * | 2022-06-16 | 2022-10-14 | 平安银行股份有限公司 | 推荐标签管理方法、装置、存储介质及电子设备 |
| CN120343082A (zh) * | 2025-03-18 | 2025-07-18 | 北京同仁堂科技发展股份有限公司 | 一种基于生产数据的自动化消息订阅推送系统的实现方法 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550184A (zh) * | 2014-10-31 | 2016-05-04 | 阿里巴巴集团控股有限公司 | 一种信息获取方法及装置 |
| CN107807943A (zh) * | 2016-09-09 | 2018-03-16 | 腾讯科技(深圳)有限公司 | 应用程序推荐方法及装置 |
| CN107895276A (zh) * | 2017-08-31 | 2018-04-10 | 中国平安人寿保险股份有限公司 | 客户行为轨迹记录方法、装置及存储介质 |
| CN109977321A (zh) * | 2019-02-19 | 2019-07-05 | 平安科技(深圳)有限公司 | 消息中心消息内容推荐方法、装置及计算机可读存储介质 |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104965890B (zh) * | 2015-06-17 | 2017-05-31 | 深圳市腾讯计算机系统有限公司 | 广告推荐的方法和装置 |
| WO2017028097A1 (zh) * | 2015-08-16 | 2017-02-23 | 常平 | 推荐附近用户时的信息提醒方法和用户推荐系统 |
| CN106446189A (zh) * | 2016-09-29 | 2017-02-22 | 广州艾媒数聚信息咨询股份有限公司 | 一种资讯推荐方法及系统 |
| CN107222566B (zh) * | 2017-07-06 | 2020-09-04 | 阿里巴巴(中国)有限公司 | 信息推送方法、装置及服务器 |
-
2019
- 2019-02-19 CN CN201910121869.4A patent/CN109977321A/zh active Pending
- 2019-11-14 WO PCT/CN2019/118423 patent/WO2020168758A1/zh not_active Ceased
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105550184A (zh) * | 2014-10-31 | 2016-05-04 | 阿里巴巴集团控股有限公司 | 一种信息获取方法及装置 |
| CN107807943A (zh) * | 2016-09-09 | 2018-03-16 | 腾讯科技(深圳)有限公司 | 应用程序推荐方法及装置 |
| CN107895276A (zh) * | 2017-08-31 | 2018-04-10 | 中国平安人寿保险股份有限公司 | 客户行为轨迹记录方法、装置及存储介质 |
| CN109977321A (zh) * | 2019-02-19 | 2019-07-05 | 平安科技(深圳)有限公司 | 消息中心消息内容推荐方法、装置及计算机可读存储介质 |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112583915A (zh) * | 2020-12-09 | 2021-03-30 | 深圳前海微众银行股份有限公司 | 一种报表数据推送方法及装置 |
| CN113590441A (zh) * | 2021-06-29 | 2021-11-02 | 中孚安全技术有限公司 | 一种基于权重计算的内网用户行为分析方法及系统 |
| CN113689259A (zh) * | 2021-08-09 | 2021-11-23 | 青岛海尔科技有限公司 | 基于用户行为的商品个性化推荐方法及系统 |
| CN113689257A (zh) * | 2021-08-09 | 2021-11-23 | 青岛海尔科技有限公司 | 一种用户行为埋点构建方法及系统 |
| CN113821596A (zh) * | 2021-08-17 | 2021-12-21 | 深圳市东信时代信息技术有限公司 | 信息推荐方法、装置、计算机设备及存储介质 |
| CN114004630A (zh) * | 2021-10-13 | 2022-02-01 | 福建福诺移动通信技术有限公司 | 基于预测模型的app智能推送引擎及方法 |
| CN115982226A (zh) * | 2021-10-15 | 2023-04-18 | 腾讯科技(深圳)有限公司 | 操作数据处理方法、装置、计算机设备和存储介质 |
| CN115982226B (zh) * | 2021-10-15 | 2025-10-10 | 腾讯科技(深圳)有限公司 | 操作数据处理方法、装置、计算机设备和存储介质 |
| CN114238799A (zh) * | 2021-12-22 | 2022-03-25 | 金现代信息产业股份有限公司 | 基于计算机软件菜单分析的智能关联推送方法及系统 |
| CN114417101A (zh) * | 2021-12-22 | 2022-04-29 | 浙江力石科技股份有限公司 | 一种基于数据埋点的用户识别方法及系统 |
| CN114519604A (zh) * | 2022-01-26 | 2022-05-20 | 北京金堤科技有限公司 | 确定用户推送首活比例的方法和装置、存储介质和电子设备 |
| CN118013129A (zh) * | 2024-03-27 | 2024-05-10 | 广东科技学院 | 基于大数据的财经资讯推送方法、系统、设备及存储介质 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109977321A (zh) | 2019-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020168758A1 (zh) | 信息推送方法、装置及计算机可读存储介质 | |
| CN110097395B (zh) | 定向广告投放方法、装置及计算机可读存储介质 | |
| AU2019204122B2 (en) | Aggregation source routing | |
| JP6470735B2 (ja) | コンテンツアイテムの視認性を推定すること | |
| US9824407B2 (en) | Calculating unique social networking system users performing an action on a social networking system object | |
| CN111552870A (zh) | 对象推荐方法、电子装置及存储介质 | |
| US20130339065A1 (en) | System and method for proactive underwriting using social data | |
| WO2019061994A1 (zh) | 电子装置、保险产品推荐方法、系统及计算机可读存储介质 | |
| US20190236107A1 (en) | Device identification for multiple device ids | |
| US20120173454A1 (en) | Financial portfolio boost evaluation | |
| US20190166216A1 (en) | Information pushing method and device | |
| CN113128773B (zh) | 地址预测模型的训练方法、地址预测方法及装置 | |
| US20140280610A1 (en) | Identification of users for initiating information spreading in a social network | |
| CN113393299A (zh) | 推荐模型训练方法、装置、电子设备和存储介质 | |
| US9898748B1 (en) | Determining popular and trending content characteristics | |
| US20140358673A1 (en) | Monitoring conversions and fee determination of online advertisements using a social networking system | |
| US20190102784A1 (en) | Modeling sequential actions | |
| JP6730002B2 (ja) | 抽出装置、抽出方法及び抽出プログラム | |
| CN110889725A (zh) | 在线广告ctr预估方法、装置、设备及存储介质 | |
| US20150348134A1 (en) | Predicting the Performance of an Advertising Campaign | |
| CN110717597A (zh) | 利用机器学习模型获取时序特征的方法和装置 | |
| CN114066513A (zh) | 一种用户分类的方法和装置 | |
| US10354287B2 (en) | Monitoring conversions and fee determination of online advertisements using a social networking system | |
| CN115511561A (zh) | 展示页面处理方法、装置、电子设备及存储介质 | |
| US20160292729A1 (en) | Identifying products related to interactions performed by an online system user with content from a third party system |
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: 19915867 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: 19915867 Country of ref document: EP Kind code of ref document: A1 |










