CN113542321A

CN113542321A - Message pushing system, related method and device

Info

Publication number: CN113542321A
Application number: CN202010297404.7A
Authority: CN
Inventors: 丁小飞; 陈思聪; 何鹏; 袁志刚; 曹涌
Original assignee: Alibaba Group Holding Ltd
Current assignee: Alibaba Group Holding Ltd
Priority date: 2020-04-15
Filing date: 2020-04-15
Publication date: 2021-10-22

Abstract

The application discloses a message pushing system, a related method and device and electronic equipment. The system determines interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods through a server; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of a user; pushing related information of user preference service to the intelligent sound box of the user in a user preference time period according to a user preference period; and the intelligent sound box displays the message, and the user interacts with the intelligent sound box according to the message. By adopting the processing mode, the speaker message delivery accuracy can be effectively improved, the message delivery conversion rate is improved, and the message delivery efficiency is improved, so that the user experience and the user activity are improved.

Description

Message pushing system, related method and device

Technical Field

The application relates to the technical field of data processing, in particular to a message pushing system, a message pushing method and a message pushing device, a user habit information determining method and a user habit information determining device, an intelligent sound box and electronic equipment.

Background

The intelligent sound box is a product of sound box upgrading, is a tool for household consumers to surf the internet by voice, such as song ordering, internet shopping or weather forecast knowing, and can also control intelligent household equipment, such as opening a curtain, setting the temperature of a refrigerator, heating a water heater in advance and the like. In order to improve user experience, the intelligent sound box system can regularly push some messages to the intelligent sound box user based on the user portrait, such as a health detection reminding message, an alarm clock reminding message and the like.

When a traditional user portrait usually depicts the behaviors of a user in some interactive fields, only simple period and frequency statistics can be achieved, such as the frequency of using an alarm clock in the past week, the average frequency of using the alarm clock once in the past week and the like. The simple statistical approach of multiple dimensions can basically meet the requirements in a general application scene, but if the user wants to be further understood, the user is unconscious to describe the very complex behavior pattern of the user. In many scenarios such as smart speaker message pushing, smart speaker user awakening, etc., there is a need for deep understanding of the user. The traditional method for pushing the smart sound box message is that a professional operator uses a processed crowd label (such as age, gender and other population attributes, and the simple behavior frequency statistics) to circle out a group of smart sound box users offline, and then the professional operator pushes the message to the group of users directionally through an online system.

However, in the process of implementing the present invention, the inventor finds that the prior art solution has at least the following problems: when determining the push user, the push time and the push content, the operator needs to judge according to the working experience of the operator, and meanwhile, the operator can only use simple crowd labels with the form of population attributes, behavior frequency statistics and the like when selecting the crowd, so that the operation mode of judging the relevance between the push message of the intelligent sound box and the user based on the simple labels seriously depends on manual work, not only greatly consumes manpower, but also is difficult to ensure the accuracy of the relevance relation, thereby influencing the message push quality, or disturbing the user too much, or missing important messages. In a big data era, extremely abundant interactive behavior data exist between a user and an intelligent sound box system, and how to utilize massive user behavior data to systematically mine the relevance between an intelligent sound box push message and the user so as to improve the message push accuracy is a problem which needs to be solved urgently by technical personnel in the field.

Disclosure of Invention

The application provides a message push system to solve the lower problem of intelligence audio amplifier message propelling movement precision that prior art exists. The application further provides a message pushing method and device, a user habit information determining method and device, an intelligent sound box and electronic equipment.

The application provides a message push system, including:

the server is used for determining interaction times data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; pushing relevant messages of user preference services to the intelligent sound box of the user in the user preference period according to the habit information;

and the intelligent sound box is used for receiving the related message and displaying the related message.

The application also provides a method for determining user habit information, which comprises the following steps:

determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods;

determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data;

and determining a user preference period of the target corresponding relation, and forming intelligent sound box service use habit information of the user according to the target corresponding relation and the user preference period.

Optionally, the method further includes:

and if the habit is a healthy habit, pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the user preference period.

Optionally, the method further includes:

and if the habit is an unhealthy habit, sending health prompt information to the intelligent sound box of the user.

The application also provides a message pushing method, which comprises the following steps:

determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user;

and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

Optionally, the data dimension reduction algorithm includes: a principal component analysis algorithm;

and determining principal component information of the corresponding relation set as the target corresponding relation according to the interaction frequency data through a principal component analysis algorithm.

Optionally, the user preference period is determined at least according to the interaction number data.

Optionally, if the user does not adopt the service use habit to interact with the smart sound box, the user pushes related information to the smart sound box of the user.

Optionally, the step of determining the service usage habits is performed on a date associated with a date type.

Optionally, the date types include: holidays;

the dates related to the date types comprise the period of the holidays, the period after the holidays are ended and the period before the holidays are started.

Optionally, the method further includes:

determining a date range type;

and determining the historical time period according to the date range type.

Optionally, the date range types include:

weekday type, weekend type, holiday type.

Optionally, the determining the type of the date range includes:

determining a message delivery conversion rate;

and determining the date range type at least according to the message launching conversion rate.

Optionally, the method further includes:

receiving a user interaction request sent by the intelligent sound box;

if the interactive request is a request caused by the message pushed by the method, marking the message as an effective message;

the determining the message placement conversion rate comprises:

and determining the message launching conversion rate according to the effective message.

Optionally, the time period includes:

and a time period formed by taking the preset number of hours as the length of the time period.

Optionally, the service includes:

time inquiry service, weather inquiry service, alarm clock reminding service, music playing service.

receiving a message which is pushed by a server and is associated with the use habit of a user on the intelligent sound box service;

displaying the message so that a user can conveniently interact with the intelligent sound box according to the message;

wherein the message is determined in the following manner: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior mode based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of user preference service to the intelligent sound box of the user in the user preference time period according to the habit information in the user preference period.

The present application further provides a message pushing apparatus, including:

the data statistics unit is used for determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods;

the user behavior mode determining unit is used for determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data through a user behavior mode determining model based on a data dimension reduction algorithm, determining a user preference period of the target corresponding relation, and forming service use habit information of the user;

and the message pushing unit is used for pushing related messages of the user preference service to the intelligent loudspeaker box of the user in the user preference time period according to the habit information in the user preference period.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the message pushing method, wherein the following steps are executed after the device is powered on and the program of the method is run by the processor: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior mode based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of user preference service to the intelligent sound box of the user in the user preference time period according to the habit information in the user preference period.

the message receiving unit is used for receiving a message which is pushed by the server and is associated with the use habit of the user on the intelligent sound box service;

the message display unit is used for displaying the message so that a user can conveniently interact with the intelligent sound box according to the message;

The application further provides an intelligent sound box, include:

a processor; and

a memory for storing a program for implementing the message pushing method, wherein the following steps are executed after the device is powered on and the program of the method is run by the processor: receiving a message which is pushed by a server and is associated with the use habit of a user on the intelligent sound box service; displaying the message so that a user can conveniently interact with the intelligent sound box according to the message; wherein the message is determined in the following manner: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent loudspeaker box of the user in the user preference time period according to the habit information in the user preference period.

The present application further provides a user habit information determining apparatus, including:

the target corresponding relation determining unit is used for determining a target corresponding relation between a user preference time period and user preference services from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data through a user behavior mode determining model based on a data dimension reduction algorithm;

and the period determining unit is used for determining a user preference period of the target corresponding relation, and the target corresponding relation and the user preference period form the intelligent sound box service use habit information of the user.

The present application further provides an electronic device, comprising:

a processor; and

a memory for storing a program for implementing the user habit information determining method, wherein the following steps are performed after the device is powered on and the program for the method is run by the processor: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data; and determining a user preference period of the target corresponding relation, and forming intelligent sound box service use habit information of the user according to the target corresponding relation and the user preference period.

The present application also provides a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform the various methods described above.

The present application also provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the various methods described above.

Compared with the prior art, the method has the following advantages:

according to the message pushing system provided by the embodiment of the application, the interaction frequency data of a user for the services of a plurality of intelligent sound boxes in a plurality of historical time periods are determined through the server according to the historical interaction behavior data between the user and the intelligent sound boxes; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; pushing relevant messages of user preference services to the intelligent sound box of the user in a user preference time period according to the habit information in the user preference period; the smart sound box displays the message so that a user can conveniently interact with the smart sound box according to the message; by the processing mode, the interactive mode information between deep users and the intelligent sound box is mined according to the user behavior data, and the mode directly describes the relationship among the users, the message delivery field (also called sound box service or skill) and the delivery time, so that the situation that the users rely on simple behavior frequency statistical data to make message delivery decision can be avoided; therefore, the intelligent loudspeaker box message delivery accuracy can be effectively improved, and the interaction between the user and the intelligent loudspeaker box can be accurately awakened, so that the message delivery conversion rate is improved, and the user experience and the user activity are improved. In addition, the processing mode enables the interaction mode information between the user and the intelligent sound box to be automatically determined, and the information is automatically pushed to the user, so that the information putting efficiency can be effectively improved, and the human resource cost is saved.

Drawings

Fig. 1 is a schematic diagram of an embodiment of a message push system provided in the present application;

fig. 2 is a schematic view of a scenario of an embodiment of a message push system provided in the present application;

fig. 3 is a schematic device interaction diagram of an embodiment of a message push system provided in the present application;

fig. 4 is a schematic diagram of a user behavior pattern determination process of an embodiment of a message push system provided in the present application;

fig. 5 is a specific schematic diagram of a user behavior pattern determination process according to an embodiment of a message push system provided in the present application.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit of this application and is therefore not limited to the specific implementations disclosed below.

The application provides a message pushing system, a message pushing method and a message pushing device, a user habit information determining method and a user habit information determining device, an intelligent sound box and electronic equipment. Each of the schemes is described in detail in the following examples.

First embodiment

Please refer to fig. 1, which is a diagram illustrating a message push system according to an embodiment of the present application. In the embodiment, the system comprises a server 1 and a smart sound box 2.

The server 1 may be a server deployed on a cloud server, or may be a server dedicated to implementing a message push service for a smart speaker user, and may be deployed in a data center. The server may be a cluster server or a single server.

The smart speaker 2 may be a tool for a household consumer to surf the internet by voice, such as ordering songs, shopping on the internet, or knowing weather forecast, and may also control smart home devices, such as opening curtains, setting the temperature of a refrigerator, raising the temperature of a water heater in advance, and the like.

Please refer to fig. 2, which is a schematic view of a scenario of the message push system of the present application. The server 1 and the intelligent sound box 2 can be connected through a network, for example, the intelligent sound box and the server can be networked through WIFI and the like. As shown in fig. 2, the smart speaker user a queries the day information of the next day through the smart speaker on a daily working day, so as to be ready for work (e.g., adding clothes according to the air temperature, etc.); then, during holidays such as spring festival and the like, the user A does not need to go to work, so that the weather information of the next day does not need to be inquired through the intelligent sound box; next, after the holiday is over, because the daily life state is not adjusted in time, the user a may forget to query the weather information of the next day through the smart speaker, which may result in that the user cannot prepare for work on the next day in time under the prior art. By the system provided by the embodiment of the application, the server 2 can automatically mine the interaction mode information between the user A and the intelligent sound box A at a deeper level according to the historical interaction behavior data between the user A and the intelligent sound box A in a period of time (such as 7 days/14 days) before the holiday when the holiday is finished, the mode directly describes the relationship among the user, the message release field and the release time, for example, a message of inquiring whether the user A inquires weather information is pushed to the intelligent sound box A in the period of time before and after the working day, and the intelligent sound box A plays the message to the user A so as to awaken the user A to inquire the weather information of the next day.

For another example, in fig. 2, a smart speaker user B often turns on a television through the smart speaker B at daily night to watch news simulcast, and after going back on vacation in summer, the smart speaker user B may forget to turn on the television through the smart speaker B to watch news simulcast.

Please refer to fig. 3, which is a schematic diagram of an interaction of devices in an embodiment of a message push system according to the present application. In this embodiment, the server is configured to determine interaction time data of the user for the plurality of smart sound box services in a plurality of historical time periods; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; pushing related information of user preference service to the intelligent sound box of the user in the user preference time period according to the habit information; and the intelligent sound box is used for receiving the related message and displaying the related message.

The user comprises a user of the intelligent sound box. The user and the smart speaker 2 can interact with each other with various speaker services, such as a time query service (the user queries time information from the smart speaker), a weather query service (the user queries weather information from the smart speaker), an alarm clock reminding service (the user sets alarm clock wake-up time through the smart speaker), a music playing service (the user instructs the smart speaker to play music), and so on.

The user behavior mode determining model comprises the steps of determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent loudspeaker box services according to the interaction times data, determining a user preference period of the target corresponding relation, and forming a model of the service use habit information of the user.

And determining the target corresponding relation between the user preference time period and the user preference service from the corresponding relation set between the plurality of historical time periods and the plurality of intelligent sound box services according to the interaction times data through the data dimension reduction algorithm. That is to say, the data dimension reduction algorithm may select the target correspondence from the correspondence set, and the number of the target correspondence may be much smaller than the number of the relationships in the correspondence set.

The data dimension reduction algorithm may be a principal component analysis algorithm. In one example, the principal component information of the corresponding relation set is determined as the target corresponding relation according to the interaction time data through a principal component analysis algorithm.

In specific implementation, the server side can determine behavior frequency statistical data (namely, the interaction frequency data) of the user for each interaction field (namely, the sound box service and also called sound box skill) in each time period within a target date range according to historical interaction behavior data between the user and the intelligent sound box; determining a user behavior mode (namely the service use habit information) according to the behavior number statistical data; pushing a message associated with the user behavior pattern (namely, a message related to the user preference service, such as health detection reminding information, alarm clock reminding information, weather inquiry information and the like) to the intelligent sound box corresponding to the user; and the intelligent sound box is used for receiving the message and displaying the message so as to facilitate the interaction of the user and the intelligent sound box according to the message.

The historical interactive behavior data comprises instruction information which is historically issued to the intelligent sound box by a user, such as when the user issues interactive instructions of which field (sound box service) to the intelligent sound box. Table 1 shows the historical interaction behavior data recorded in the server 2 of the present embodiment.

TABLE 1 historical interactive behavior data between a user and a smart speaker

As shown in table 1, the server 2 collects and records historical interactive behavior data of multiple smart speaker users. According to the system provided by the embodiment of the application, based on the historical interactive behavior data shown in the table 1, the service end 2 can mine deeper interaction mode information between the user and the intelligent sound box from the historical interactive behavior data of the user aiming at each user, and the mode directly describes the relationship among the user, the message release field and the release time.

The target date range may be a date range from the user to all interactions between the smart sound box, for example, from 2018/10/5 to a date of mining the user behavior pattern; the target date range may also be a time range of recent interaction between the user and the smart speaker, such as a last work day of one week, a last work day of two weeks, a last holiday period, a last day of one month, and the like.

The time period may be a small time period in one day with preset hours as intervals, and if every 4 hours in one day is taken as a time period, one day includes 6 time periods, which may be 0-4 o 'clock, 4-8 o' clock, 8-12 o 'clock, 12-16 o' clock, 16-20 o 'clock, and 20-24 o' clock. In specific implementation, the time period may be one time period per hour, or one time period per 8 hours, and so on.

The behavior frequency statistical data may be behavior frequency data of the user for each interaction field in each time period of a day. Table 2 shows the statistics data of the behavior times recorded in the server 2 of this embodiment.

TABLE 2 statistical data of behavior times of users

As can be seen from table 2, the server 2 may count, for each user, behavior frequency data of the user for each interaction domain in each time period of each day.

Please refer to fig. 4, which is a schematic diagram illustrating a user behavior pattern (i.e. service usage habit information of the user) determining process according to an embodiment of the message push system of the present application. In this embodiment, the process of the server 2 determining the statistical data of the behavior times is as follows.

Step one, according to the behavior log of each user, respectively constructing a high-dimensional user behavior statistical matrix for each user, wherein the user log can be specifically designed to relate to m fields, and the form is as follows: weather, time, alarm clock, music, etc., and divide 24 hours per day into n sub-periods, can divide by fixed hour interval (such as 4 hour interval), subdivide the field and time quantum and carry out the statistics of action number, namely: the access times of the user in m fields and n time periods respectively. As can be seen from FIG. 4, the high-dimensional matrix constructed by this step may include three dimensions: the data of the three dimensions can be used as index values, and each element value on the three-dimensional matrix can be specific behavior time data.

Second, date window (the target date range) selection: the statistical data are arranged in sequence in a certain type of date window to obtain the user behavior statistical matrix shown in fig. 4. In fig. 4, the date window type is the last three days and the matrix includes statistics of the number of behaviors for the last three days.

Thirdly, obtaining principal components in the user behavior statistical matrix through a dimension reduction module, namely: the user behavior patterns and pattern scores are selected as candidates. That is, the present embodiment determines a candidate user behavior pattern and a pattern score according to the behavior number statistical data.

When the method is specifically implemented, the behavior mode and the mode score of the candidate user can be determined according to the behavior frequency statistical data through a principal component analysis algorithm; other dimension reduction techniques may also be employed to determine the candidate user behavior patterns and pattern scores.

In this embodiment, the principal component matrix is determined by the following process, and the principal component may be a matrix under the most dominant date index. 1) Assuming that the size of a date window is d, selecting the date window to obtain a three-dimensional user behavior statistical matrix of m X n X d, wherein m represents a field dimension (namely a sound box service dimension), n represents a time period dimension, d represents a date dimension, behavior frequency statistical data is data of one point under the three coordinates, modifying the dimension of the matrix, and flattening the first two dimensions into one dimension to form a two-dimensional matrix X of mn X d; 2) normalizing the preprocessing matrix X, X ═ (X-min (X))/(max (X) -min (X)); 3) performing a process of de-column averaging on the matrix X, X ═ X-mean (X); 4) and (3) reducing the dimensionality: the dimensionality reduction is performed by using a PCA method, and singular value SVD decomposition is performed on the X, so that U, S, V, are SVD (X), wherein each column of U is a component, only the most important component is reserved, the first column of main component of U is c (norm is 1), the U is converted into a matrix A of mxn, the matrix A is a main component matrix, and the mode corresponding to each element position in the matrix is a candidate user behavior mode (namely a candidate target corresponding relation).

And step four, extracting a main mode from the main component, namely: user behavior patterns (i.e., target correspondence). The user behavior patterns include: and the corresponding relation between the interactive field and the time period. The main mode, i.e. the positions with the element values in the main component in front, the indexes of the positions are the domain and the time period respectively, and the value of the position, i.e. the weight of the mode (the mode score). The present embodiment determines the user behavior pattern according to the pattern score.

In particular, the user behavior pattern may be determined according to the pattern score and a score threshold. The server 2 may determine the score threshold according to the user behavior pattern quantity threshold. For example, if there are k (e.g., k is determined to be 4 manually) patterns at most, the score threshold may be 1/(k × k), and the coordinates (i, j) of the element a in the principal component matrix that is greater than or equal to the threshold are selected, where i represents a domain and j represents a time period, so as to obtain the domain and the time period corresponding to the user behavior pattern.

And fifthly, if the user does not adopt the user behavior mode to interact with the intelligent sound box, pushing a message associated with the user behavior mode to the intelligent sound box corresponding to the user.

The message associated with the user behavior pattern may be system reply information triggered by the user behavior pattern, or prompt information for inquiring whether the user interacts with the sound box by using the user behavior pattern.

In specific implementation, the corresponding message delivery can be performed according to the field where the user behavior pattern is located. For example, if the user a mode is obtained as above: weather period 8: 00-10: 00, although the user interacts with the system in the time period on a certain day, the user does not visit the weather, and then the weather information can be put into the sound box APP of the user sound box or the mobile phone end. Alternatively, when detecting that the pattern in which the user should appear does not appear, the user is appropriately wakened up, as shown in fig. 2, by asking the speaker whether to inquire about the weather of the next day.

In one example, the server 2 performs the step of determining the service usage habit on a date associated with the date type. The date types include, but are not limited to: holidays, a certain fixed date, etc.; the dates associated with the date type include, but are not limited to, during holidays, after holidays end, and before holidays begin. As shown in fig. 2, the server 2 may mine the user behavior pattern after the vacation is over.

In one example, the server determines the user preference period at least according to the interaction time data. In specific implementation, the user preference period of the target corresponding relation can be determined in a time window sliding mode. For example, the target correspondence of the user a includes: weather period 8: 00-10: 00, preference period is "day before workday", e.g. sunday to thursday.

In addition, in order to continuously mine the user behavior pattern, in the specific implementation, the server 2 may further be configured to slide the date window of the second step, so as to obtain a time interval of the user behavior pattern duration, which may include a forming time of the user behavior pattern, a last occurrence time, and the like.

For example, in a second step of selecting a weekday window type, with a window size of 10 days, the target time range includes: from 10 working days, i.e., 6/2020/1/17/2020, a user behavior pattern can be determined based on the statistics of behavior times for the 10 working days, and from this pattern it can be determined when to push what kind of message to the user, and at this time the time of formation of the user behavior pattern can also be determined. And then, sliding the time window one day backward to obtain the next time window from 1 month and 7 days in 2020 to 1 month and 20 days in 2020, and determining whether the user behavior pattern appearing in the previous day appears again according to the user behavior statistical matrix in the window.

Please refer to fig. 5, which is a specific diagram illustrating a user behavior pattern determining process according to an embodiment of the message pushing system of the present application. In this embodiment, in order to dig a user behavior pattern with higher accuracy according to the behavior frequency statistical data in the reasonable target date range, the server 2 may be further configured to obtain a reconstruction matrix of the user behavior statistical matrix through the dimension reduction module; and constructing a reconstructed image according to the reconstruction matrix, and displaying the image to system related personnel, wherein the image can comprise behavior times statistical data which embody a user behavior mode. The pattern type, including the date window type, may be determined using an annotated method. And (4) marking, namely randomly sampling a batch of samples for the reconstructed matrix, and directly displaying visually, so that a marking person marks which obvious behavior types, and the matrix has exploration properties.

The dimension of the reconstruction matrix can be the same as that of the user behavior statistical matrix, but only comprises partial data in the user behavior statistical matrix, and the partial data can be behavior time statistical data related to the determined user behavior pattern, that is, the reconstruction matrix can comprise data capable of reflecting the user behavior pattern.

Furthermore, the related personnel of the system can also give out marking information whether the mode is available or not for the mined user behavior mode according to the image. If the system related personnel think that the excavated mode has higher message putting conversion rate according to the data in the image, the system can be put into use; otherwise, if the message pushed by the system cannot reach a certain message launching conversion rate, the parameters of the system can be adjusted, and the aim of the adjustment is as follows: the message pushed by the system reaches a certain message launching conversion rate.

In specific implementation, a clustering method can be adopted to determine the type of the mode, including the type of the date window. If the clustering can be performed into X types, the reconstructed matrixes can be selected, each matrix is used as a sample, after all samples are obtained, a certain clustering algorithm is selected to obtain X clustering centers, the clustering centers are visualized, behavior habits of each type of field can be analyzed, and then the most typical X type behavior types can be extracted.

To sum up, the idea of the system provided in the embodiment of the present application may be to describe the complicated historical behaviors of the user by using a high-dimensional representation vector (a user behavior statistical matrix), perform dimension reduction on the representation vector by using a dimension reduction technology (for example, a three-dimensional user behavior statistical matrix is converted into a two-dimensional user behavior statistical matrix), reconstruct (for example, the above reconstruction matrix), visualize the representation vector to explore possible pattern types (for example, a date window type is a last working day of a week, and the like), and finally extract the user behavior pattern from the principal component.

In one example, the server 2 is further configured to determine a date range type (i.e., the date window type); determining the historical time period based at least on the date range type. The date range types include, but are not limited to, the following: weekday type, weekend type, holiday type.

In specific implementation, the date range type and the target date range can be firstly determined; then, the historical time period is determined according to the target date range.

In this embodiment, some date range types may be set manually, and then the date range types may be adjusted according to the message drop conversion rate. The message delivery conversion rate may be a ratio between the number of system push messages causing interaction between the user and the smart speaker and the total number of system push messages.

In specific implementation, the server 2 is further configured to receive a user interaction request sent by the smart sound box; if the interactive request is a request caused by the message pushed by the method, marking the message as an effective message; the server 2 is specifically configured to determine a message delivery conversion rate according to the effective message and the total number of messages. In addition, the server 2 is further configured to generate the historical interaction behavior data according to user instruction data carried by the interaction request.

After determining the date range type, the target date range may be determined based on the date range type. In specific implementation, a plurality of date range types can be determined, so that a plurality of target date ranges are determined, and then user behavior patterns are mined respectively according to the behavior frequency statistical data of each target date range. In addition, different target date ranges can be determined for different domains, so that more accurate user behavior patterns can be mined.

The system provided by this embodiment, by adopting the processing manner shown in fig. 5, determines a more reasonable target date range, and may avoid that the range is too large or too small, if the range is too large, some user behavior patterns that have failed may be excessively mined, and if the range is too small, some actually valid user behavior patterns may be missed. After a more reasonable date range is determined, a more accurate user behavior mode can be mined according to the behavior frequency statistical data in the range; therefore, the message pushing precision can be effectively improved.

As can be seen from the above embodiments, in the message pushing system provided in the embodiment of the application, the interaction frequency data of the user for the services of the plurality of intelligent sound boxes in a plurality of historical time periods is determined according to the historical interaction behavior data between the user and the intelligent sound boxes by the server; determining a model through a user behavior mode based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user; pushing related information of user preference service to the intelligent sound box of the user in a user preference time period according to the habit information in the user preference period; the smart sound box displays the message so that a user can conveniently interact with the smart sound box according to the message; by the processing mode, the interaction mode information between deep users and the intelligent sound box is mined according to the user behavior data, and the mode directly describes the relationship among the users, the message delivery field (also called sound box service or skill) and the delivery time, so that the situation that the users rely on simple behavior frequency statistical data to carry out message delivery decision artificially can be avoided; therefore, the intelligent sound box message delivery accuracy can be effectively improved, the interaction between the user and the intelligent sound box can be accurately awakened, the message delivery conversion rate is improved, and the user experience and the user activity are improved. In addition, the processing mode enables the interaction mode information between the user and the intelligent sound box to be automatically determined, and the information is automatically pushed to the user, so that the information putting efficiency can be effectively improved, and the human resource cost is saved.

Second embodiment

The embodiment of the application also provides a message pushing method. The execution subject of the method may be a server, or may be any device capable of executing the method. In this embodiment, the method includes the steps of:

step 1: determining interaction times data of a user for a plurality of intelligent loudspeaker box services in a plurality of historical time periods;

step 2: determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data, and determining a user preference period of the target corresponding relation to form service use habit information of the user;

and step 3: and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

In one example, the data dimension reduction algorithm comprises: a principal component analysis algorithm; and determining principal component information of the corresponding relation set as the target corresponding relation according to the interaction frequency data through a principal component analysis algorithm.

In one example, the user preference period is determined based at least on the interaction times data.

In one example, if the user does not adopt the service use habit to interact with the smart speaker, pushing a relevant message to the smart speaker of the user.

In one example, the step of determining the service usage habits is performed on a date associated with a date type.

The date types, including but not limited to: holidays; the dates associated with the date type include, but are not limited to, during holidays, after holidays end, and before holidays begin.

In one example, the method may further comprise the steps of: determining a date range type; determining the historical time period according to the date range type.

The date range types include, but are not limited to: weekday type, weekend type, holiday type.

In one example, the determining the date range type may include the sub-steps of: determining a message putting conversion rate; and determining the date range type at least according to the message launching conversion rate.

In one example, the method may further comprise the steps of: receiving a user interaction request sent by the intelligent sound box; if the interactive request is a request caused by the message pushed by the method, marking the message as an effective message; correspondingly, the determining of the message drop conversion rate may be performed as follows: and determining the message launching conversion rate according to the effective message.

The time period includes, but is not limited to: and a time period formed by taking the preset number of hours as the length of the time period.

The services, including but not limited to: time inquiry service, weather inquiry service, alarm clock reminding service, music playing service.

Third embodiment

Corresponding to the message pushing method, the application also provides a message pushing device. Since the device embodiment is basically similar to the method embodiment one, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment. The device embodiments described below are merely illustrative.

The application provides a message push device, includes:

Fourth embodiment

Corresponding to the message pushing method, the application also provides an electronic device. Since the apparatus embodiment is substantially similar to the method embodiment one, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The device embodiments described below are merely illustrative.

The application provides an electronic device, including: a processor and a memory for storing a program for implementing the message pushing method, the device being powered on and executing the program of the method by the processor, the following steps being performed: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

Fifth embodiment

The embodiment of the application also provides a message pushing method. The execution subject of the method can be a smart sound box, and can also be any device capable of executing the method, such as a smart phone and the like. In this embodiment, the method includes the steps of:

step 1: receiving a message which is pushed by a server and is associated with the use habit of a user on the intelligent sound box service;

step 2: displaying the message so that a user can conveniently interact with the intelligent sound box according to the message;

Sixth embodiment

The application provides a message push device, includes:

Seventh embodiment

Corresponding to the message pushing method, the application also provides an intelligent sound box. Since the apparatus embodiment is substantially similar to the method embodiment one, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The device embodiments described below are merely illustrative.

The application provides an intelligent sound box, include: a processor and a memory for storing a program for implementing the message pushing method, the device being powered on and executing the program of the method by the processor, the following steps being performed: receiving a message which is pushed by a server and is associated with the use habit of a user on the intelligent sound box service; displaying the message so that a user can conveniently interact with the intelligent sound box according to the message; wherein the message is determined as follows: determining interaction times data of a user for a plurality of intelligent loudspeaker box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

Eighth embodiment

The embodiment of the application also provides a flow chart of the user habit information determining method. The execution subject of the method may be a server, or may be any device capable of executing the method. In this embodiment, the method includes the steps of:

step 2: determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data;

and step 3: and determining a user preference period of the target corresponding relation, and forming intelligent sound box service use habit information of the user according to the target corresponding relation and the user preference period.

In one example, the method may further comprise the steps of: and if the habit is a healthy habit, pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the user preference period.

In one example, the method may further comprise the steps of: and if the habit is an unhealthy habit, sending health prompt information to the intelligent sound box of the user.

By adopting the processing mode, the service related information can be pushed to the user according to the using habit of the healthy user for the intelligent sound box, and the unhealthy using habit of the user can be reminded; therefore, the user experience can be effectively improved.

When the method is specifically implemented, a health habit can be preset, for example, the habit is the habit when a certain sound box service is used from 5 am to 11 pm, and if the target corresponding relation of the user accords with the preset health habit, the related message of the user preference service related to the target corresponding relation can be pushed to the sound box of the user; if the target corresponding relation of the user does not accord with the preset health habit, the health prompt information of the user preference service related to the target corresponding relation can be pushed to the sound box of the user.

As can be seen from the above embodiments, the method for determining user habit information provided by the embodiments of the present application determines interaction frequency data of a user for a plurality of smart sound box services in a plurality of historical time periods; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data; determining a user preference period of a target corresponding relation, and forming intelligent sound box service use habit information of the user according to the target corresponding relation and the user preference period; by the processing mode, the interaction mode information between the deep-level user and the intelligent sound box is mined according to the user behavior data, and the mode directly describes the relationship among the user, the message delivery field (also called sound box service or skill) and the delivery time, so that the situation that the user relies on simple behavior frequency statistical data to make a message delivery decision can be avoided; therefore, the intelligent sound box message delivery accuracy can be effectively improved, the interaction between the user and the intelligent sound box is ensured to be accurately awakened, and the message delivery conversion rate is improved.

Ninth embodiment

Corresponding to the user habit information determining method, the application also provides a user habit information determining device. Since the device embodiment is basically similar to the first method embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment. The device embodiments described below are merely illustrative.

The application provides a user habit information determination device, includes:

Tenth embodiment

Corresponding to the user behavior mode method, the application also provides the electronic equipment. Since the apparatus embodiment is substantially similar to the method embodiment one, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The device embodiments described below are merely illustrative.

The application provides an electronic device, including: a processor and a memory for storing a program for implementing a method for determining user habit information, the device being powered on and running the program of the method via the processor for performing the following steps: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data; and determining a user preference period of the target corresponding relation, and forming the intelligent sound box service use habit information of the user by the target corresponding relation and the user preference period.

Although the present application has been described with reference to the preferred embodiments, it is not intended to limit the present application, and any person skilled in the art can make variations and modifications without departing from the spirit and scope of the present application, therefore, the scope of the present application should be limited by the scope of the claims.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

1. Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include non-transitory computer readable media (transient media), such as modulated data signals and carrier waves.

2. As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

Claims

1. A message push system, comprising:

the server is used for determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; pushing related information of user preference service to the intelligent sound box of the user in a user preference time period according to the habit information in the user preference period;

2. A method for determining user habit information is characterized by comprising the following steps:

3. The method of claim 2, further comprising:

4. The method of claim 2, further comprising:

5. A message pushing method, comprising:

6. The method of claim 5,

the data dimension reduction algorithm comprises the following steps: a principal component analysis algorithm;

7. The method of claim 5,

and determining the user preference period at least according to the interaction frequency data.

8. The method of claim 5,

and if the user does not adopt the service use habit to interact with the intelligent sound box, pushing related information to the intelligent sound box of the user.

9. The method of claim 5,

the step of determining the usage habits of said service is performed on a date associated with a date type.

10. The method of claim 9,

the date types include: holidays;

11. The method of claim 5, further comprising:

determining a date range type;

and determining the historical time period according to the date range type.

12. The method of claim 11, wherein the date range types include:

weekday type, weekend type, holiday type.

13. The method of claim 11, wherein the determining a date range type comprises:

determining a message delivery conversion rate;

14. The method of claim 13, further comprising:

receiving a user interaction request sent by the intelligent sound box;

the determining the message placement conversion rate comprises:

15. The method of claim 5, wherein the time period comprises:

16. The method of claim 5, wherein the service comprises:

17. A message pushing method, comprising:

wherein the message is determined in the following manner: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

18. A message push apparatus, comprising:

19. An electronic device, comprising:

a processor; and

a memory for storing a program for implementing the message pushing method, wherein the following steps are executed after the device is powered on and the program of the method is run by the processor: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

20. A message push apparatus, comprising:

21. An intelligent sound box, comprising:

a processor; and

a memory for storing a program for implementing the message pushing method, wherein the following steps are executed after the device is powered on and the program of the method is run by the processor: receiving a message which is pushed by a server and is associated with the use habit of a user on the intelligent sound box service; displaying the message so that a user can conveniently interact with the intelligent sound box according to the message; wherein the message is determined in the following manner: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data and a user behavior mode determination model based on a data dimension reduction algorithm, and determining a user preference period of the target corresponding relation to form service use habit information of the user; and pushing related information of the user preference service to the intelligent sound box of the user in the user preference time period according to the habit information.

22. A user habit information determining apparatus, comprising:

23. An electronic device, comprising:

a processor; and

a memory for storing a program for implementing the user habit information determining method, wherein the following steps are performed after the device is powered on and the program for the method is run by the processor: determining interaction frequency data of a user for a plurality of intelligent sound box services in a plurality of historical time periods; determining a model through a user behavior mode determination model based on a data dimension reduction algorithm, and determining a target corresponding relation between a user preference time period and a user preference service from a corresponding relation set between a plurality of historical time periods and a plurality of intelligent sound box services according to the interaction times data; and determining a user preference period of the target corresponding relation, and forming intelligent sound box service use habit information of the user according to the target corresponding relation and the user preference period.