Service push method, service push device, storage medium and electronic device

Abstract

The present invention discloses a service push method, a service push device, a storage medium and an electronic device, wherein the service push method comprises: obtaining the target user's own state data and the scene data; constructing the user portrait of the target user according to the own state data and the scene data; obtaining a similar user group according to the user portrait; predicting the target user's preference value for the target service according to the preference information of the similar user group for the target service based on a Bayesian model; and pushing the target service according to the preference value. The method can effectively improve the accuracy of pushing the target service of interest to the target user, and provide the user with a good service experience.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to a service push method, a service push device, a storage medium and an electronic device. Background Art

[0002] With the development of Internet technology, the application of business push is becoming more and more extensive. Business push is a technology that provides users with services that may suit their interests and hobbies based on the content characteristics of the service itself and the user's experience behavior.

[0003] In the related art, the algorithm for business push is usually based on historical data, similar user habits, preference habits and other factors.

[0004] However, the existing service push algorithms do not fully consider the factors affecting the push services and cannot provide users with a good service experience. Summary of the invention

[0005] Based on this, it is necessary to provide a service push method, a service push device, a storage medium and an electronic device to address the above technical issues.

[0006] The present invention provides a method for pushing a service, which includes:

[0007] Obtain the target user's own status data and the scene data they are in;

[0008] Building a user profile of the target user according to the own state data and the scenario data;

[0009] Obtaining a similar user group according to the user portrait;

[0010] Based on the Bayesian model, predicting the preference value of the target user for the target service according to the preference information of the similar user group for the target service;

[0011] The target service is pushed according to the preference value.

[0012] In some embodiments, based on a Bayesian model, predicting the target user's preference value for the target service according to the preference information of the similar user group for the target service includes:

[0013] Based on the Bayesian model, similar users in the similar user group that meet each tag of the user portrait are obtained, and a first repetition probability and a first non-repetition probability of similar users corresponding to each tag experiencing the target business are obtained, wherein when constructing the user portrait of the target user, the target user's own state data and the scene data in which the target user is located form the tag of the user portrait;

[0014] Obtaining a second repetition probability and a second non-repetition probability of historical experience of the similar user group for the target service;

[0015] Obtaining a total repetition probability according to the first repetition probability and the second repetition probability, and obtaining a total non-repetition probability according to the second non-repetition probability and the second non-repetition probability;

[0016] A cloud push index is obtained according to the total duplication probability and the total non-duplication probability to serve as a preference value of the target user for the target service.

[0017] In some embodiments, pushing the target service according to the preference value includes:

[0018] If it is determined that the preference value is greater than or equal to the preference threshold, the target service is pushed to the target user;

[0019] If it is determined that the preference value is less than the preference threshold, the target service is not pushed to the target user.

[0020] In some embodiments, the service push method further includes:

[0021] Obtain the number of times the target user has experienced the target service pushed this time and the degree of completion of each experience;

[0022] Determining a user feedback index of the target user for the target service pushed this time according to the number of experiences and the degree of completion of each experience;

[0023] The target service to be pushed next time is adjusted according to the user feedback index.

[0024] In some embodiments, adjusting the target service to be pushed next time according to the user feedback index includes:

[0025] Obtaining the first target service whose user feedback index is less than the feedback index threshold among the target services pushed this time;

[0026] The first target service is removed from the target service whitelist to be pushed next time, and the second target service in the target service pushed this time except the first target service is kept in the target service whitelist to be pushed next time.

[0027] In some embodiments, the service push method further includes:

[0028] According to the preference value predicted for the next push, determine whether there is a new target business other than the target business pushed this time;

[0029] If so, the new target service is added to the target service whitelist to be pushed next time.

[0030] In some embodiments, the service push method further includes:

[0031] Obtaining a first update push index according to a user feedback index and a preference value corresponding to the second target service, and obtaining a second update push index according to a user feedback index and a preference value corresponding to the new target service;

[0032] The second target service and the new target service are sorted according to the first update push index and the second update push index, and used as the final target service whitelist for the next push.

[0033] The embodiment of the present application provides a service push device, the service push device comprising:

[0034] The first acquisition module is used to acquire the target user's own state data and the scene data in which the target user is located;

[0035] A construction module, used to construct a user portrait of the target user according to the own state data and the scenario data;

[0036] A second acquisition module is used to obtain a similar user group according to the user portrait;

[0037] A prediction module, configured to predict the target user's preference value for the target service based on the preference information of the similar user group for the target service based on a Bayesian model;

[0038] A push module is used to push the target service according to the preference value.

[0039] An embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the service push method provided in any embodiment of the present application are implemented.

[0040] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the service recommendation method provided in any embodiment of the present application are implemented.

[0041] The service push method, service push device, storage medium and electronic device provided in the embodiments of the present application comprehensively consider the target user's own status data and the scene data in which he is located, such as the target user's age, gender, identity, time, city, interests and hobbies, etc., to construct a user portrait of the target user, and based on the similar user group obtained according to the user portrait, a Bayesian model is used to predict the target user's preference value for the target service for the user portrait of the target user, so as to determine whether to push the target service to the target user through the preference value. In this way, the present application considers the influencing factors of the target user's preference for the target service from many aspects, and combines the similar user group and the Bayesian model to effectively improve the accuracy of pushing the target service of interest to the target user, thereby providing the user with a good service experience.

[0042] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0044] Figure 1 A schematic diagram of a process flow of a service push method in an embodiment;

[0045] Figure 2 It is a flowchart of a service push method in another embodiment;

[0046] Figure 3 It is a structural block diagram of a service push device in an embodiment;

[0047] Figure 4 FIG. 4 is a structural block diagram of an electronic device in an embodiment.

[0048] Reference numerals:

[0049] Service push device 10; electronic device 20;

[0050] First acquisition module 1; construction module 2; second acquisition module 3; prediction module 4; push module 5; memory 6; processor 7. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] In one embodiment, Figure 1As shown, a service push method is provided. The embodiment of the present application takes the method applied to an electronic device as an example, wherein the electronic device may be, but is not limited to, a personal computer, a laptop, a smart phone, and a tablet computer. It is understandable that the method may also be applied to a server, and may also be applied to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes at least the following steps S1 to S5.

[0053] Step S1, obtaining the target user's own state data and the scene data in which the target user is located.

[0054] Among them, the self-status data is the status data related to the actual situation of the target user itself, for example, the self-status data at least includes data on user attributes, social attributes, behavioral habits, interest preferences and psychological attributes, such as the target user's age, gender, ethnicity, identity, interests, preferences, marital status, fertility status, exercise status, mood and social relationships. The scene data refers to the specific life scene of the target user in the current time and space, for example, the scene data can be data such as geographic location, date, time, weather and temperature.

[0055] In the embodiments, different users have different preferences for services and their preferences for the same service also vary. Taking music push as an example, users like to listen to light music when sleeping, youth music when taking the subway in the early morning, parents like to listen to nostalgic classics, and middle school students like to listen to fashionable songs, etc. In addition, for the same music, the user's preference decreases with the number of repeated listenings. Therefore, the present application considers the actual situation of the users from many aspects to determine the factors affecting the target users' service preferences, that is, to realize service push based on the acquired target users' own status data and the scene data in which they are located, which helps to more accurately understand the target users' demand for services and is conducive to more accurately digging out potential favorite services for users.

[0056] Specifically, the method of the embodiment of the present application is applied to an electronic device, and the target user's own status data and the scene data are collected through the information recorded and stored in the electronic device. For example, the geographical location of the target user can be calculated based on the GPS positioning, gyroscope and map set in the electronic device; the current time and date of the target user can be obtained based on the timetable and calendar set in the electronic device, so as to judge whether the date is a holiday and whether the current time is morning, noon or evening; the climate conditions of the city where the target user is located can be judged based on the weather information of the electronic device; the target user's age, gender, nationality, identity, whether he is married and whether he has children can be obtained based on the registration information and avatar stored in the electronic device; the target user's interests, preferences and social data can be obtained based on the target user's daily operations in the electronic device, such as Baidu or social situations. Therefore, the influencing factors when pushing services to the target user are comprehensively considered, so as to facilitate a more accurate understanding of the target user's demand for services, and to help users more accurately dig out potential favorite services, so as to achieve the effect of accurately pushing the target user's interested services.

[0057] Step S2, constructing a user profile of the target user based on its own status data and scenario data.

[0058] Among them, the user portrait is a fictitious representation of the target user created based on its own status data and scenario data. Through the user portrait, each specific data of the target user can be abstracted into a label, and these labels can be used to concretize the image of the target user, so as to facilitate the provision of targeted services to the target user. It is understandable that in the process of building a user portrait, it is necessary to pay attention to the richness, diversity, scientificity and real-time nature of the data, so as to ensure the perfection of the user portrait, so as to more accurately understand the target user's demand for the service and achieve the effect of accurately pushing the target user's interested services.

[0059] Table 1

[0060] Label user Name Zhang San age 25 City Shanghai nationality Chinese gender male identity student interest football weather sunny temperature 25～30 Do you have children? none Married? no Do you like opera? no Do you like JAZZ? yes Do you like rock? yes

[0061] Specifically, based on the collected own status data and scenario data, key portrait data is designed according to the needs of the push business. That is to say, different data needs to be extracted for different businesses and different types of users. For example, for music push, different users have different preferences for music. Therefore, it is necessary to consider the target user's activities, age, mood, time, interests, preferences, exercise status, mood, social relationships, geographic location, date, time, weather and temperature, etc., which obviously affect the user's business preference factors. There is no need to analyze unnecessary data. Therefore, the basic framework for depicting user portraits is constructed based on key portrait data related to the business, and the construction of user portraits is completed by purifying and enriching the portrait content. Table 1 shows the user portrait constructed for music push based on the target user's own status data and scenario data.

[0062] Step S3, obtaining a similar user group based on the user portrait.

[0063] Among them, the similar user group refers to a group collection composed of other users with similar label information as in the user portrait, and the similar user group will be updated as the user portrait changes.

[0064] Specifically, based on the data of other users in the big data and combined with the content of each label in the user portrait, it is determined that other users with similar label information form a similar user group. For example, the user portrait of the target user has label information such as "age 25 years old, city is Shanghai, interest is football, married, temperature is 25-30, weather is sunny, time is noon, and mood is good", then users with the same age of about 25 years old, city is Shanghai, interest is football, married, temperature is 25-30, weather is sunny, time is noon, and mood is good are regarded as similar users, and all similar users are classified into similar user groups.

[0065] Step S4: Based on the Bayesian model, the preference value of the target user for the target service is predicted according to the preference information of similar user groups for the target service.

[0066] Among them, the Bayesian model is a probability model, which is a graphical model based on probabilistic reasoning, and the commonly used Bayesian classification method is the naive Bayesian classification. With the development of Internet technology, there are many services in the network, such as music, search, video, news, advertising and other services. It can be understood that for music, search, video, news, advertising and other services, the methods provided in the embodiments of the present application can be applied. The preference value can be understood as a quantitative value of the target user's preference for the target service.

[0067] Specifically, without knowing the target user's preference for the target business, a Bayesian model is used to predict the target user's preference value for different businesses based on the user portrait of the target user based on similar user groups. In this way, we can try to find businesses that are closer to user behavior and in line with user tastes, so that we can filter out businesses that the target user does not like, and recommend businesses that the target user likes to achieve accurate push.

[0068] Step S5: Push the target service according to the preference value.

[0069] In an embodiment, the conditions of services that meet the tastes of target users can be pre-set according to actual conditions, and the preference value can be used to determine whether to push the target service to the target user. Specifically, if the preference value predicted for the target service does not meet the preset conditions, the target service is filtered and not pushed to the target user; conversely, if the preference value predicted for the target service meets the preset conditions, the target service is pushed to the target user. In this way, by judging the preference value obtained by the user portrait and the Bayesian model to determine whether to push the service, it is possible to effectively filter out the recommended services from thousands of services, and achieve the effect of accurately pushing the latest, hottest, and most favorite services to users, and effectively filter out the services that users are not interested in, and provide users with a good service experience.

[0070] In the above-mentioned service push method, the target user's own status data and the scenario data, such as the target user's age, gender, identity, time, city, interests and hobbies, are comprehensively considered to construct a user portrait of the target user, and based on the similar user group obtained according to the user portrait, a Bayesian model is used to predict the target user's preference value for the target service for the user portrait of the target user, so as to determine whether to push the target service to the target user based on the preference value. In this way, the present application considers the factors affecting the target user's preference for the target service from many aspects, and combines similar user groups and Bayesian models to effectively improve the accuracy of pushing the target service of interest to the target user, thereby providing users with a good service experience.

[0071] Table 2

[0072]

[0073] Table 3

[0074]

[0075] In some embodiments, based on the Bayesian model, similar users with each label that meets the user portrait in the similar user group are obtained, and the first repetition probability and the first non-repetition probability of the target business experienced by similar users corresponding to each label are obtained, wherein, when constructing the user portrait of the target user, the target user's own state data and the scene data in which the user is located form the label of the user portrait; the second repetition probability and the second non-repetition probability of the historical experience of the similar user group for the target business are obtained; the total repetition probability is obtained according to the first repetition probability and the second repetition probability, and the total non-repetition probability is obtained according to the second non-repetition probability and the second non-repetition probability; the cloud push index is obtained according to the total repetition probability and the total non-repetition probability as the preference value of the target user for the target business. In this way, based on the Bayesian model, we can try to find a business that is closer to user behavior and in line with user tastes, so as to effectively screen out the business worth pushing to the target user from thousands of businesses, and achieve accurate push.

[0076] Table 4

[0077]

[0078] The target experience service may include but is not limited to playing music, searching on Baidu, watching videos, browsing news or advertisements. Historical experience can be understood as the number of times similar users repeat operations on the same service, such as the number of times a user repeatedly listens to the same song, the number of times a user repeatedly watches the same video, or the number of times a user repeatedly browses the same news.

[0079] Specifically, since the Bayesian model is based on the assumption that conditional probabilities are independent of each other, the process of obtaining the posterior probability by calculating the prior probability and using the Bayesian formula is specific. In the embodiment of the present application, based on the similar user group obtained based on big data, the calculated first repetition probability, the first non-repetition probability, the second repetition probability and the second non-repetition probability are used as prior probabilities, and the posterior probability, i.e., the total repetition probability and the total non-repetition probability, is calculated using the Bayesian formula. The total repetition probability and the total non-repetition probability are then divided, and the calculation result is the cloud push index, i.e., the cloud push index K = total repetition probability / total non-repetition probability. The cloud push index is used as the preference value of the target user for the target service, and this preference value is used to determine whether to push the target service to the target user, thereby improving the accuracy of pushing the target service of interest to the target user.

[0080] For example, taking music push as an example, Table 2 shows the first repetition probability and the first non-repetition probability table of the target service of similar user experience corresponding to each tag obtained for the user portrait, Table 3 shows the second repetition probability and the second non-repetition probability table of the historical experience of the target service by the similar user group, and Table 4 shows the cloud push index table obtained according to the total repetition probability and the total non-repetition probability. Specifically, for "Stupid Kid", based on the network big data such as a similar user group composed of 100,000 similar users, the Bayesian model is used to predict whether the target user likes the music, that is, the preference value. Among them, for the first time the user listens to this song, the total repetition probability corresponding to the first preference degree is obtained by multiplying all the first repetition probabilities shown in Table 1 with the second repetition probability when the song is recommended to the user for the first time, that is, X1 = a1*30%*20%*40%*50%*20%*80%*90%*50%*60%*60%*70%, a1 = 80% , X1=0.017418240000%, and multiply all the first non-repetition probabilities shown in Table 1 by the second non-repetition probabilities when the song is recommended to the user for the first time to obtain the total non-repetition probability corresponding to the first preference degree, that is, Y1=b1*70%*80%*60%*50%*80%*20%*10%*50%*40%*40%*30%, b1=20%, X1=0.001290240000%, and then divide the total repetition probability and the total non-repetition probability to obtain the first cloud recommendation index K1 value for listening to the song ; For the user listening to the song for the second time, the total repetition probability corresponding to the second preference degree is obtained by multiplying all the first repetition probabilities shown in Table 1 with the second repetition probability when the song is recommended to the user for the second time, and the total non-repetition probability corresponding to the second preference degree is obtained by multiplying all the first non-repetition probabilities shown in Table 1 with the second non-repetition probability when the song is recommended to the user for the second time, and then the total repetition probability and the total non-repetition probability are divided to obtain the second time listening cloud recommendation index K2 value; For the user listening to the song for the third time, the total repetition probability corresponding to the third preference degree is obtained by multiplying all the first repetition probabilities shown in Table 1 with the second repetition probability when the song is recommended to the user for the third time, and the total non-repetition probability corresponding to the third preference degree is obtained by multiplying all the first non-repetition probabilities shown in Table 1 with the second non-repetition probability when the song is recommended to the user for the third time, and then the total repetition probability and the total non-repetition probability are divided to obtain the third time listening cloud recommendation index K3 value, and so on, for the same service, each preference degree is obtained based on the second repetition probability and the second non-repetition probability corresponding to each recommendation for similar users. Therefore, through the above method, based on the Bayesian model, we can try to find businesses that are closer to user behavior and in line with user tastes, so as to effectively screen out businesses worth pushing to target users from thousands of businesses and achieve accurate push.

[0081] Among them, referring to Table 4, the cloud recommendation index K1=13.50 when the user listens to this song for the first time, the cloud recommendation index K2=5.06 when the user listens to this song for the second time, and the cloud recommendation index K3=3.38 when the user listens to this song for the third time. Based on the Bayesian model, it can be predicted that the user's preference for this song is gradually decreasing.

[0082] In some embodiments, if it is determined that the preference value is greater than or equal to the preference threshold, the target service is pushed to the target user; if it is determined that the preference value is less than the preference threshold, the target service is not pushed to the target user. In this way, whether to push the target service to the target user can be determined, which can effectively filter out the recommended services among thousands of services, and achieve the effect of accurately pushing the latest, hottest, and most favorite services to users, and can also effectively filter out the services that users are not interested in, providing users with a good service experience.

[0083] The preference threshold may be preset according to actual conditions such as business type, and there is no restriction on this.

[0084] For example, take music push as an example, the music is "Stupid Child", the preset preference threshold is 3, if "Stupid Child" is pushed for the first time, and the target user's preference value K for "Stupid Child" is less than 3, then "Stupid Child" is not recommended, on the contrary, if the target user's preference value K for "Stupid Child" is greater than or equal to 3, then "Stupid Child" is recommended. It can be understood that when judging whether to push the target service to the target user by the preference value, the preference value corresponding to the current number of recommendations of the service needs to be compared with the preference threshold. For example, if "Stupid Child" is pushed for the first time, the preference value K1 is compared with the preference to determine whether to push the target service, and if "Stupid Child" is not pushed for the target user for the first time, such as "Stupid Child" is pushed for the second time, the preference value K2 is compared with the preference to determine whether to push the target service, and so on, and finally the target user is screened out songs worth recommending.

[0085] In some embodiments, in order to further push preferred services to the target user, the embodiments of the present application can obtain the number of experiences of the target user for the target service pushed this time and the degree of completion of each experience, determine the user feedback index of the target user for the target service pushed this time according to the number of experiences and the degree of completion of each experience, and adjust the target service pushed next time according to the user feedback index. In other words, considering whether the target user likes the target service pushed this time and whether the service pushed this time is successful, the target user's experience of the target service pushed this time is evaluated through historical records and user feedback index, that is, the user feedback index is used to feedback the target user's feelings when experiencing the target service, so as to judge whether the target user likes the target service pushed this time, and thus the target service pushed next time is adjusted according to the user feedback index, so as to customize the preferred service for the target user.

[0086] The experience completion can be understood as the user's experience of the current service. For example, for music, the experience completion can be the listening time each time the music is listened to; for news, the experience completion can be the content volume each time the news is browsed.

[0087] In some embodiments, if the first target service whose user feedback index is less than the feedback index threshold value is obtained among the target services pushed this time, it is considered that the target user does not like the target service pushed this time, and therefore, the first target service is removed from the target service whitelist for the next push, that is, the first target service is put into the non-recommended list or blacklist, and the second target service other than the first target service among the target services pushed this time is retained in the target service whitelist for the next push. In this way, it can be ensured that the target service pushed each time is the service that the target user is interested in, avoiding the problem of repeatedly pushing services whose preference has decreased due to multiple experiences to the user, and improving the user experience.

[0088] Table 5

[0089] Number of auditions Trial duration User Feedback IndexA 1 Less than 30 seconds 0.5 2 Every time more than 30 seconds 1 3～5 Every time more than 30 seconds 2 5～10 Every time more than 30 seconds 3 More than 10 times Every time more than 30 seconds 4

[0090] For example, taking music push as an example, Table 5 shows a pre-set user feedback index table, in which the listening time is used to determine whether the recommendation is successful. If the user listens to the music for less than 30 seconds, it is considered that the user does not like the song pushed this time and the recommendation fails; if the user listens to the music for more than 30 seconds, it is considered that the user likes the song pushed this time and the recommendation is successful. Based on this, the user feedback index A of the target user for the target service pushed this time is obtained according to the relationship between the number of auditions, the audition duration and the user feedback index. For example, for the pushed music, if the target user has auditioned 3 times and each audition duration exceeds 30 seconds, the user feedback index A can be determined to be 2, and then according to the user feedback index A=2, it is determined whether the target service pushed this time is included in the target service whitelist for the next push. For another example, if the cloud recommendation index K2 of the song is 3, but the last user feedback index is 0.5, that is, the user's audition duration does not exceed 30 seconds, and the user switches songs, then the update push index of the song will drop to 1.5, and the song will not be included in the target service whitelist for the next push. Therefore, this method can facilitate the customization of preferred services for target users.

[0091] It should be noted that the user feedback index A is the user feedback index of the last time the target service was experienced. However, when the service is pushed to the user for the first time, the user does not have any experience record of the service, and thus does not have any experience result of the service. In this case, the user feedback index A adopts a preset default value, such as A=1.

[0092] In some embodiments, according to the preference value predicted for the next push, it is determined whether there is a new target service other than the target service pushed this time; if so, the new target service is added to the target service whitelist for the next push. That is to say, each time the target service is pushed, the method can calculate the target user's preference value for the newly added service as the service in the network data changes, such as the newly added service. If the preference value of the newly added service meets the recommendation condition, the newly added service is taken as the new target service and added to the target service whitelist for the next push, so as to provide the target user with a good and latest service experience.

[0093] In some embodiments, the first update push index is obtained according to the user feedback index and preference value corresponding to the second target service, and the second update push index is obtained according to the user feedback index and preference value corresponding to the new target service; the second target service and the new target service are sorted according to the first update push index and the second update push index, and used as the final target service whitelist for the next push. In this way, the purpose of cyclically updating the target service whitelist for each push is achieved, thereby providing a good service experience for the target user.

[0094] Please refer to the following Figure 2The service push method of the embodiment of the present application is illustrated by taking music push as an example, which specifically includes the following steps.

[0095] Step S6, constructing a user portrait.

[0096] Step S7, obtaining the cloud recommendation index K of the song.

[0097] Step S8, setting the preference threshold to 3, and adding songs with a cloud recommendation index K greater than or equal to 3 to the target service whitelist for this push.

[0098] Step S9, calculate the first updated recommendation index Ga, Ga = cloud recommendation index K1 * user feedback index A1, where, since the song is recommended for the first time, the default starting user feedback index A1 of each song is 1; K1 is the cloud recommendation index when the song is recommended for the first time based on a similar user group.

[0099] Step S10, sorting and pushing according to the first updated recommendation index Ga, specifically, the songs in the target service whitelist are sorted from high to low according to Ga.

[0100] Step S11, pushing the target service whitelist to the target user for the first time, and determining the audition result of the target user.

[0101] Step S12, refreshing the user feedback index A according to the number of auditions and the audition duration of the target user, and removing songs whose audition duration is less than 30 seconds from the target service whitelist.

[0102] Step S13, recalculate the first updated recommendation index Ga and the second updated recommendation index Gb according to the latest cloud recommendation index k and the user feedback index A. Specifically, the first updated recommendation index Ga = the user feedback index A2 and the cloud recommendation index K2 corresponding to the second target service, wherein A2 is the user feedback index of the last user audition, and K2 is the cloud recommendation index of the second recommended song obtained based on a similar user group; considering the situation of some new recommended songs added each time the song is pushed, the second updated recommendation index Gb = cloud recommendation index K1*user feedback index A1 when recalculating. Furthermore, the target service whitelist to be pushed next time is refreshed according to the calculated first updated recommendation index Ga and the second updated recommendation index Gb, and the songs in the target service whitelist to be pushed next time are sorted from high to low according to Ga and Gb, so as to push the target service whitelist to be pushed next time to the user.

[0103] Step S14, shuffling and pushing the target service whitelist recommended each time through the above steps in a loop.

[0104] It should be understood that although Figure 1 and Figure 2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0105] In one embodiment, Figure 3 As shown, a service push device 10 is provided, comprising: a first acquisition module 1, a construction module 2, a second acquisition module 3, a prediction module 4 and a push module 5.

[0106] The first acquisition module 1 is used to acquire the target user's own state data and the scene data in which the target user is located.

[0107] Construction module 2 is used to construct a user portrait of the target user based on its own status data and scenario data.

[0108] The second acquisition module 3 is used to obtain similar user groups based on user portraits.

[0109] The prediction module 4 is used to predict the target user's preference value for the target service based on the Bayesian model and the preference information of similar user groups for the target service.

[0110] The push module 5 is used to push the target service according to the preference value.

[0111] The above-mentioned service push device 10 comprehensively considers the target user's own status data and the scenario data in which he is located, such as the target user's age, gender, identity, time, city, interests and hobbies, etc., to construct a user portrait of the target user, and based on the similar user group obtained according to the user portrait, a Bayesian model is used to predict the target user's preference value for the target service for the user portrait of the target user, so as to determine whether to push the target service to the target user based on the preference value. In this way, the present application considers the influencing factors of the target user's preference for the target service from many aspects, and combines similar user groups and Bayesian models to effectively improve the accuracy of pushing the target service of interest to the target user, thereby providing users with a good service experience.

[0112] In one embodiment, the prediction module 4 is used to obtain similar users for each label that meets the user portrait in a similar user group based on a Bayesian model, and obtain a first repetition probability and a first non-repetition probability of similar users experiencing the target business corresponding to each label, wherein when constructing a user portrait of the target user, the target user's own state data and the scenario data in which the target user is located form the label of the user portrait; obtain a second repetition probability and a second non-repetition probability of the historical experience of the similar user group for the target business; obtain a total repetition probability based on the first repetition probability and the second repetition probability, and obtain a total non-repetition probability based on the second non-repetition probability and the second non-repetition probability; obtain a cloud push index based on the total repetition probability and the total non-repetition probability as the target user's preference value for the target business.

[0113] In some embodiments, the push module 5 is used to push the target service to the target user if the preference value is greater than or equal to the preference threshold, or not push the target service to the target user if the preference value is less than the preference threshold.

[0114] In some embodiments, the service push device 10 further includes a third acquisition module, a first determination module, and an adjustment module. The third acquisition module is used to obtain the number of experiences of the target user for the target service pushed this time and the degree of completion of each experience; the determination module is used to determine the user feedback index of the target user for the target service pushed this time according to the number of experiences and the degree of completion of each experience; and the adjustment module is used to adjust the target service pushed next time according to the user feedback index.

[0115] In some embodiments, the adjustment module is used to obtain the first target business among the target businesses pushed this time, whose user feedback index is less than the feedback index threshold; remove the first target business from the target business whitelist to be pushed next time, and retain the second target business among the target businesses pushed this time except the first target business in the target business whitelist to be pushed next time.

[0116] In some embodiments, the service push device 10 further includes a second determination module for determining whether there is a new target service other than the target service pushed this time according to the preference value predicted for the next push, and if so, adding the new target service to the target service whitelist for the next push.

[0117] In some embodiments, the service push device 10 further includes a fourth acquisition module. The fourth acquisition module is used to obtain a first update push index according to a user feedback index and a preference value corresponding to the second target service, and to obtain a second update push index according to a user feedback index and a preference value corresponding to the new target service; the push module 5 is used to sort the second target service and the new target service according to the first update push index and the second update push index, and use them as the final target service whitelist for the next push.

[0118] For the specific definition of the service push device 10, please refer to the definition of the service push method above, which will not be repeated here. Each module in the above service push device 10 can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0119] In one embodiment, Figure 4 As shown, an electronic device 20 is provided, including a memory 6 and a processor 7.

[0120] Specifically, the memory 6 stores a computer program, and the processor 7 implements the steps of the service push method provided in any embodiment of the present application when executing the computer program.

[0121] It is understandable that the electronic device 20 can be a terminal, such as a personal computer, a laptop, a smart phone, and a tablet computer. The electronic device 20 includes a processor 7, a memory 6, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor 7 of the electronic device 20 is used to provide computing and control capabilities. The memory 6 of the electronic device 20 includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The memory 6 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device 20 is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, an operator network, near field communication (NFC) or other technologies. When the computer program is executed by the processor 7, a service push method is implemented. The display screen of the electronic device 20 can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device 20 can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the housing of the electronic device 20, or an external keyboard, touchpad or mouse.

[0122] In one embodiment, the processor 7 implements the following steps when executing the computer program: based on the Bayesian model, obtain similar users for each tag that meet the user portrait in the similar user group, and obtain a first repetition probability and a first non-repetition probability of similar users experiencing the target business corresponding to each tag, wherein when constructing the user portrait of the target user, the target user's own state data and the scenario data in which the target user is located form the label of the user portrait; obtain a second repetition probability and a second non-repetition probability of the historical experience of the similar user group for the target business; obtain a total repetition probability based on the first repetition probability and the second repetition probability, and obtain a total non-repetition probability based on the second non-repetition probability and the second non-repetition probability; obtain a cloud push index based on the total repetition probability and the total non-repetition probability as the preference value of the target user for the target business.

[0123] In one embodiment, the processor 7 implements the following steps when executing the computer program: if it is determined that the preference value is greater than or equal to the preference threshold, the target service is pushed to the target user; if it is determined that the preference value is less than the preference threshold, the target service is not pushed to the target user.

[0124] In one embodiment, the processor 7 implements the following steps when executing the computer program: obtaining the number of experiences of the target user for the target service pushed this time and the degree of completion of each experience; determining the user feedback index of the target user for the target service pushed this time based on the number of experiences and the degree of completion of each experience; and adjusting the target service to be pushed next time based on the user feedback index.

[0125] In one embodiment, the processor 7 implements the following steps when executing the computer program: obtaining the first target business whose user feedback index is less than the feedback index threshold among the target businesses pushed this time; removing the first target business from the target business whitelist to be pushed next time, and retaining the second target business among the target businesses pushed this time except the first target business in the target business whitelist to be pushed next time.

[0126] In one embodiment, the processor 7 implements the following steps when executing the computer program: determine whether there is a new target service other than the target service pushed this time according to the preference value predicted for the next push; if so, add the new target service to the target service whitelist for the next push.

[0127] In one embodiment, the processor 7 implements the following steps when executing the computer program: obtaining a first update push index based on the user feedback index and preference value corresponding to the second target business, and obtaining a second update push index based on the user feedback index and preference value corresponding to the new target business; sorting the second target business and the new target business according to the first update push index and the second update push index, and using them as the final target business whitelist for the next push.

[0128] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0129] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the service recommendation method provided in any embodiment of the present application are implemented.

[0130] In one embodiment, the computer program implements the following steps when executed by a processor: based on a Bayesian model, similar users for each tag that satisfy the user portrait in a similar user group are obtained, and a first repetition probability and a first non-repetition probability of similar users experiencing the target business corresponding to each tag are obtained, wherein when constructing a user portrait of the target user, the target user's own state data and the scenario data in which the user is located form the label of the user portrait; a second repetition probability and a second non-repetition probability of the historical experience of the similar user group for the target business are obtained; a total repetition probability is obtained based on the first repetition probability and the second repetition probability, and a total non-repetition probability is obtained based on the second non-repetition probability and the second non-repetition probability; a cloud push index is obtained based on the total repetition probability and the total non-repetition probability as the preference value of the target user for the target business.

[0131] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: if it is determined that the preference value is greater than or equal to the preference threshold, the target service is pushed to the target user; if it is determined that the preference value is less than the preference threshold, the target service is not pushed to the target user.

[0132] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: obtaining the number of times the target user has experienced the target service pushed this time and the degree of completion of each experience; determining the user feedback index of the target user for the target service pushed this time based on the number of experiences and the degree of completion of each experience; and adjusting the target service to be pushed next time based on the user feedback index.

[0133] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: obtaining a first target business whose user feedback index is less than a feedback index threshold among the target businesses pushed this time; removing the first target business from a whitelist of target businesses to be pushed next time, and retaining a second target business among the target businesses pushed this time except the first target business in the whitelist of target businesses to be pushed next time.

[0134] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: determining whether there is a new target service other than the target service pushed this time according to the preference value predicted for the next push; and adding the new target service to the target service whitelist for the next push.

[0135] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: obtaining a first update push index based on a user feedback index and a preference value corresponding to the second target business, and obtaining a second update push index based on a user feedback index and a preference value corresponding to the new target business; sorting the second target business and the new target business according to the first update push index and the second update push index, and using them as the final target business whitelist for the next push.

[0136] In summary, the service push method, service push device, storage medium and electronic device provided in the embodiment of the present application comprehensively consider the target user's own state data and the scene data in which it is located, such as the target user's age, gender, identity, time, city and hobbies, etc., to build a user portrait of the target user, and based on the similar user group obtained according to the user portrait, the Bayesian model is used to predict the target user's preference value for the target service for the user portrait of the target user, so as to determine whether to push the target service to the target user by the preference value. In this way, the present application considers the factors affecting the target user's preference for the target service in many aspects, and combines the similar user group and the Bayesian model to effectively improve the accuracy of pushing the target service of interest to the target user, and provide the user with a good service experience. In addition, the user feedback index is used to feedback the feelings of the target user when experiencing the service, so as to determine whether the target user likes the target service pushed this time, and thus the target service pushed next time is adjusted according to the user feedback index, so as to customize the preferred service for the target user, ensure that the target service pushed each time is the service of interest to the target user, avoid the problem of repeatedly pushing the service whose preference is reduced due to multiple experiences, and improve the user experience.

[0137] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static random access memory (SRAM) and dynamic random access memory (DRAM).

[0138] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A service push method, characterized in that: include: Obtain the target user's own status data and the scene data he is in; Building a user profile of the target user according to the own state data and the scenario data; Obtaining a similar user group according to the user portrait; Based on the Bayesian model, predicting the preference value of the target user for the target service according to the preference information of the similar user group for the target service; Pushing the target service according to the preference value; Wherein, based on the Bayesian model, predicting the preference value of the target user for the target service according to the preference information of the similar user group for the target service includes: Based on the Bayesian model, similar users in the similar user group that meet each tag of the user portrait are obtained, and a first repetition probability and a first non-repetition probability of similar users corresponding to each tag experiencing the target business are obtained, wherein when constructing the user portrait of the target user, the target user's own state data and the scene data in which the target user is located form the tag of the user portrait; Obtaining a second repetition probability and a second non-repetition probability of historical experience of the similar user group for the target service; Obtain a total repetition probability according to the first repetition probability and the second repetition probability, and obtain a total non-repetition probability according to the second non-repetition probability and the second non-repetition probability; A cloud push index is obtained according to the total duplication probability and the total non-duplication probability to serve as a preference value of the target user for the target service.

2. The service push method according to claim 1, characterized in that: Pushing the target service according to the preference value includes: If it is determined that the preference value is greater than or equal to the preference threshold, the target service is pushed to the target user; If it is determined that the preference value is less than the preference threshold, the target service is not pushed to the target user.

3. The service push method according to claim 1 or 2, characterized in that: The service push method also includes: Obtain the number of times the target user has experienced the target service pushed this time and the degree of completion of each experience; Determine the user feedback index of the target user for the target service pushed this time according to the number of experiences and the completion degree of each experience; The target service to be pushed next time is adjusted according to the user feedback index.

4. The service push method according to claim 3, characterized in that: Adjusting the target service to be pushed next time according to the user feedback index includes: Obtaining the first target service whose user feedback index is less than the feedback index threshold among the target services pushed this time; The first target service is removed from the target service whitelist to be pushed next time, and the second target service in the target service pushed this time except the first target service is kept in the target service whitelist to be pushed next time.

5. The service push method according to claim 4, characterized in that: The service push method also includes: According to the preference value predicted for the next push, determine whether there is a new target business other than the target business pushed this time; If so, the new target service is added to the target service whitelist to be pushed next time.

6. The service push method according to claim 5, characterized in that: The service push method also includes: Obtaining a first update push index according to a user feedback index and a preference value corresponding to the second target service, and obtaining a second update push index according to a user feedback index and a preference value corresponding to the new target service; The second target service and the new target service are sorted according to the first update push index and the second update push index, and used as the final target service whitelist for the next push.

7. A service push device, characterized in that: include: The first acquisition module is used to acquire the target user's own state data and the scene data in which the target user is located; A construction module, used to construct a user portrait of the target user according to the own state data and the scenario data; A second acquisition module is used to obtain a similar user group according to the user portrait; A prediction module, configured to predict the target user's preference value for the target service based on the preference information of the similar user group for the target service based on a Bayesian model; A push module, used for pushing the target service according to the preference value; Wherein, the prediction module is specifically used for: Based on the Bayesian model, similar users in the similar user group that meet each tag of the user portrait are obtained, and a first repetition probability and a first non-repetition probability of similar users corresponding to each tag experiencing the target business are obtained, wherein when constructing the user portrait of the target user, the target user's own state data and the scene data in which the target user is located form the tag of the user portrait; Obtaining a second repetition probability and a second non-repetition probability of historical experience of the similar user group for the target service; Obtain a total repetition probability according to the first repetition probability and the second repetition probability, and obtain a total non-repetition probability according to the second non-repetition probability and the second non-repetition probability; A cloud push index is obtained according to the total duplication probability and the total non-duplication probability to serve as a preference value of the target user for the target service.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the service pushing method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the service push method according to any one of claims 1 to 6 are implemented.

Images