5g message personalized pushing method and system based on ai model

Abstract

The present application relates to the technical field of wireless communication, and particularly relates to a 5G message individualization pushing method and system based on an AI model, which comprises: obtaining the behavior record of a user in the process of using a system; analyzing the browsing content preference condition of the user after breaking the information cocoon according to the behavior record of the user; and pushing individualized messages to the user according to the browsing content preference condition of the user after breaking the information cocoon. The present application analyzes the behavior record of the user based on the AI model, and pushes individualized messages to the 5G terminal user through the 5G network. The present application analyzes the browsing content preference condition of the user after breaking the information cocoon, predicts and analyzes the new favorite type that the user may exhibit after breaking the information cocoon, so as to optimize the individualized pushing strategy of the system to the favorite of the user.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, in particular to a 5G message personalized pushing method and system based on an AI model. BACKGROUND

[0002] Message pushing has become a key strategy for application platforms to attract users and improve use frequency. The system aims to optimize user experience by in-depth analysis of user daily use of the application.

[0003] However, in the process of users obtaining information through forums and other systems, there is an information cocoon phenomenon. This may be due to the user's own long-term focus on and preference for a certain type of information, and the spontaneous construction of an information cocoon; it may also be due to improper system pushing mechanism, i.e. the system excessively focuses on a certain type of information to push to the user, thereby guiding the user to gradually prefer to receive this type of information in the use process, and thus forming an externally constructed information cocoon. If the information pushed by the system deviates from the user's true preference due to the "information cocoon" effect, it may cause user's resentment and dissatisfaction over a long period of time, which is not conducive to the use and promotion of the application. SUMMARY

[0004] In order to solve the technical problem that the pushed information deviates from the user's true preference due to the "information cocoon" effect, the purpose of the present application is to provide a 5G message personalized pushing method and system based on an AI model, and the technical solution adopted is as follows:

[0005] In the first aspect of the present application, the 5G message personalized pushing method based on an AI model comprises: step S1, obtaining the behavior record of the user in the use process of the system; the behavior record comprises data records of the user operating different content types of data information; step S2, analyzing the browsing content preference of the user after breaking the information cocoon according to the behavior record of the user; obtaining the preference degree sequence of the user for different content types of browsing according to the data records of the user operating different content types of data information; analyzing the change trend of the user's preference according to the preference degree sequence of the user for different content types of browsing, and screening out a user group with similar preferences; screening out a user group with similar preferences and consistent preferences from the user group with similar preferences; obtaining the browsing content preference of the user after breaking the information cocoon from the user group with similar preferences and consistent preferences; step S3, performing personalized message pushing to the user according to the browsing content preference of the user after breaking the information cocoon.

[0006] Further, according to the user's preference degree sequence of different content types, the change trend of the user's preference is analyzed, and a user group with similar preferences is screened out, specifically including: obtaining the preference change of the user according to the change between the adjacent date preference degree sequences of the user; analyzing the preference change of the user in a certain period of time to obtain the new preference degree sequence of the user; and obtaining a user group with similar preferences according to the new preference degree sequence of the user.

[0007] Further, according to the data record of the user's operation on different content types, the user's preference degree sequence of different content types is obtained, specifically including: the data record includes: browsing time, browsing times and browsing content; obtaining the user's preference coefficient for each content type according to the browsing time, browsing times and browsing content of the user's operation on different content types; and sorting each content type browsed by the user according to the size of the preference coefficient to obtain the user's preference degree sequence of different content types.

[0008] Further, the preference change of the user is obtained according to the change between the adjacent date preference degree sequences of the user, specifically including: generating the previous day preference content of the user according to the top 95% preference in the previous day preference degree sequence of the user; pushing the current day content to the user; the current day content includes the previous day preference content and associated content, the associated content is different from the previous day preference content, and the proportion of the associated content in the current day content is 5%; obtaining the similarity parameter between the adjacent two-day preference degree sequences according to the difference between the current day preference degree sequence and the previous day preference degree sequence of the user; and taking the similarity parameter between the adjacent two-day preference degree sequences as the preference change of the user.

[0009] Further, the preference change of the user includes browsing content similarity of a first similarity level and browsing content similarity of a second similarity level; when the similarity parameter between the adjacent two-day preference degree sequences is greater than or equal to a first threshold, it indicates that the browsing content similarity of the user in the adjacent two days is of the first similarity level; and when the similarity parameter between the adjacent two-day preference degree sequences is less than the first threshold, it indicates that the browsing content similarity of the user in the adjacent two days is of the second similarity level.

[0010] Further, the change of the preference of the user in a period of time is analyzed to obtain a new preference degree sequence of the user; and a user group with similar preferences and consistent preferences is obtained according to the new preference degree sequence of the user, specifically including: when the similarity of the browsing contents of the user in two adjacent days is a second similarity level, the similarity parameter between the user and other users in the preference degree sequence is obtained according to the difference between the preference degree sequence of the user and the preference degree sequence of other users; and the other users with the similarity parameter between the preference degree sequence of the user and the preference degree sequence of other users greater than a second threshold value are determined as the user group with similar preferences.

[0011] Further, the user group with similar preferences and consistent preferences is screened from the user group with similar preferences, specifically including: the difference index between the browsing record sequences of two adjacent days of each user is obtained by analyzing the preference degree sequence of each user to different content types in the current preset number of days and the similarity parameter between the preference degree sequences of two adjacent days of each user in the user group with similar preferences; and the user with the difference index between the browsing record sequences of two adjacent days less than or equal to a third threshold value is determined as the user group with similar preferences and consistent preferences.

[0012] Further, the browsing content preference of the user after breaking the information cocoon is obtained from the user group with similar preferences and consistent preferences, specifically including: the data records of the user group with similar preferences and consistent preferences when operating data information of different content types are analyzed to obtain the total number of people and the total time of browsing of each content type; and the popular coefficient of each content type in the user group with similar preferences and consistent preferences is obtained according to the total number of people and the total time of browsing of each content type, and then the browsing content preference of the user after breaking the information cocoon is obtained.

[0013] Further, the personalized message is pushed to the user according to the browsing content preference of the user after breaking the information cocoon, specifically including: the popular coefficient table is formed by sorting the popular coefficients of each content type in the user group with similar preferences and consistent preferences according to the size; and the associated content is replaced by the content with the popular coefficient greater than a fourth threshold value in the popular coefficient table when the user uses the system next time.

[0014] The second aspect of the application is a 5G message personalized pushing system based on an AI model, the pushing system comprising: a behavior record module for obtaining the behavior record of a user in the process of using the system; the behavior record comprising the data record of the user when operating different content type data information; a preference analysis module for analyzing the browsing content preference of the user after breaking the information cocoon according to the behavior record of the user; for obtaining the preference degree sequence of the user for different content types according to the data record of the user when operating different content type data information; analyzing the change trend of the user's preference according to the preference degree sequence of the user for different content types, and screening out a user group with similar preferences; screening out a user group with similar preferences and consistent preferences from the user group with similar preferences; obtaining the browsing content preference of the user after breaking the information cocoon from the user group with similar preferences and consistent preferences; a message pushing module for pushing personalized messages to the user according to the browsing content preference of the user after breaking the information cocoon.

[0015] The application has the following beneficial effects:

[0016] The application is aimed at the defects that the information obtained by the user when using the forum and other information exchange platforms is single or one-sided due to the information cocoon phenomenon, and it is difficult to accurately obtain the real preference of the user, so it is difficult to provide the best personalized message pushing method. The application analyzes the browsing content preference of the user after breaking the information cocoon, such as a large change in the proportion of original content types and new content types, and then predicts and analyzes the new preference type that the user may exhibit after breaking the information cocoon according to the actual behavior record of the user and the preference characteristics of the user group with similar preferences, so as to optimize the personalized pushing strategy of the system for the user's preference.

[0017] The application records and analyzes the operation behavior of the user by introducing a small amount of associated content in the message pushing process, such as the browsing frequency and duration of the user, and the proportion of the original content type and the new content type is greatly changed. The application obtains the browsing duration and frequency of the user for the new content type by recording and analyzing the behavior of the user after breaking the information cocoon, so as to evaluate the preference degree of the user for the new content, and then compare the information of the user with similar preference bias to optimize the personalized pushing strategy, so as to realize the best user preference recommendation pushing, thereby improving the pushing effect of the system and continuously attracting users. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art and the advantages thereof, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the following description only only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings.

[0019] Figure 1 The application provides a 5G message personalized pushing method based on an AI model.

[0020] Figure 2 The application provides a 5G message personalized pushing system based on an AI model. DETAILED DESCRIPTION

[0021] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object, the following describes the specific implementation, structure, features and effects of the 5G message personalized pushing method and system based on an AI model according to the present application in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0023] The following specifically describes the specific scheme of the 5G message personalized pushing method based on an AI model provided by the present application in combination with the accompanying drawings. Please refer to Figure 1 , which shows the flowchart of the 5G message personalized pushing method based on an AI model provided by one embodiment of the present application. The 5G message personalized pushing method based on an AI model in the embodiment of the present application includes the following steps:

[0024] Step S1, obtaining the behavior record of the user in the process of using the system; the behavior record includes the data record of the user operating different content types of data information. The system here is an application platform on various mobile terminals or a self-media operation platform, any platform with information promotion function such as recommending content to the user through historical data analysis. The behavior record of the user in the process of using the system refers to the record traces left by the user after entering the system for relevant operation through the user ID (Identification, identity number), including but not limited to browsing content, browsing times, browsing time, likes, collections, forwarding, attention and the like. The content type refers to the category of the browsed content, such as popular science or variety. According to the needs of the actual scene, it can also be a more detailed category, such as natural science, technology invention, health and health, environmental protection, social science, culture and art, education and teaching, and agricultural science and technology.

[0025] Step S2, according to the behavior record of the user, analyzing the browsing content preference of the user after breaking the information cocoon.

[0026] According to the data record of the user operating different content types of data information, the preference degree sequence of the user for different content types browsed is obtained.

[0027] According to the browsing time, browsing times and browsing content of the user operating different content types, the preference coefficient of the user for each content type is obtained; according to the size of the preference coefficient, the ranking of each content type browsed by the user is sorted, and the preference degree sequence of the user for different content types browsed is obtained.

[0028] According to the preference degree sequence of the user for different content types browsed, the change trend of the user's preference is analyzed, and the user group with similar preference is screened out. According to the change between the preference degree sequences of adjacent dates of the user, the preference change of the user is obtained. According to the preference in the top 95% of the preference degree sequence of the user in the previous day, the preferred content of the user in the previous day is generated.

[0029] Pushing the content of the day to the user; the content of the day includes the preferred content of the previous day and the associated content, the associated content is different from the preferred content of the previous day, and the proportion of the associated content in the content of the day is 5%. Here, 95% and 5% can be changed according to the needs of the actual scene.

[0030] According to the difference between the daily preference degree sequence of the user and the previous day preference degree sequence, a similarity degree parameter between the adjacent two-day preference degree sequences is obtained; and the similarity degree parameter between the adjacent two-day preference degree sequences represents the preference change of the user. The above-mentioned related content in the implementation of the application refers to the content of the non-direct preference or weak preference information of the user.

[0031] The preference change of the user includes a first similar level of browsing content similarity and a second similar level of browsing content similarity; when the similarity degree parameter between the adjacent two-day preference degree sequences is greater than or equal to a first threshold value, it indicates that the similarity of the browsing content of the user in the adjacent two days is the first similar level; and when the similarity degree parameter between the adjacent two-day preference degree sequences is less than the first threshold value, it indicates that the similarity of the browsing content of the user in the adjacent two days is the second similar level. The first similar level represents a higher similarity of the browsing content, and the second similar level represents a lower similarity of the browsing content.

[0032] The preference change of the user in a certain period of time is analyzed to obtain a new preference degree sequence of the user; and a user group with similar preferences is obtained according to the new preference degree sequence of the user.

[0033] When the similarity of the browsing content of the user in the adjacent two days is the second similar level, a similarity degree parameter between the preference degree sequence of the user and the preference degree sequence of other users is obtained according to the difference between the preference degree sequence of the user and the preference degree sequence of other users; other users with the similarity degree parameter between the preference degree sequence of the user and the preference degree sequence of other users greater than a second threshold value are identified as a similar preference user group; and a user group with similar preferences and consistent preferences is obtained from the similar preference user group. In the embodiment of the application, the first threshold value and the second threshold value can be the same or different, and are preferably the same, and the value is 0.8.

[0034] The difference index between the adjacent two-day browsing record sequences of each user is obtained by analyzing the preference degree sequence of each user for different content types browsed in the current preset number of days in the similar preference user group and the similarity degree parameter between the adjacent two-day preference degree sequences of each user. Users with a difference index between the adjacent two-day browsing record sequences less than or equal to a third threshold value are identified as a user group with similar preferences and consistent preferences. In the embodiment of the application, the third threshold value is 0.3. The current preset number of days is set to 15 days before the current day in the embodiment of the application. In actual scenarios, other days can be set according to different demand purposes.

[0035] The browsing content preference of the user after breaking the information cocoon is obtained from the user group with similar preferences and consistent preferences.

[0036] The data records of the data information of different content types are operated by analyzing the user group with similar preferences and consistent preferences, to obtain the total number of people and the total time length of browsing of each content type. According to the total number of people and the total time length of browsing of each content type, the popular coefficient of each content type in the user group with similar preferences and consistent preferences is obtained, and then the browsing content preference of the user after breaking the information cocoon is obtained.

[0037] In step S3, according to the browsing content preference of the user after breaking the information cocoon, personalized message pushing is performed to the user. According to the popular coefficient of each content type in the user group with similar preferences and consistent preferences, the popular coefficient table is sorted in size, and the associated content in the front is replaced with the content with a popular coefficient greater than the fourth threshold value in the popular coefficient table when the user uses the system next time. That is, the content with a popular coefficient in the front of a certain percentage in the popular coefficient table replaces the original associated content. The proportion of the replaced associated content in the content of the day is unchanged.

[0038] In the embodiment of the present application, whether the user breaks the information cocoon is judged according to the difference between the preference degree sequence of the day and the previous day, and whether the user is consistent in preference is judged according to the change of the preference degree sequence of half a month. But in some actual scenarios, this frequency time can have other possibilities, such as judging whether the user breaks the information cocoon according to the difference between the preference degree sequence of the current month and the previous month, and judging whether the user is consistent in preference according to the change of the preference degree sequence of half a year; such as judging whether the user breaks the information cocoon according to the difference between the preference degree sequence of the current hour and the previous hour, and judging whether the user is consistent in preference according to the change of the preference degree sequence of the day; such as judging whether the user breaks the information cocoon according to the difference between the preference degree sequence of the current login period and the previous login period, and judging whether the user is consistent in preference according to the change of the preference degree sequence of the previous ten login periods; and so on, not listed.

[0039] In another possible embodiment of the present application, the information obtained by the user when using the forum and other information exchange platforms is single or one-sided due to the information cocoon phenomenon, and it is difficult to accurately obtain the real preferences of the user, thereby making it difficult to provide the best personalized message pushing method. In the message pushing process, a small amount of non-user direct preference or weak preference information is introduced, the operation behavior of the user is recorded and analyzed, such as the number of browsing times, the time length and the like of the user, and the proportion of the original content type and the new content type is greatly changed, and then according to the actual behavior record of the user and the preference characteristics of the user group with similar preferences, the new preference content type that the user may exhibit after breaking the information cocoon is predicted and analyzed, thereby optimizing the personalized pushing strategy of the system to the user's preferences. By recording and analyzing the behavior of the user after breaking the information cocoon, the browsing time length, frequency and the like of the user to the new content type are obtained, so as to evaluate the preference degree of the user to the new content, and then the user information with similar preference bias is compared, the personalized pushing strategy is optimized, and the best user preference recommendation pushing is realized, thereby improving the pushing effect of the system and continuously attracting the user. The present application analyzes the behavior record of the user based on the AI model, and finally performs personalized message pushing to the 5G terminal user through the 5G network.

[0040] In the embodiment of the present application, the behavior data of the user after breaking the information cocoon is recorded and analyzed in detail, and the browsing time length, frequency and the like of the user to the new content type are accurately obtained as key indicators, so as to evaluate the preference degree of the user to the new type. Then, the user information with similar preference bias is compared and analyzed, the personalized pushing content is optimized, the accurate and efficient user preference recommendation pushing is realized, the pushing effect of the system is improved, and the user is continuously attracted and retained.

[0041] The specific scene to which the embodiment of the present application is directed is that there may be a case of pushing non-user preference content due to the information cocoon phenomenon in the system, which reduces the preference degree of the user to the system. Therefore, the preference bias of the user is analyzed by pushing a small amount of weakly related information to the user, and the preference type of the user is further analyzed by combining the preference content of the same type of user, thereby realizing the personalized pushing of the user message. The overall logic of the 5G message personalized pushing method based on the AI model of the embodiment of the present application is as follows:

[0042] I. The system obtains the behavior record of the user in the process of using the system.

[0043] The forum system is a platform for users to obtain and exchange information, and after the user logs in the system using the 5G terminal, the user can obtain the news, hot topics and content messages in the system that meet the own preferences after AI model data processing.

[0044] Any operation on the system messages by the user after logging in the system will be automatically recorded by the system, including the operation type C, operation time T and operation frequency N for the user ID.

[0045] The recorded data will be transmitted to the system background and then used to make a personal behavior record form of the user. The form will be updated in real time according to the operation of the user on the system content.

[0046] Meanwhile, the system will also generate a single-day behavior record form for the operation of the user on the system content every day, so as to facilitate the management and analysis of the behavior state of the user.

[0047] II. Analyzing the browsing content preference of the user after breaking the information cocoon according to the behavior record of the user.

[0048] a. According to the data record of the operation of the user on different types of data information, the overall change is obtained.

[0049] When using the forum and other self-media operation platform, such as the platform for information promotion and the like, the user will pay more attention to the content preferred by him. For example, if the user likes to browse news, the browsing times of the news type content in the behavior record of the user will be more, and the browsing time will be relatively longer. In comparison, the browsing times and time of other types of content are relatively less and short.

[0050] Therefore, by analyzing the browsing time and browsing content of the user on each part of the content type, the preference degree of the user on each part of the content type can be obtained.

[0051]

[0052] In the formula, v is the preference coefficient of the user on each content type, i is the number of the operation times of the user on the content type, N is the number of the operation times of the user on the module, T i is the time length of the i-th operation of the user on the module, T c is the total time length of using the system platform, N c is the total number of the operation times of the user on all content types of the system, is the ratio of the number of the operation times of the user on the content type to the total number of the operation times of the user on all content types of the system, which is used to describe the preference degree of the user, the greater the value, the higher the preference degree, and vice versa. is the total time length of the operation of the user on the module. is the total time length of the operation of the user on the module, which is used to represent the preference degree of the user on the information of this type.

[0053] Type, module and system, the relationship between the three: type refers to several similar data, such as several videos belong to science or variety, module refers to the part that users can operate when using the system, a type is a module, type and module can be replaced by type, system is a software application system, including information type and module.

[0054] All content types browsed by the user are traversed using the above method, the user's preference degree for all content types is obtained, and the v values are compared, the content types C browsed by the user are sorted in descending order of preference degree, and the user's preference degree sequence is obtained. At the same time, record the user's preference coefficient v corresponding to each content type C in the sequence.

[0055] b. According to the user's preference degree sequence of different browsing content types, analyze the change trend of user's preference, and filter out the user group with similar preference, so as to deeply study and understand the user's preference.

[0056] When users use forums and other platforms to obtain information, whether out of their own preference or influenced by platform recommendations, they are prone to information cocoon phenomenon. If the information cocoon is formed due to the user's own preference, the system pushed content will usually meet the user's interest, which can not only improve the user's browsing times, but also attract the user. However, if the information cocoon is mainly caused by the platform's recommendation algorithm, the user's interest in these pushed messages may be relatively flat. In the long run, this situation not only may not achieve the expected effect, but also may cause the user's resentment and dissatisfaction.

[0057] Therefore, in order to more accurately analyze the user's preference, and avoid the deviation of the pushed message from the user's true preference due to the information cocoon phenomenon, it is necessary to reanalyze and evaluate the user's preference bias.

[0058] ① According to the change between the user's preference degree sequence of adjacent dates, the change of the user's preference is obtained.

[0059] For users with information cocoon phenomenon, the system adds or replaces 5% of the content with low preference degree to the user's recommended content every day, and replaces it with other content with related types. And detect the user's browsing situation of the newly added content, if the user's browsing frequency of this part of content is high or the browsing time is long, then the user's preference degree of this part of content is large, on the contrary, if the user's browsing frequency of this part of content is low, the time is short, then the user's interest in this part of content is low.

[0060] The operation frequency and time length of the user on different content types in the day are analyzed in depth by using the method of step a to construct the user preference degree sequence of different content types. If the user shows high frequency and long time operation behavior on a new content type, the content type is ranked in the front in the preference degree sequence, and the sequence in the day is significantly different from the previous day; on the contrary, if the user is not active in operating the new content type, the preference degree sequence in the day is less different from the previous day.

[0061] By comparing and analyzing the change of the preference degree of each content type in the user preference degree sequence of adjacent two days, whether the user is interested in the new content is analyzed. If the difference is small, it means that the user is not interested in the new content; if the difference is significant, it means that the user is highly interested in the new content, and the user's information cocoon is generated by system recommendation, and the subsequent recommendation content strategy needs to be adjusted accordingly.

[0062] The content type information in each sequence is extracted, the similarity between the user preference degree sequences of adjacent two days is obtained by using cosine similarity, and the difference between the corresponding values of the same content type preference coefficient, i.e. the difference value Δf=v t -v y (v t is the preference coefficient of the user for the content type in the day, v y is the preference coefficient of the user for the content type in the previous day) is taken as a reference value to obtain the similarity parameter between the operation sequences of adjacent two days of the user:

[0063]

[0064] In the formula, δ is the similarity degree parameter between the two sequences, z is the sequence number of each content type in the current sequence, n is the sequence length, d is the cosine value between the current sequence and the historical sequence, which is used to represent the similarity between the two sequences. The higher the cosine value, the more similar the two sequences, and vice versa. Δf z is the change difference value between the corresponding values of the preference coefficients of the content types in the two sequences. The larger the value, the greater the change in the proportion of the user's browsing of the same content type, and vice versa. is the sum of the difference values between the corresponding values of the preference coefficients of the same content type in the two sequences. The larger the value, the less similar the two sequences, and vice versa. is the mean of the difference values between the preference coefficients of the same content type in the two sequences, which is used to represent the similarity between the two sequences. In order to ensure that the value is proportional to the similarity between the two sequences, the value is operated by negative one power, so that The larger the value, the higher the similarity between the sequences, and vice versa. The similarity between two sequences is represented by the value, which is normalized by the norm() function to a value range of [0, 1].

[0065] Subsequently, the threshold is set to 0.8, and when δ≥0.8, the similarity of the user's browsing content between adjacent days is the first similarity level, otherwise, the similarity of the user's browsing content between adjacent days is the second similarity level.

[0066] ② Analyze the change of the user's preference during browsing to obtain the user's new preference degree sequence, and obtain the user group with similar preference sequence and stability according to the sequence.

[0067] When the similarity of the user's browsing content between adjacent days is the second similarity level, that is, there is a significant difference in the user's browsing of different types of content in the forum system between adjacent days, in order to ensure that the subsequent messages pushed to the user are more close to his personal preferences, it is necessary to re-analyze the user's preferences. In this process, the user's real favorite content may not have been detected, therefore, it is necessary to analyze the user's actual preference type in a deeper level, and in order to ensure that the subsequent pushed content is the content that the user is interested in, the above method is realized by means of AI model to identify other user sequences with similar user preference content types in the system, and to match the other user preference degree sequence with the user's preference degree sequence. The similarity parameter δ between the user preference degree sequence and each other user preference degree sequence is obtained, and if δ≥0.8, it is divided into the classification of similar preference types.

[0068] Among these users, part of the users' preference types are still in the analysis stage, and another part of the users' preference types have been analyzed. For the user data still in the change process, its instability may have a bad effect on the data result. Therefore, it is necessary to use the user data whose analysis is completed and the preference type is relatively stable as the main object of study.

[0069] Therefore, the preference degree sequence of each user in the past half month is collected, and these sequences are analyzed based on the difference between adjacent dates to understand the change of the user's preference between adjacent dates. If the user's browsing state changes little between adjacent dates, it indicates that the user's preference type is relatively stable; otherwise, it indicates that the user's preference is still in a large change.

[0070]

[0071] In the formula, F is the difference index between the preference degree sequences of adjacent two days of each user, j is the browsing date number of the user, m is the total browsing date number of the user, k is the number of the detected new content type, l is the number of the new content type, v j,k is the preference coefficient of the user on the content of the new content type k on the jth day. is the difference value between the two adjacent days of the user preference coefficient of the same content type, the closer to 1, the more stable the user's favorite type area, otherwise it is still in the analysis process; δ j+1 is the similarity parameter between the two sequences of the content type browsed by the user in the last day and the previous day, the larger the value, the greater the difference between the two adjacent days of the user, otherwise, the smaller the difference, in order to ensure that each part in the formula is positively correlated, the reciprocal of δ j+1 is obtained.

[0072] The threshold is set to 0.3, when F≤0.3, the user's browsing of the new content type tends to be stable, otherwise it is still in the process of change.

[0073] c. For the information type with more browsing times in the user group with similar preferences and stable browsing. When analyzing, the influence of the new type should be reduced, so the new type should be excluded and the original information type should be analyzed.

[0074] The user group with similar preferences and stable browsing is obtained through the above step. In order to better understand the real preferences of the user who breaks the information cocoon, and at the same time ensure that the pushed content is as close to the user's preferences as possible, the content type that the user who breaks the information cocoon prefers should be referred to, and the information of this part of the content type should be pushed to the user, and the user's browsing should be observed.

[0075] The content types browsed by each user are sorted and summarized, and the browsing times of each user are sorted according to the content type. If the same content type appears, further sorting is performed according to the browsing time.

[0076] At this time, the browsing times and browsing time of each content type in the sequence are obtained, and the popularity coefficient is determined by the size of the browsing times and time of each content type information, and the most popular content type is screened out, which is determined as the content type pushed to the user who breaks the information cocoon.

[0077] P=max(a c ,b c );

[0078] In the formula, P is the popularity coefficient of content type C in the screened user, a c , b c are the user times and total browsing time of content type C in the sorted sequence, and max(a c , b cThe product of the number of user visits and the total browsing time of the sequence is filtered out as the maximum value of the information type that needs to be pushed.

[0079] III. Personalized message pushing for users according to the changed preferences based on user behavior records.

[0080] The system transmits the operation records of the user using the 5G terminal to the background database through the function management module using efficient and secure 5G network equipment, and saves them in the personal behavior record table corresponding to the user in the database. When analyzing the operation behavior of the user, the AI model starts to run, retrieves all the recent behavior record forms of the user from the system background database through the user ID to obtain the behavior data of the user.

[0081] Then, the AI model analyzes the behavior data of the user to detect whether there is a user who breaks the information cocoon phenomenon, and analyzes the change of the user's favorite information type. For the change of the user's favorite information type, the high-similarity users existing in the database are retrieved by comparing the similarity of the favorite types. After performing the operation in step (b), the AI model further screens out users with stable favorite information types, and obtains the popularization coefficients of each information type in the overall operation information type of these users through the method in step (c).

[0082] Then, the AI model sorts the popularization coefficients of each information type by size, makes a popularization coefficient form, and delivers the form to the personalized message pushing module. After receiving the data form delivered by the AI model, the message pushing module pushes the 5G message to the user according to the first 5 popularization information types in the form through the 5G base station, and records the operation of the user on these contents.

[0083] Finally, when the user logs in next time, the system will replace the last 5% of the content in the user's favorite list with the first 5 information types in the popularization coefficient form in the recommended content, and record the operation of the user.

[0084] This process will be repeated until the most favorite content type of the user is obtained, so as to realize the accurate pushing of personalized messages for the user.

[0085] Based on the same inventive concept as the above method embodiment, the embodiments of the present application also provide a 5G message personalized pushing system based on an AI model, as shown in Figure 2 The pushing system comprises:

[0086] A behavior record module for obtaining the behavior record of the user during the use of the system; the behavior record includes the data record of the user operating different content types of data information.

[0087] The preference analysis module is configured to analyze the user's browsing content preference after breaking out of the information cocoon according to the user's behavior record, obtain a user's preference degree sequence for different content types according to the user's data record when operating different content types of data information, analyze the user's preference change trend according to the user's preference degree sequence for different content types, and screen out a user group with similar preferences; and obtain the user's browsing content preference after breaking out of the information cocoon from the user group with similar preferences and consistent preferences.

[0088] The message pushing module is configured to push personalized messages to the user according to the user's browsing content preference after breaking out of the information cocoon.

[0089] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0090] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

Claims

1. A 5G personalized message push method based on an AI model, characterized in that, The push method includes: Step S1: Obtain user behavior records during system usage; the behavior records include data records of user operations on different types of content information. Step S2: Analyze the user's browsing content preferences after breaking out of the information cocoon, based on the user's behavior records; Based on the data records of users' operations on different content types, a sequence of user preferences for different content types is obtained; based on the sequence of user preferences for different content types, the changing trend of user preferences is analyzed, and user groups with similar preferences are screened out; from the user groups with similar preferences, a user group with similar and consistent preferences is screened out; from the user group with similar and consistent preferences, the user's browsing content preferences after breaking out of the information cocoon are obtained. Step S3: Based on the user's browsing preferences after breaking out of the information cocoon, personalized message pushes are sent to the user. From the aforementioned user groups with similar preferences, a user group with similar and consistent preferences is selected, including: By analyzing the user group with similar preferences, the sequence of preferences for different content types browsed by each user within the current preset number of days, and the similarity parameter between the preference sequences of each user on two adjacent days, the difference index between the browsing record sequences of each user on two adjacent days is obtained. Users whose difference index between two adjacent days' browsing record sequences is less than or equal to the third threshold are identified as a user group with similar and consistent preferences. From user groups with similar and consistent preferences, we obtained users' browsing content preferences after breaking out of their information cocoons, including: Analyze user groups with similar and consistent preferences, and record data when operating on data information of different content types to obtain the total number of views and total viewing time for each content type; Based on the total number of views and total viewing time for each content type, the popularity coefficient of each content type in the user group with similar and consistent preferences is obtained, thereby revealing the user's browsing content preferences after breaking out of the information cocoon.

2. The 5G message personalized push method based on an AI model according to claim 1, characterized in that, Based on the sequence of user preferences for different types of content viewed, we analyze the changing trends of user preferences and filter out user groups with similar preferences, specifically including: The user's preference changes are obtained based on the changes in the sequence of preference levels between adjacent dates. Analyze the changes in users' preferences over a certain period of time to obtain a new sequence of user preferences; based on the new sequence of user preferences, identify user groups with similar preferences.

3. The 5G message personalized push method based on an AI model according to claim 2, characterized in that, Based on the data records of users interacting with different content types, a sequence of user preferences for different content types is obtained, specifically including: The data records include: browsing duration, number of browsing sessions, and browsing content; Based on the browsing time, number of browsing sessions, and content viewed when users interact with different content types, we obtain the user's preference coefficient for each content type. Based on the magnitude of the preference coefficient, the content types browsed by the user are sorted to obtain a sequence of the user's preference for different content types.

4. The 5G message personalized push method based on an AI model according to claim 3, characterized in that, Based on the changes in the user's preference intensity sequence between adjacent dates, the user's preference changes are obtained, specifically including: Generate user preference content based on the top 95% of the user's preferences in the previous day's preference sequence. Push daily content to users; the daily content includes the previous day's preferred content and related content, the related content is different from the previous day's preferred content, and the related content accounts for 5% of the daily content; Based on the difference between the user's daily preference sequence and the previous day's preference sequence, a similarity parameter between two adjacent days' preference sequences is obtained; The similarity parameter between the preference sequences of two adjacent days represents the changes in the user's preferences.

5. The 5G message personalized push method based on an AI model according to claim 4, characterized in that, Changes in user preferences include browsing content with the first level of similarity and browsing content with the second level of similarity; When the similarity parameter between the preference sequences of two adjacent days is greater than or equal to the first threshold, it indicates that the similarity of the browsing content of the user on two adjacent days is at the first similarity level; when the similarity parameter between the preference sequences of two adjacent days is less than the first threshold, it indicates that the similarity of the browsing content of the user on two adjacent days is at the second similarity level.

6. The 5G message personalized push method based on an AI model according to claim 5, characterized in that, Analyze changes in user preferences over a certain period to obtain a new sequence of user preference levels; based on this new sequence, identify user groups with similar preferences, specifically including: When the similarity of a user's browsing content on two consecutive days is at the second similarity level, the similarity parameter between the user's preference sequence and other users' preference sequences is obtained based on the difference between the user's preference sequence and other users' preference sequences. Users whose similarity parameter between this user and other users' preference sequences is greater than a second threshold are identified as a user group with similar preferences.

7. The 5G message personalized push method based on an AI model according to claim 4, characterized in that, Personalized message pushes are delivered to users based on their browsing preferences after breaking out of their information cocoons, specifically including: Based on the popularity coefficient of each content type among user groups with similar and consistent preferences, the content types are sorted by size to form a popularity coefficient table. The next time a user uses the system, the associated content will be replaced with content from the welcome coefficient form whose popularity coefficient is greater than the fourth threshold.

8. A 5G personalized message push system based on an AI model, characterized in that, The push system includes: The behavior recording module is used to acquire user behavior records during system usage; the behavior records include data records of user operations on different types of content information; The preference analysis module is used to analyze users' browsing content preferences after breaking out of their information cocoons based on their behavior records; to obtain a sequence of user preferences for different content types based on data records of user interactions with different content types; to analyze the changing trends of user preferences based on the sequence of user preferences for different content types, and to filter out user groups with similar preferences; from the user groups with similar preferences, to filter out user groups with similar and consistent preferences; and from the user groups with similar and consistent preferences, to obtain users' browsing content preferences after breaking out of their information cocoons. The message push module is used to push personalized messages to users based on their browsing preferences after breaking out of their information cocoon. From the aforementioned user groups with similar preferences, a user group with similar and consistent preferences is selected, including: By analyzing the user group with similar preferences, the sequence of preferences for different content types browsed by each user within the current preset number of days, and the similarity parameter between the preference sequences of each user on two adjacent days, the difference index between the browsing record sequences of each user on two adjacent days is obtained. Users whose difference index between two adjacent days' browsing record sequences is less than or equal to the third threshold are identified as a user group with similar and consistent preferences. From user groups with similar and consistent preferences, we obtained users' browsing content preferences after breaking out of their information cocoons, including: Analyze user groups with similar and consistent preferences, and record data when operating on data information of different content types to obtain the total number of views and total viewing time for each content type; Based on the total number of views and total viewing time for each content type, the popularity coefficient of each content type in the user group with similar and consistent preferences is obtained, thereby revealing the user's browsing content preferences after breaking out of the information cocoon.

Images