An information processing method and device based on user multi-source data, and a medium
By integrating multi-source data to generate data feature vectors and combining them with sorting algorithms to optimize the video display order, the problem of unpredictable interest preferences of new or low-frequency users in existing technologies is solved, achieving higher recommendation accuracy and user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING CENT BIOLOGY CO LTD
- Filing Date
- 2025-08-21
- Publication Date
- 2026-06-23
AI Technical Summary
In existing technologies, personalized recommendation methods based on a single data source are difficult to accurately predict the interests and preferences of new users or users with low frequency of use, and cannot reflect users' cross-application behavior across different applications, resulting in low recommendation accuracy and affecting user experience.
By acquiring multi-source data from target users, including basic personal attribute information, operational behavior data of the first application, and operational data of the second application on the target device, a data feature vector is generated. This vector is then combined with a ranking algorithm to determine the model, dynamically select the target ranking algorithm, and optimize the video display order.
It improves the ability to capture the interests of new users or users with low frequency of use, enhances the relevance and matching degree of video recommendations, and improves the user experience.
Smart Images
Figure CN121071237B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information processing technology, and in particular to an information processing method, device and medium based on multi-source user data. Background Technology
[0002] With the explosive growth of internet content, such as the millions of new videos added daily on short video platforms, the cost for users to manually filter the information they need has increased significantly. Therefore, personalized recommendations are needed. These recommendations use algorithms to filter out irrelevant or low-relevance content and only show users content that highly matches their interests and needs, thereby improving the user experience. In existing technologies, personalized recommendation methods often rely on a single data source for recommendation ranking. For example, they predict users' interests and preferences based on their historical actions within the application (such as clicks, browsing, likes, and shares), determine the recommendation ranking based on the predicted interests and preferences, and then display several pieces of content to the user according to the recommended ranking.
[0003] However, the above method also has the following technical problems:
[0004] Relying solely on historical user behavior within an application to predict user interests is insufficient for new users or those who use the application infrequently. Furthermore, user actions in other applications may imply underlying interests, and this method ignores cross-application behavior information. Additionally, users' intentions or purposes for launching an application may differ under different circumstances; relying solely on historical user behavior within the application cannot predict the user's current true interest state, or their current intention or purpose for launching the application. Therefore, the accuracy of interests predicted using this method is low, and it cannot predict the user's current true interest state. Consequently, the accuracy of the recommendation order determined by this method is also low, potentially impacting user experience. Summary of the Invention
[0005] To address the aforementioned technical problems, the present invention provides an information processing method, device, and medium based on multi-source user data.
[0006] According to a first aspect of the present invention, an information processing method based on multi-source user data is provided. The method is applied to a first application and includes the following steps:
[0007] S1. In response to the target user launching the first application, acquire multi-source data of the target user; the multi-source data includes: the target user's basic personal attribute information data R1, the target user's operation behavior data of the first application R2, and the second application operation data of the target device R3; wherein, R3 includes the application type and foreground dwell time of each second application that has been run in the foreground of the target device within the target time period; the first application is used to display video; the second application is an application other than the first application in the target device, and the target device is the device where the first application is located; the end time of the target time period is the current time point, and the duration of the target time period is the preset duration.
[0008] S2. Obtain data feature vector T based on multi-source data of target users, T = (A, B, C); A is the basic attribute feature obtained based on R1, B is the first operational behavior feature obtained based on R2, and C is the second operational behavior feature obtained based on R3.
[0009] S3. Based on T and the sorting algorithm, determine the model and select the target sorting algorithm from several preset sorting algorithms.
[0010] S4. Use a target sorting algorithm to determine the video display order of several videos to be displayed, and display the videos to be displayed to the target user according to the video display order.
[0011] According to a second aspect of the present invention, a non-transitory computer-readable storage medium is provided, wherein a computer program is stored in the storage medium, and the computer program is loaded and executed by a processor to implement the aforementioned method.
[0012] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method.
[0013] The present invention has at least the following beneficial effects:
[0014] This invention provides an information processing method, device, and medium based on multi-source user data. The method is applied to a first application. In the method, in response to a target user launching the first application, multi-source data of the target user is acquired. The multi-source data includes: the target user's basic personal attribute information, the target user's operational behavior data in the first application, and the second application operation data of the target device. The second application operation data includes the application type and duration of each second application that has been run in the foreground of the target device within a target time period. Data feature vectors are obtained based on the multi-source data of the target user. A model is determined based on the data feature vectors and a sorting algorithm, and a target sorting algorithm is determined from several preset sorting algorithms. The target sorting algorithm is used to determine the video display order of several videos to be displayed, and the videos are displayed to the target user according to the video display order. As can be seen, this invention integrates data from multiple data sources, namely, the target user's basic personal attribute information data, the target user's operational behavior data of the first application, and the target device's second application operation data to generate a data feature vector. This helps to capture the user's multi-layered needs and potential interests. Even if the target user is a new user or a user who uses the first application infrequently, the multi-layered needs and potential interests of the target user can still be captured. Furthermore, the second application operation data includes the application type and the duration of each second application that has been run in the foreground of the target device within the target time period. The end time of the target time period is the current time. Therefore, the second application operation data can reflect the target user's current intention or purpose in launching the application, so that the data feature vector can reflect the target user's current intention or purpose in launching the application. Furthermore, based on the data feature vector and the ranking algorithm, a model is determined, and a target ranking algorithm is determined from several preset ranking algorithms. This helps to improve the accuracy of the determined target ranking algorithm. Further, the target ranking algorithm is used to determine the video display order of several videos to be displayed, and the videos to be displayed are shown to the target user according to the video display order. This allows videos with higher relevance and stronger matching to be displayed to be presented to the target user first, which helps to improve the user experience. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a flowchart of an information processing method based on multi-source user data provided in an embodiment of the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] It should be noted that the terms "first," "second," etc., in the technical solutions of this invention and the above-described drawings are used to distinguish similar tasks and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0019] Embodiments of the present invention provide an information processing method based on multi-source user data. The method is applied to a first application and includes the following steps: Figure 1 As shown:
[0020] S1. In response to the target user launching the first application, acquire the target user's multi-source data; the multi-source data includes: the target user's basic personal attribute information data R1, the target user's operation behavior data for the first application R2, and the target device's second application operation data R3; wherein, R3 includes the application type and foreground dwell time of each second application that has been run in the foreground of the target device within the target time period; the first application is used to display video; the second application is an application other than the first application in the target device, and the target device is the device where the first application is located; the end time of the target time period is the current time point, and the duration of the target time period is a preset duration.
[0021] In a preferred embodiment, when the target user launches the first application, the multi-source data of the target user is obtained and the process proceeds to step S2; this enables the video display order of several videos to be displayed to be determined as quickly as possible, and the several videos to be displayed to the target user according to the video display order; this is beneficial to improving the user experience.
[0022] Specifically, the preset duration is a pre-set duration, such as 10 minutes or 20 minutes, which will not be elaborated here. By setting the preset duration, it is helpful to capture short-term changes in user interests in real time, which is beneficial to improve the timeliness and relevance of the video display order of several videos to be displayed.
[0023] Specifically, the target user is the user corresponding to the account associated with the first application.
[0024] In one specific embodiment, R1 includes at least the target user's age, gender, height, current location, and place of origin.
[0025] In one specific embodiment, R2 includes at least data reflecting the operational behavior of the target user when watching videos using the first application within a historical time period, such as the completion rate of each type of video, the 50% completion rate, etc.
[0026] Specifically, the historical time period refers to the period prior to the current time.
[0027] Furthermore, the end time of the historical time period is earlier than the current time, and the duration of the historical time period is longer than the preset duration.
[0028] In a specific embodiment, the foreground dwell time of a second application that runs in the foreground of the target device within a target time period can be understood as: the duration between the start time of the second application's current foreground running on the target device and the end time of its current foreground running on the target device. For example: if the target time period is from 8:00 to 9:00, and the second application runs in the foreground of the target device from 7:40 to 8:10, then the second application belongs to the second application that runs in the foreground of the target device within the target time period, and the foreground dwell time of the second application is 30 minutes; if the target time period is from 8:00 to 9:00, and the second application runs in the foreground of the target device from 8:20 to 8:30, then the second application belongs to the second application that runs in the foreground of the target device within the target time period, and the foreground dwell time of the second application is 10 minutes.
[0029] Optionally, the first application is a short video application.
[0030] S2. Obtain a data feature vector T based on the multi-source data of the target user, T = (A, B, C); A is the basic attribute feature obtained based on R1, B is the first operational behavior feature obtained based on R2, and C is the second operational behavior feature obtained based on R3.
[0031] Specifically, step S2 includes the following steps S21-S22:
[0032] S21. Obtain the foreground application related data list G = (G1, G2, ..., G3) from R3. i , ..., G m ), G i =(G i1 G i2 );G i For the i-th foreground application-related data group, 1≤i≤m, where m is the preset number of foreground application-related data groups; G i1 To determine the application type of the i-th foreground application after sorting all foreground applications in ascending order of their start time in the foreground on the target device; a foreground application is the second application that runs in the foreground on the target device within the target time period; G i2 With G i1 The foreground duration of the corresponding foreground application; for example: if the second applications running in the foreground on the target device during the target time period are application 1, application 2, and application 3; where application 1 starts running in the foreground on the target device at 8:00; application 2 starts running in the foreground on the target device at 8:20; and application 3 starts running in the foreground on the target device at 8:40; then, application 1, application 2, and application 3 are all foreground applications, and application 3 is the application type of the first foreground application after sorting all foreground applications in the order of their start times in the foreground on the target device from latest to earliest; application 2 is the application type of the second foreground application after sorting all foreground applications in the order of their start times in the foreground on the target device from latest to earliest; and application 1 is the application type of the third foreground application after sorting all foreground applications in the order of their start times in the foreground on the target device from latest to earliest; that is, G 11 For application type 3, G 12 Foreground dwell time of application 3; G 21 For application type 2, G 22 Foreground dwell time of application 2; G 31 For application type 1, G 32 The duration of time application 1 remains in the foreground.
[0033] Specifically, in step S21, if k < m, then G is determined. x1 NULL, G x2 =0; where k is the number of second applications that ran in the foreground of the target device during the target time period; G x1To determine the application type of the x-th foreground application after sorting all foreground applications in ascending order of their start time in the foreground on the target device; G x2 For G x1 The foreground dwell time of the corresponding foreground application; k+1≤x≤m.
[0034] Through the above steps, when k < m, after sorting all foreground applications in ascending order of their start time on the target device's foreground, there is no x-th foreground application. Therefore, setting G to determine... x1 NULL, G x2 =0, to prevent errors or anomalies in the data groups related to the foreground application.
[0035] S22. Input G into the RNN feature extraction module to obtain C; C = (C1, C2, ..., C g , ..., C h) C g Let g be the g-th second operation behavior feature value in C, 1≤g≤h, where h is the number of second operation behavior feature values in C.
[0036] In one specific embodiment, other feature extraction modules capable of capturing temporal dynamics and contextual information in sequence data can be used instead of the RNN feature extraction module in step S22 to obtain C.
[0037] Through the above steps, since there are time dependencies between the various front-end application-related data groups in the front-end application-related data list, it is necessary to use an RNN feature extraction module that can capture the temporal dynamics and contextual information in the sequence data to obtain the second operation behavior features, so that the second operation behavior features can more accurately reflect the characteristics of the various front-end application-related data groups in the front-end application-related data list.
[0038] In one specific embodiment, step S2 further includes:
[0039] Input R1 into the first feature extraction module to obtain A.
[0040] Specifically, the first feature extraction module is a module capable of extracting features from personal basic attribute information data. Those skilled in the art can select a suitable feature extraction module as the first feature extraction module according to the technical purpose of this solution; for example, a feature extraction module capable of extracting text features, such as the feature extraction module in the LSTM model, can be used as the initial module and the first feature extraction module can be obtained through training.
[0041] In one specific embodiment, step S2 further includes:
[0042] R2 is input into the second feature extraction module to obtain B.
[0043] Specifically, the second feature extraction module is a module capable of extracting features from operational behavior feature data. Those skilled in the art can select a suitable feature extraction module as the second feature extraction module according to the technical purpose of this solution; for example, a feature extraction module capable of extracting behavioral features, such as the feature extraction module in the LSTM model, can be used as the initial module and the second feature extraction module can be obtained through training.
[0044] Specifically, A = (A1, A2, ..., A...) j , ..., A n ), A j Let be the j-th basic attribute feature value in A, 1≤j≤n, where n is the number of basic attribute feature values in A.
[0045] Specifically, B = (B1, B2, ..., B e , ..., B f ), B e Let f be the e-th first operation behavior feature value in B, where 1≤e≤f, and f is the number of first operation behavior feature values in B.
[0046] Specifically, A j B e With C g All data are in numeric format, representing A. j B e With C g It can be used in mathematical operations.
[0047] S3. Based on T and the sorting algorithm, determine the model and select the target sorting algorithm from several preset sorting algorithms.
[0048] Specifically, step S3 includes the following steps S31-S35:
[0049] S31. Input G into the target linear regression model to obtain the intention influence weight β corresponding to C; where 0≤β≤1.
[0050] Specifically, the target linear regression model is a linear regression model pre-trained using relevant data by relevant personnel. It is used to determine the magnitude of the impact of the target user's actions on the second application before launching the first application on the target user's current video viewing intention. Those skilled in the art can determine the specific training method according to their needs, and this embodiment does not limit it.
[0051] Specifically, β is used to measure the impact of the target user's actions on the second application before launching the first application on the target user's current video viewing intention; the larger β is, the greater the impact of the target user's actions on the second application before launching the first application on the target user's current video viewing intention; and vice versa.
[0052] S32. Based on β, obtain the preset influence weight α corresponding to A and the preset influence weight γ corresponding to B, where α and γ meet the following conditions:
[0053] α=γ=(1-β) / 2.
[0054] S33, Let A j =A j ×α、B e =B e ×γ、C g =C g ×β is used to update T and use the updated T as the target feature vector MT.
[0055] S34. Input MT into the ranking algorithm determination model to obtain the target ranking algorithm label D corresponding to the target user's current video viewing intention; D∈(P1, P2, ..., P... r ..., P s ), P r Let be the sorting algorithm label for the r-th preset sorting algorithm, where 1 ≤ r ≤ s, and s is the number of preset sorting algorithms.
[0056] Specifically, the target user's current video viewing intent can be understood as: the purpose for which the target user is currently watching the video, such as shopping, relaxing, learning, etc.
[0057] Specifically, the ranking algorithm determines the model as a neural network model trained for the target ranking algorithm label acquisition task.
[0058] Specifically, the preset sorting algorithm is a pre-set sorting algorithm. Different preset sorting algorithms have different operation logic and parameters, which will not be elaborated here.
[0059] S35. Use the preset sorting algorithm corresponding to D as the target sorting algorithm.
[0060] Through the above steps, based on the relevant data list of the foreground application and the target linear regression model, the intent influence weight corresponding to the second operation behavior feature is obtained. This intent influence weight measures the magnitude of the impact of the target user's operation on the second application before launching the first application on the target user's current video viewing intent. Based on the intent influence weight, preset influence weights corresponding to the basic attribute features and the first operation behavior feature are obtained. The data feature vector is updated based on the intent influence weights corresponding to the second operation behavior feature, the basic attribute features, and the preset influence weights corresponding to the first operation behavior feature. The updated data feature vector is used as the target feature vector, comprehensively considering the second operation behavior. The influence of features, basic attribute features, and first operational behavior features on the target user's current video viewing intention can also be understood as comprehensively considering the influence of second operational behavior features, basic attribute features, and first operational behavior features on the target user's current video viewing purpose or interest, which is conducive to improving the accuracy of the target feature vector. Inputting the target feature vector into the ranking algorithm determination model to obtain the target ranking algorithm label corresponding to the target user's current video viewing intention, and using the preset ranking algorithm corresponding to the target ranking algorithm label as the target ranking algorithm to achieve dynamic selection of the ranking algorithm, is conducive to improving the accuracy of the obtained target ranking algorithm, which can improve the recommendation effect of different user groups and improve the user experience.
[0061] S4. Use the target sorting algorithm to determine the video display order of several videos to be displayed, and display the several videos to be displayed to the target user according to the video display order.
[0062] Through the above steps, in response to the target user launching the first application, multi-source data of the target user is acquired. This multi-source data includes: the target user's basic personal attribute information, the target user's operational behavior data regarding the first application, and the target device's second application operational data. The second application operational data includes the application type and duration of each second application run in the foreground of the target device within the target time period. A data feature vector is obtained based on the target user's multi-source data. By integrating data from multiple data sources—that is, fusing the target user's basic personal attribute information, the target user's operational behavior data regarding the first application, and the target device's second application operational data—a data feature vector is generated. This helps to capture the user's multi-layered needs and potential interests. Even if the target user is a new user or a user who uses the first application infrequently, the multi-layered needs and potential interests of the target user can still be captured. Interests; and the second application operation data includes the application type and duration of each second application that ran in the foreground of the target device within the target time period; the end time of the target time period is the current time; therefore, the second application operation data can reflect the target user's current intention or purpose in launching the application, so that the data feature vector can reflect the target user's current intention or purpose in launching the application; based on the data feature vector and the sorting algorithm, a model is determined, and a target sorting algorithm is determined from several preset sorting algorithms; it can realize dynamic selection of sorting algorithms, which is beneficial to improving the accuracy of the obtained target sorting algorithm; furthermore, the target sorting algorithm is used to determine the video display order of several videos to be displayed, and several videos to be displayed are displayed to the target user according to the video display order; so that videos with higher relevance and stronger matching degree are presented to the target user first, which is beneficial to improving the user experience.
[0063] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store a computer program related to implementing a method in the method embodiments, the computer program being loaded and executed by the processor to implement the method provided in the above embodiments.
[0064] Embodiments of the present invention also provide an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method provided in the above embodiments.
[0065] Embodiments of the present invention also provide a computer program product including program code, which, when the program product is run on an electronic device, causes the electronic device to perform the steps of the methods described above in various exemplary embodiments of the present invention.
[0066] This invention provides an information processing method, device, and medium based on multi-source user data. The method is applied to a first application. In the method, in response to a target user launching the first application, multi-source data of the target user is acquired. The multi-source data includes: the target user's basic personal attribute information, the target user's operational behavior data in the first application, and the second application operation data of the target device. The second application operation data includes the application type and duration of each second application that has been run in the foreground of the target device within a target time period. Data feature vectors are obtained based on the multi-source data of the target user. A model is determined based on the data feature vectors and a sorting algorithm, and a target sorting algorithm is determined from several preset sorting algorithms. The target sorting algorithm is used to determine the video display order of several videos to be displayed, and the videos are displayed to the target user according to the video display order. As can be seen, this invention integrates data from multiple data sources, namely, the target user's basic personal attribute information data, the target user's operational behavior data of the first application, and the target device's second application operation data to generate a data feature vector. This helps to capture the user's multi-layered needs and potential interests. Even if the target user is a new user or a user who uses the first application infrequently, the multi-layered needs and potential interests of the target user can still be captured. Furthermore, the second application operation data includes the application type and the duration of each second application that has been run in the foreground of the target device within the target time period. The end time of the target time period is the current time. Therefore, the second application operation data can reflect the target user's current intention or purpose in launching the application, so that the data feature vector can reflect the target user's current intention or purpose in launching the application. Furthermore, based on the data feature vector and the ranking algorithm, a model is determined, and a target ranking algorithm is determined from several preset ranking algorithms. This helps to improve the accuracy of the determined target ranking algorithm. Further, the target ranking algorithm is used to determine the video display order of several videos to be displayed, and the videos to be displayed are shown to the target user according to the video display order. This allows videos with higher relevance and stronger matching to be displayed to be presented to the target user first, which helps to improve the user experience.
[0067] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.
Claims
1. An information processing method based on multi-source user data, the method being applied to a first application, characterized in that, The method includes the following steps: S1. In response to the target user launching the first application, acquire the target user's multi-source data; the multi-source data includes: the target user's basic personal attribute information data R1, the target user's operation behavior data for the first application R2, and the target device's second application operation data R3; wherein, R3 includes the application type and foreground dwell time of each second application that has been running in the foreground of the target device within the target time period; the first application is used to display video; the second application is an application other than the first application in the target device, and the target device is the device where the first application is located; the end time of the target time period is the current time point, and the duration of the target time period is a preset duration; S2. Obtain a data feature vector T, T=(A, B, C), based on the multi-source data of the target user; A is the basic attribute feature obtained based on R1, B is the first operational behavior feature obtained based on R2, and C is the second operational behavior feature obtained based on R3; including the following steps S21-S22: S21. Obtain the foreground application related data list G=(G1, G2, ..., G3) from R3. i , ..., G m ), G i =(G i1 G i2 );G i For the i-th foreground application-related data group, 1≤i≤m, where m is the preset number of foreground application-related data groups; G i1 To determine the application type of the i-th foreground application after sorting all foreground applications in ascending order of their start time in the foreground on the target device; a foreground application is the second application that runs in the foreground on the target device within the target time period; G i2 With G i1 The foreground dwell time of the corresponding foreground application; S22. Input G into the RNN feature extraction module to obtain C; C = (C1, C2, ..., C...). g , ..., C h ), C g Let g be the g-th second operation behavior feature value in C, 1≤g≤h, where h is the number of second operation behavior feature values in C; S3. Based on T and the sorting algorithm, determine the model and select the target sorting algorithm from several preset sorting algorithms; S4. Use the target sorting algorithm to determine the video display order of several videos to be displayed, and display the several videos to be displayed to the target user according to the video display order.
2. The information processing method based on multi-source user data according to claim 1, characterized in that, A = (A1, A2, ..., A...) j , ..., A n A j Let be the j-th basic attribute feature value in A, 1≤j≤n, where n is the number of basic attribute feature values in A.
3. The information processing method based on multi-source user data according to claim 2, characterized in that, B = (B1, B2, ..., B) e , ..., B f ), B e Let f be the e-th first operation behavior feature value in B, where 1≤e≤f, and f is the number of first operation behavior feature values in B.
4. The information processing method based on multi-source user data according to claim 3, characterized in that, Step S3 includes the following steps S31-S35: S31. Input G into the target linear regression model to obtain the intention influence weight β corresponding to C; where 0≤β≤1; S32. Based on β, obtain the preset influence weight α corresponding to A and the preset influence weight γ corresponding to B, where α and γ meet the following conditions: α=γ=(1-β) / 2; S33, Let A j =A j ×α、B e =B e ×γ、C g =C g ×β so that T is updated and the updated T is used as the target feature vector MT; S34. Input MT into the ranking algorithm determination model to obtain the target ranking algorithm label D corresponding to the target user's current video viewing intention; D∈(P1, P2, ..., P... r ..., P s ), P r Let s be the sorting algorithm label for the r-th preset sorting algorithm, where 1 ≤ r ≤ s, and s is the number of preset sorting algorithms; S35. Use the preset sorting algorithm corresponding to D as the target sorting algorithm.
5. The information processing method based on multi-source user data according to claim 1, characterized in that, In step S21, if k < m, then G is determined. x1 NULL, G x2 =0; where k is the number of second applications that have run in the foreground of the target device during the target time period; G x1 To determine the application type of the x-th foreground application after sorting all foreground applications in ascending order of their start time in the foreground on the target device; G x2 For G x1 The foreground dwell time of the corresponding foreground application; k+1≤x≤m.
6. The information processing method based on multi-source user data according to claim 1, characterized in that, R1 includes at least the target user's age, gender, height, current location, and place of origin.
7. The information processing method based on multi-source user data according to claim 4, characterized in that, β is used to measure the impact of the target user's actions on the second application before launching the first application on the target user's current video viewing intention; The larger β is, the greater the impact of the target user's actions on the second application before launching the first application on the target user's current video viewing intention.
8. A non-transitory computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by a processor to implement the information processing method based on user multi-source data as described in any one of claims 1-7.
9. An electronic device, comprising: A processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the information processing method based on user multi-source data as described in any one of claims 1-7.