A personalized advertisement delivery method based on user portrait

Abstract

The application discloses a personalized advertisement putting method based on user portrait, comprising: collecting basic data, calculating the total use frequency of software, and dividing the putting characteristic time period; setting a time window, using a sensor to obtain posture data in the time window, calculating the screen rotation frequency, holding posture and device tilt angle change rate, and determining the micro-scene category; putting advertisements based on the generated advertisement putting data; according to the putting characteristic time period, calculating the time length proportion of temporary operation situation and fixed entertainment situation, pre-setting the advertisement putting strategy based on the time length proportion, dynamically generating advertisement parameters according to posture and motion inertia, realizing more refined advertisement putting, using the situation proportion of the same type period in history to actively pre-adjust the initial strategy of the next period, realizing behavior continuity utilization, adapting to the user behavior mode coming soon in advance, improving the coherence and effectiveness of advertisement putting, and improving the accuracy and user experience of advertisement putting.

Description

Technical Field

[0001] This invention relates to the field of advertising delivery technology, and in particular to a personalized advertising delivery method based on user profiles. Background Technology

[0002] With the rapid development of internet technology and the widespread adoption of smart devices, digital advertising has become an important means of brand promotion and product marketing. However, traditional advertising methods often employ a "one-size-fits-all" strategy, displaying the same ad content to all users. This approach ignores the differences and personalized needs among users, resulting in poor advertising effectiveness and a damaged user experience. To improve the accuracy and effectiveness of advertising, personalized advertising technology based on user profiles has emerged.

[0003] User profiling, as the foundation of personalized advertising, involves collecting and analyzing multi-dimensional user data (such as device usage time, software usage type, and software usage frequency) to build models of user interests and behaviors. This data not only reflects users' explicit needs but also implies their latent interests and preferences. Through user profiling, advertising systems can more accurately understand users, thereby achieving precise ad delivery.

[0004] However, existing personalized advertising technologies still have many shortcomings. First, most advertising systems rely solely on users' historical behavioral data, neglecting the user's current usage context and state. For example, users' acceptance and preferences for advertisements may differ drastically during busy work hours and leisurely weekends. Second, traditional advertising systems often fail to adequately consider users' privacy needs and social environments. Displaying overly private or awkward advertising content in public places not only degrades the user experience but may also provoke user resentment. Furthermore, with the diversification of smart devices, users are switching between different devices more frequently, and traditional advertising systems often cannot achieve seamless cross-device migration of advertising strategies, forcing users to readjust to different advertising experiences on different devices. Summary of the Invention

[0005] This application provides a personalized advertising method based on user profiles. It dynamically generates advertising parameters based on posture and motion inertia to achieve more refined advertising. By utilizing the proportion of similar scenarios in historical time periods, it proactively adjusts the initial strategy for the next time period to achieve continuous utilization of behavior, adapts in advance to upcoming user behavior patterns, improves the consistency and effectiveness of advertising, and enhances the accuracy of advertising and user experience.

[0006] This application provides a personalized advertising method based on user profiles, including: S101, collect basic data, calculate the total usage frequency of the software, and divide the delivery characteristic time period; wherein, the delivery characteristic time period includes inactive time period, general time period and active time period; the basic data includes device usage time, software usage type and software usage frequency; S102, Set a time window, use sensors to acquire posture data within the time window, calculate the screen rotation frequency, grip posture and device tilt angle change rate based on the posture data, and determine the micro-scene category; the micro-scene category includes temporary operation scenarios and fixed entertainment scenarios; S103, Generate ad delivery data based on the micro-scenario category determination results, and deliver ads based on the generated ad delivery data; wherein, the ad delivery data includes ad format, maximum display duration, initial volume, interaction method, display frequency limit and whether automatic video playback is allowed. S104. Calculate the time ratio of temporary operation scenarios and fixed entertainment scenarios based on the time period of the ad placement, and pre-set the ad placement strategy based on the time ratio.

[0007] Preferably, the method for determining the micro-scenario category is as follows: when the holding posture is that the device is placed flat on the table, the tilt angle change rate is less than the minimum tilt angle change rate threshold, and the screen is continuously lit for more than a time threshold, it is determined to be a fixed entertainment scenario in the form of desktop viewing; when the holding posture is that the device is held with both hands, the screen rotation frequency is less than the minimum screen rotation frequency threshold, and the tilt angle change rate is less than the minimum tilt angle change rate threshold, it is determined to be a fixed entertainment scenario in the form of handheld immersion; when the holding posture is that the device is held with one hand, the screen rotation frequency is greater than or equal to the maximum screen rotation threshold, and the tilt angle change rate is greater than or equal to the maximum tilt angle change rate threshold, it is determined to be a temporary operation scenario.

[0008] Preferably, the advertising delivery method further includes: S201, collect usage scenario data and operation data, determine the scenario type according to the usage judgment rules of the scenario data, and identify the operation comfort zone based on the operation data; wherein, the usage scenario data includes proximity sensor distance, ambient light data and screen brightness data; the operation data includes operation time, operation coordinates and operation duration; S202, cross-combine scenario types and operational comfort zones to form combined scenarios, adjust the advertising data generated in step S103 according to the combined scenarios, and perform advertising based on the adjusted advertising data.

[0009] Preferably, the steps for determining the scenario type using the judgment rules are as follows: Set a face count threshold, a ratio threshold range, and a distance threshold range. The ratio threshold range includes a maximum ratio threshold and a minimum ratio threshold. The distance threshold range also includes a minimum distance threshold and a maximum distance threshold. If the face count is greater than or equal to the preset face count threshold, it indicates that the front-facing camera has detected at least two different face outlines, and the user is necessarily sharing the screen with others. This is determined to be shared browsing in a multi-person scenario. If the face count is less than the preset face count threshold, the proximity sensor distance is less than the minimum distance threshold, and the screen brightness to ambient light ratio is greater than the maximum ratio threshold, it is determined to be private browsing. If the proximity sensor distance is greater than the maximum distance threshold and the screen brightness to ambient light ratio is less than the minimum ratio threshold, it is determined to be shared browsing with the device placed horizontally at a distance. If the proximity sensor distance is greater than or equal to the minimum distance threshold and less than or equal to the maximum distance threshold, or the screen brightness to ambient light ratio is greater than or equal to the minimum ratio threshold and less than or equal to the maximum ratio threshold, it is determined to be a normal scenario.

[0010] Preferably, the step of identifying the operating comfort zone based on operational data includes calculating the coordinates of the hot zone center and the discrete radius of the hot zone, determining the dominant hand tendency based on the hot zone center coordinates, identifying the fatigue state, and the formula for calculating the discrete radius of the hot zone based on the hot zone center coordinates is: ,in, Let M be the discrete radius of the hot zone, M be the total number of operation points in the current operation point queue, and i be the operation point index. , and are the horizontal and vertical coordinates of the screen for the i-th operation point, respectively.

[0011] Preferably, the dominant hand preference is determined based on the coordinates of the center of the hot zone and the screen width W. < This indicates that most touches occurred in the left third of the screen, suggesting left-hand operation; if > This indicates that most touches occurred in the right third of the screen, suggesting right-hand operation; if < < , The center position suggests that the operation was performed with both hands or the index finger.

[0012] Preferably, the advertising delivery method further includes: S301 identifies devices under the same account, builds a proximity sensing network, and identifies continuous patterns across devices based on the proximity sensing network; S302: Calculate contextual confidence and inheritance weight based on the cross-device continuity pattern. Determine whether to perform migration based on contextual confidence and inheritance weight. If the determination result is yes, implement the advertising strategy migration and deliver the advertisement; otherwise, do not perform migration.

[0013] Preferably, the method for constructing a proximity sensing network consists of three steps: local area network heartbeat broadcasting, Bluetooth proximity scanning, and spatial determination of proximity devices. The spatial determination of proximity devices is as follows: when device A and device B can detect each other in two consecutive scans, and the average value of their Bluetooth signal strength is greater than the strength threshold, then A and B are considered to be in the same physical space.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages: Dynamically generating advertising parameters based on posture and motion inertia to achieve more refined advertising delivery; proactively adjusting the initial strategy for the next period by utilizing the proportion of similar historical time periods to achieve continuous use of behavior, adapting in advance to upcoming user behavior patterns, improving the coherence and effectiveness of advertising delivery, and enhancing the accuracy and user experience of advertising delivery; identifying private / shared browsing to adjust parameters such as advertising content and volume according to different privacy scenarios, avoiding embarrassing advertising in public places, and improving the user's social experience; recognizing hand operation comfort zones and fatigue zones to dynamically adjust the position and size of advertising interaction buttons according to the user's hand operation state, reducing operation difficulty, protecting user privacy, improving operation convenience, and adapting to hand fatigue; and more accurately determining when to trigger advertising strategy migration and the degree of inheritance of migration parameters by the target device through multi-device Bluetooth proximity sensing and cross-device continuous pattern recognition, achieving seamless and adaptive cross-device migration of advertising strategies, avoiding users experiencing cold starts or inappropriate advertising experiences after switching devices, and improving the accuracy and user satisfaction of advertising delivery. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a personalized advertising delivery method based on user profiles according to the present invention. Figure 2 This is a flowchart illustrating the process of forming a combined scenario according to the present invention; Figure 3 This is a schematic diagram of the cross-device ad migration process of the present invention. Detailed Implementation

[0016] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0017] It should be noted that the terms "vertical," "horizontal," "up," "down," "left," "right," and similar expressions used in this article are for illustrative purposes only and do not represent the only possible implementation.

[0018] 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 this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0019] Example 1: Figure 1 This is a flowchart illustrating a personalized advertising delivery method based on user profiles according to an embodiment of the present invention, including: S101, collect basic data, calculate the total usage frequency of the software, and divide the delivery characteristic time periods; among which, the delivery characteristic time periods include inactive time periods, general time periods, and active time periods; Specifically, basic data is collected, including device usage time, software usage type, and software usage frequency. This is achieved by recording the device's unlock and lock timestamps, calculating the difference between the lock and unlock timestamps as the device usage time, and categorizing applications on the device into social entertainment, office tools, lifestyle services, and education / culture categories. Software usage type is identified based on the software used. For software usage frequency, the total number of foreground switching events within each time window (every hour) is counted, and then divided by the duration of that time window (in hours) to obtain the software's usage frequency within that time period. The total usage frequency of all software is then calculated based on the software's usage frequency using the following formula: ,in, This represents the total usage frequency of the software within the window. The total number of all software installed on this device. To determine the usage frequency of the k-th software within the i-th time window, inactive and active frequency thresholds are set based on user group behavior characteristics. All time windows throughout the day are iterated through, and characteristic time periods are divided according to the relationship between the total usage frequency of the time window and the inactive and active frequency thresholds. Specifically: when the total software usage frequency of a time window is less than the inactive frequency threshold, the window is classified as an inactive time window; when the total software usage frequency of a time window is greater than or equal to the inactive frequency threshold but less than the active frequency threshold, the window is classified as a normal time window; when the total software usage frequency of a time window is greater than or equal to the active frequency threshold, the window is classified as an active time window. Adjacent time windows of the same type are merged into a continuous time period. For example, multiple consecutive 15-minute windows from 02:00 to 05:00 are marked as inactive and merged into a 3-hour inactive time period. The merged time period has a start and end time. For isolated windows with extremely short durations, they are merged into adjacent longer time periods to avoid overly fragmented time period divisions. S102, Set a time window, use sensors to acquire posture data within the time window, calculate the screen rotation frequency, grip posture and device tilt angle change rate based on the posture data, and determine the micro-scenario category; the micro-scenario category includes temporary operation scenarios and fixed entertainment scenarios; Furthermore, a sliding time window mechanism is adopted to collect sensor data. The time window is set by setting the window length, sliding step size, and sampling interval. The time window length is set to 6, meaning that each window contains 6 consecutive sampling points. The total duration of the window is the window length multiplied by the sampling interval. The sliding step size is set to 0.5 seconds, meaning that the window slides forward once every 0.5 seconds (synchronized with the sampling interval). That is, when a new data point is collected, the oldest data point in the window is removed to form a new window. The sampling interval is set to 0.5 seconds, meaning that the system collects data from the device sensors once every 0.5 seconds. At each sampling moment, the system obtains acceleration data, screen orientation status data, and angular velocity data from the accelerometer, rotation vector sensor, and gyroscope, respectively. The acceleration data includes linear acceleration values ​​in the X, Y, and Z axes, which are used to determine the device's orientation, tilt angle, and grip posture. The screen orientation status data records whether the current screen is portrait, landscape left (90 degrees), landscape right (-90 degrees), or inverted. The angular velocity data is used to assist in calculating the tilt angle change rate.

[0020] The screen rotation frequency is calculated based on the collected posture data, using the following formula: ,in, Screen rotation frequency, measured in Hertz (Hz), represents the number of times the screen orientation changes per unit of time, used to quantify how frequently the user flips the device. This is the number of times the screen orientation changes within a time window. Each change from the current screen orientation to another different orientation (from portrait to landscape, or from landscape to portrait) is counted as one change. Two consecutive changes in the same orientation are not counted. N is the number of sampling points contained within the sliding window, i.e., N=6. The sampling interval is in seconds. In this scheme, =0.5 seconds. Based on the above screen rotation frequency, a screen rotation threshold is set. If the screen rotation frequency is greater than or equal to the maximum screen rotation threshold, it is judged as high-frequency rotation, indicating that the user is frequently flipping the device, which usually occurs during movement or temporary operation scenarios. The grip posture is calculated by collecting three-axis acceleration data through an accelerometer and calculating the average values ​​of the X, Y, and Z axes using the following formula: ,in, This is the average value of all sampling points within the current sliding window of the accelerometer along the X-axis, with the same units as acceleration. This represents the number of sampling points contained within the sliding window. This is the raw measurement value of the accelerometer on the X-axis at the i-th sampling time, in m / s². 2 , Using the sampling point index, the average value of the sampling points on the Y-axis and Z-axis is calculated using the same method based on the above formula, resulting in... and The variances of the X, Y, and Z axes are calculated based on the average values ​​of the sampled points, using the following formula: ,in, The variance of the accelerometer along the X-axis represents the degree of dispersion of the X-axis acceleration data within the current sliding window. This represents the average value of all sampling points within the current sliding window of the accelerometer along the X-axis. This represents the number of sampling points contained within the sliding window. For the i-th sampling time, the original measurement value of the accelerometer on the X-axis is used. Based on the above formula, the variances of the Y-axis and Z-axis are calculated using the same method to obtain... and Based on the variances of the X, Y, and Z axes, the maximum and second largest variances are identified, and the ratio of the maximum to the second largest variance is calculated. Simultaneously, the arithmetic mean of the three-axis variances is calculated. An arithmetic mean threshold and a ratio threshold range are set. The ratio threshold range includes a minimum ratio threshold and a maximum ratio threshold. If the arithmetic mean is less than the preset arithmetic mean threshold, it indicates that the acceleration variances of the three axes are very small, and the device is judged to be placed flat on the table, i.e., H=0. If the ratio is greater than or equal to the maximum ratio threshold and the axis corresponding to the maximum variance is the Z-axis, it indicates that the device's fluctuation in the Z-axis direction is much greater than that of the other two axes, and the device is judged to be held with one hand, i.e., H=1. If the ratio is less than or equal to the minimum ratio threshold, it indicates that the variances of the three axes are close to each other, and the device is judged to be held with both hands, i.e., H=2. The tilt angle change rate is calculated based on the acquired angle data. The formulas for calculating the pitch angle and roll angle are: , ,in, Let be the pitch angle (around the X-axis) of the i-th sampling point. For the i-th sampling time, the accelerometer measurement value on the X-axis is... For the i-th sampling time, the accelerometer measurement value on the Y-axis is... The accelerometer reading on the Z-axis is the value measured at the i-th sampling time. The arctangent function is used to calculate the angle from the ratio of acceleration components. Let be the conversion factor for radians to angles, and i be the sampling point index. The sum of the absolute values ​​of the pitch and roll angle changes is calculated based on the pitch and roll angles. The formula is: , ,in, The sum of the absolute values ​​of the pitch angle changes between adjacent sampling points is given by N, where N is the number of sampling points contained within the sliding window, and i is the index of the sampling point. The roll angle change rate is calculated based on the sum of the absolute values ​​of the pitch angle changes and roll angle changes, using the following formula: ,in, A tilt angle threshold range is set for the device tilt angle change rate, which is the average rate of change of pitch angle and roll angle. The tilt angle change rate threshold range includes a maximum tilt angle change rate threshold and a minimum tilt angle change rate threshold. When the tilt angle change rate is greater than or equal to the maximum tilt angle change rate threshold, it is judged as a large tilt angle change, which usually occurs when walking, running or moving the device quickly. When the tilt angle change rate is less than the minimum tilt angle change rate threshold, it is defined as a stable tilt angle, which usually occurs when the device is placed on a table or when the user holds it stably for viewing.

[0021] When the grip posture H=0, the tilt angle change rate is less than the minimum tilt angle change rate threshold, and the screen remains lit for more than the time threshold, it is determined to be a fixed entertainment scenario in desktop viewing mode; when the grip posture H=2 (two-handed grip), the screen rotation frequency is less than the minimum screen rotation frequency threshold, and the tilt angle change rate is less than the minimum tilt angle change rate threshold, it is determined to be a fixed entertainment scenario in handheld immersive mode; when the grip posture H=1 (one-handed grip), the screen rotation frequency is greater than or equal to the maximum screen rotation threshold, and the tilt angle change rate is greater than or equal to the maximum tilt angle change rate threshold, it is determined to be a temporary operation scenario.

[0022] S103, Generate ad delivery data based on the micro-scenario category determination results, and deliver ads based on the generated ad delivery data; wherein, the ad delivery data includes ad format, maximum display duration, initial volume, interaction method, display frequency limit and whether automatic video playback is allowed. Furthermore, based on the micro-scenario category determination results, corresponding advertising delivery data is generated. This data includes the ad format, maximum display duration, initial volume, interaction method, display frequency limit, and whether automatic video playback is allowed. When the determination result is a temporary operation scenario, it is assumed that the user is currently in a state of rapid movement, one-handed operation, or distraction. The ad format is set to banner ads and static image ads that can be skipped after 0 seconds of screen opening, prohibiting full-screen ads, interstitial ads, and rewarded video ads. The maximum display duration is set to 5 seconds. Simultaneously, a skip button is displayed within or at the edge of the ad area at the same time the ad appears (i.e., at second 0), and the user immediately closes the ad after clicking it. The initial volume is set to completely silent to avoid interfering with users in public or noisy environments. The interaction method disables all interaction methods requiring mobile devices or large gestures, supporting only click interactions. The display frequency limit is set to 1, meaning that in a temporary operation scenario, the maximum display frequency is 1 every two minutes. Only one ad will be displayed at a time to avoid frequent interruptions; autoplay of videos is not allowed; when the judgment result is a fixed entertainment scenario, assuming the user is currently in a stable and focused state, the ad format will be set to full-screen video ad, landscape immersive ad, and interactive ad, without restriction on banner or interstitial format; the maximum display time is set to 45 seconds, but a skip button must be provided within the first 5 seconds of ad playback, which the user can click to skip at any time. If the user does not skip, the ad will continue to play until the 45-second mark; the initial volume is set between 40% and 60%, with the specific value determined by the device's ambient light sensor. When the ambient light is bright (such as outdoors during the day), the volume will automatically increase to close to 60%, and when the ambient light is dim (such as in a room at night), the volume will automatically decrease to close to 40%; the interaction method is to activate all immersive interactive functions, including shake, gravity sensing, and sliding view, with the maximum display frequency set to once every 30 seconds; autoplay of videos is allowed.

[0023] Based on the generated ad delivery data, ad delivery is performed as follows: First, a frequency check is conducted by querying historical display records to determine if the display frequency limit has been exceeded within the current time window. If it has, the current delivery is abandoned, ad creatives are not requested, and the process ends directly. If the frequency allows, a request is sent to the ad server, including the ad delivery type for the current time period and whether reverse delivery is needed. The server returns ad creatives that meet the content requirements. The generated six parameters are then bound to the returned ad creatives. For example, if the current operation is temporary, the display duration of the creatives is forcibly shortened or... Limit the ad to 5 seconds and override the built-in volume settings of the ad material (force mute). Also disable any shake-to-play interaction commands that may be included in the ad material. Render the ad at the specified location (top, bottom, or full screen) according to the bound parameters. For banners in temporary operation scenarios, ensure the height does not exceed 80 pixels. Place the skip button in the upper right corner of the banner, with a size of at least 44×44 pixels for easy clicking. For full-screen videos in fixed entertainment scenarios, autoplay but limit the volume to 40%-60%, and display the skip button within 5 seconds before playback. Continuously monitor user interaction behavior during ad display. Based on the above process, complete the precise targeting of the ad.

[0024] S104. Calculate the time ratio of temporary operation scenarios and fixed entertainment scenarios based on the time period of the ad placement, and pre-set the ad placement strategy based on the time ratio.

[0025] Specifically, within each targeted time period, the system is further divided into several basic statistical units. Starting from the beginning of the time period, a continuous statistical unit is formed every 5 minutes. If the last unit is less than 5 minutes, it is recorded according to the actual duration. Within each statistical unit, the duration of users in temporary operation scenarios and fixed entertainment scenarios is recorded in real time. Based on the statistically recorded duration of temporary operation scenarios and fixed entertainment scenarios, the percentage of time spent in temporary operation scenarios and the percentage of time spent in fixed entertainment scenarios are obtained respectively. Specifically, the total number of seconds of temporary operation within the unit is divided by the total duration of the basic statistical unit. Similarly, the total number of seconds of fixed entertainment within the unit is divided by the total duration of the basic statistical unit. Before each targeted time period is about to begin, the system executes pre-adjustment logic, performing cross-day comparisons for the same type of targeted time periods, and pre-adjusting the strategy based on the percentage of time spent in temporary operation scenarios and the percentage of time spent in fixed entertainment scenarios. For temporary operation scenarios, if the duration of temporary operation scenarios accounts for more than 0.7% of the total time, it means that users spent more than 70% of their time in temporary operation scenarios in the previous time period of the same type. It is inferred that users will continue this high-frequency mobile behavior pattern in the current time period. Therefore, for a period of time after the start of the current time period, the advertising parameters for temporary operation scenarios are forcibly adopted. If the duration of fixed entertainment scenarios accounts for more than 0.8% of the total time, it means that users spent more than 80% of their time in fixed entertainment scenarios in the previous time period of the same type. It is inferred that users are likely to continue this stable and focused state in the current time period. Therefore, for a period of time after the start of the current time period, an enhanced fixed entertainment strategy is adopted. Specifically, the upper limit of the ad display frequency is temporarily increased from once every 30 seconds to once every 20 seconds to take advantage of users' high level of focus and increase ad exposure opportunities, prioritizing the display of brand ads with a duration of more than 15 seconds.

[0026] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: dynamically generating advertising parameters based on posture and motion inertia to achieve more refined advertising placement; proactively adjusting the initial strategy for the next period by utilizing the proportion of similar historical time periods to achieve continuous utilization of behavior; adapting in advance to upcoming user behavior patterns; improving the coherence and effectiveness of advertising placement; and improving the accuracy of advertising placement and user experience.

[0027] Example 2: The delivery method in Example 1 above did not differentiate between user scenarios (private / shared), making it easy to deliver awkward content in public places. Furthermore, the fixed position of the ad interaction button did not consider user hand operating habits (such as dominant hand, one-handed operation) and fatigue levels, leading to difficulty in operating large-screen devices and a high rate of accidental touches. This example dynamically adjusts the position and size of the ad interaction button by calculating the center of the heat zone and the operating radius, and identifies hand fatigue levels, improving the stability of ad reception fluctuation parameters. Figure 2 As shown.

[0028] S201, collect usage scenario data and operation data, determine the scenario type according to the usage judgment rules of the scenario data, and identify the operation comfort zone based on the operation data; wherein, the usage scenario data includes proximity sensor distance, ambient light data and screen brightness data; the operation data includes operation time, operation coordinates and operation duration; Specifically, a proximity sensor (VCNL4020) is used to measure the proximity distance, which is the distance between the object and the screen. An ambient light sensor (TSL2586) is used to detect ambient light data, i.e., the light intensity of the surrounding environment. The screen brightness value is collected through the system brightness management interface. Based on the ambient light intensity and the screen brightness value, the screen brightness to ambient light ratio is calculated using the following formula: ,in, The screen brightness to ambient light ratio reflects whether the user manually increases the screen brightness in low-light environments or whether the screen is relatively dim in bright environments. This is the screen brightness value. The ambient light intensity is set to 0.1 to prevent division by zero errors when the ambient light is zero. The device's camera detects the maximum number of faces within the window (faces may briefly enter and exit the frame). Based on the ratio of screen brightness to ambient light and the number of faces, a judgment rule is used to determine the scenario. This rule sets a face count threshold, a ratio threshold range, and a distance threshold range. The ratio threshold range includes a maximum and a minimum ratio threshold. Similarly, the distance threshold range also has a minimum and a maximum distance threshold. If the number of faces is greater than or equal to the preset face count threshold, it indicates that the front-facing camera has detected at least two different face outlines, meaning the user is sharing the screen with others. This is thus determined as screen sharing. For browsing (multi-person scenarios), P=0, this rule has the highest priority. If the number of faces is less than the preset face number threshold, the proximity sensor distance is less than the minimum distance threshold, and the screen brightness to ambient light ratio is greater than the maximum ratio threshold, it is determined to be private browsing, and P=1. If the proximity sensor distance is greater than the maximum distance threshold and the screen brightness to ambient light ratio is less than the minimum ratio threshold, it is determined to be shared browsing (far-distance flat placement). If the proximity sensor distance is greater than or equal to the minimum distance threshold and less than or equal to the maximum distance threshold, or the screen brightness to ambient light ratio is greater than or equal to the minimum ratio threshold and less than or equal to the maximum ratio threshold, it means that both the distance and brightness ratio are in the fuzzy range, and it is impossible to clearly determine whether it is private or shared. This is recorded as a normal scenario, and P=2.

[0029] Operation data is acquired through the operation event interface, and each operation is recorded in real time. This operation data includes the operation time, operation coordinates, and operation duration. Operation time refers to the start time of the operation. Operation coordinates are defined with the top-left corner of the screen as the origin, with x-axis pointing right and y-axis pointing down. The operation duration is the number of milliseconds elapsed from pressing the finger to releasing it. For swipe operations, typically only the total duration from pressing to releasing is recorded. Based on the collected operation coordinates and operation duration, an operation point queue is formed in timestamp order. The steps for identifying the comfort zone based on operation data include calculating the center coordinates and discrete radius of the comfort zone, determining the dominant hand tendency based on the center coordinates, and identifying fatigue state. The center coordinates of the comfort zone are calculated based on the operation coordinates and operation duration using the following formula: , ,in, Here, x and y represent the x and y coordinates of the center of the hot zone. The hot zone refers to the main area of ​​user interaction on the screen, with the center being the centroid of the dense area, reflecting the screen position where the user is currently most focused or interacts most frequently. M is the total number of operation points in the current operation point queue, and i is the operation point index, ranging from 1 to M. , Let x and y be the horizontal and vertical coordinates of the i-th touch point on the screen, respectively. The origin is the top-left corner of the screen, with x positive to the right and y positive downwards. The discrete radius of the hot zone is calculated based on the coordinates of the hot zone center, using the following formula: ,in, The radius of the hot zone represents the dispersion of all operation points around the center of the hot zone. A larger value indicates a more dispersed distribution of user touch points (larger range of finger movement); a smaller value indicates that touch points are concentrated near the center of the hot zone (smaller range of finger movement). M is the total number of operation points in the current operation point queue, and i is the operation point index, from 1 to M. , Let be the horizontal and vertical coordinates of the i-th touch point, respectively. Based on the center coordinates of the hot zone and the screen width W, determine the dominant hand preference. < This indicates that most touches occurred in the left third of the screen, suggesting left-hand operation. =-1; if > This indicates that most touches occurred in the right third of the screen, suggesting right-handed operation. =+1; if < < , Centered, it is inferred that it was operated by both hands or the index finger, remember. =0; Fatigue is identified by calculating the Euclidean distance between the center coordinates of two adjacent hot zones, and simultaneously counting the number of negative feedbacks generated by the user within the corresponding time windows of the two adjacent hot zones. The identification rule is: when the Euclidean distance between the center coordinates of the hot zone is greater than a distance threshold and the number of negative feedbacks is greater than or equal to a number threshold, it is determined to be a state of hand fatigue, recorded as 0. Otherwise, not tired. 0.

[0030] S202, cross-combine scenario types and operational comfort zones to form combined scenarios, adjust the advertising data generated in step S103 according to the combined scenarios, and perform advertising based on the adjusted advertising data.

[0031] Furthermore, when the scenario type is determined to be a normal scenario, the advertising data from step S103 is used directly; when the scenario type is determined to be shared browsing and private browsing, shared browsing and private browsing are combined with hand fatigue status to form four combined scenarios. The first is a combination of private browsing and hand comfort, where the user uses the device alone and operates it easily, making full use of screen space and interactive convenience; the second is a combination of private browsing and hand fatigue, where the user is in a private environment but feels fatigued after operating with one hand for a long time, and needs to reduce the burden of operation; the third is a combination of shared browsing and hand comfort, where the user uses the device in front of others and is in good condition, but needs to avoid social embarrassment and excessive movements; the fourth is a combination of shared browsing and hand fatigue, where the user is already fatigued in a shared environment, and at this time, it is necessary to minimize advertising interference and actively provide fatigue relief options.

[0032] In the first scenario, the ad delivery data is adjusted to allow any private ads, the volume remains normal (40%–60%), and the interactive buttons are dynamically placed near the center of the hot zone (≤50 pixels from the center, offset inwards at the edges). If the hot zone radius is less than 100 pixels, the button size increases by 20%, and fatigue relief is not triggered. In the second scenario, private ads are allowed, the volume is reduced to 20%–30%, and the button layout consists of a main button in the center of the hot zone plus a copy button on each side of the screen (coordinates (50,ȳ) and (W-50,ȳ)). All buttons are 10% larger than C1 (cumulative increase of 30%). In a fixed entertainment scenario, the maximum ad duration is shortened from 45 seconds to 25 seconds, and the skip button is moved to the 3rd second. In the third scenario, the screen is forced to... To block private ads, only general ads will be displayed. The initial volume will be 0% and autoplay videos will be disabled. Users will need to actively click to turn on the sound. The interaction button will be fixed in the lower right corner of the screen and enlarged by 20%. The skip ad button will always be visible and the text will be enlarged. If it is a fixed entertainment scenario, the ad duration will be compressed to within 15 seconds, and large interactions such as shaking and gravity sensing will be disabled. When it is the fourth combination scenario, private ads will be strictly blocked, completely muted, and no option to turn on the sound will be provided. Only the fixed button in the lower right corner will be retained, and its size will be increased by another 30% (cumulative increase of 50%). A tap to close prompt will be displayed. Only static image or text ads are allowed, videos are prohibited, and the display frequency limit will be reduced to once every 3 minutes. At the same time, a fatigue relief prompt will pop up, and users can choose to halve the ad frequency of the same type of time period in the next 30 minutes.

[0033] The ad delivery process based on the adjusted ad delivery data is as follows: Check the historical records based on the maximum display frequency of the current combination (once every 3 minutes for the fourth combination, other combinations use the frequency of the first option). If the maximum has been reached, abandon the current delivery; otherwise, request ad creatives that meet the content requirements from the ad server (no content restrictions for the first and second combinations, but privacy categories must be masked for the third and fourth combinations). After obtaining the creatives, bind all adjusted parameters (such as volume, button position, skip time, etc.) to the creatives, overriding the default settings of the creatives. Then, render the ad in the designated location: For the second combination, display an interactive button identical to the one in the center on each of the left and right sides of the screen; for the fourth combination, display the "Tap to Close" prompt next to the large button in the lower right corner. During the ad display, monitor user interactions such as clicking, skipping, and turning on sound, and record viewing time and negative feedback. Finally, if the user clicks the fatigue relief prompt in the fourth combination, set a 30-minute timer, after which the original ad frequency will be automatically restored.

[0034] The technical solutions described in the above embodiments of this application have at least the following technical effects or advantages: By recognizing private / shared browsing, the content and volume of advertisements are adjusted according to different privacy scenarios to avoid displaying embarrassing advertisements in public places and improve the user's social experience; by recognizing the comfort zone and fatigue zone of hand operation, the position and size of the advertisement interaction button can be dynamically adjusted according to the user's hand operation state, reducing the difficulty of operation, protecting user privacy, improving the convenience of operation, and adapting to hand fatigue state.

[0035] Example 3: While Examples 1 and 2 both involve ad delivery on a single device, this example achieves seamless and adaptive migration of ad strategies across different devices through cross-device migration. This improves ad delivery accuracy and user satisfaction, and solves the cold start and inconsistent ad experience problems faced by users when switching between different devices. Figure 3 As shown.

[0036] S301 identifies devices under the same account, builds a proximity sensing network, and identifies continuous patterns across devices based on the proximity sensing network; Specifically, when each device first installs the Ad Software Development Kit (Ad SDK), the user is required to log in to a unified account. After successful login, the SDK retrieves a list of all registered devices under that account from the server and generates a globally unique device ID for the current device. Each time the device starts the SDK, it reports its current status (online, offline, IP address, etc.) to the server. The goal of building a proximity-aware network is to determine whether two devices under the same account are currently in the same physical space. The method for building a proximity-aware network consists of three steps: local area network heartbeat broadcast, Bluetooth proximity scanning, and spatial determination of nearby devices. The local area network heartbeat broadcast is used for device discovery. Every 30 seconds, each device broadcasts an encrypted heartbeat packet via the local area network. The heartbeat packet contains: device ID, device type, current timestamp, and screen status. Other devices within the same subnet can receive these heartbeat packets. By receiving the heartbeat packets, the device can know which devices share the same account. The devices are currently connected to the same local area network (LAN). However, LAN broadcast alone cannot determine the distance between devices. Bluetooth proximity scanning is used to accurately determine the distance. Each device simultaneously scans for nearby paired Bluetooth devices. The Bluetooth signal strength index (RSSI) is used to determine whether the devices are in the same room. When the RSSI is greater than a strength threshold, the devices are considered to be in the same room and are considered neighboring devices. When the RSSI is less than or equal to the strength threshold, the devices are far apart and not in the same room. For devices that cannot be paired via Bluetooth, the Wi-Fi signal strength from heartbeat packets is used to determine whether they are in the same room. Based on the above determination results, each device generates a neighboring device table. The neighboring device table includes the discovered device ID, device type, the timestamp of the most recent detection, the exponentially weighted average of the most recent three RSSI measurements, and a consecutive hit counter. The rule for determining whether they are in the same physical space is: when devices A and B are in the same physical space, if they are detected in two consecutive scans (with a 30-second interval)... In this process, if both sides can detect each other (i.e., A can receive B's heartbeat packet or Bluetooth signal, and B can also receive A's), and the average RSSI of both sides is greater than the strength threshold, then A and B are considered to be in the same physical space. Based on the above steps, a proximity sensing network is constructed.

[0037] Based on the constructed proximity sensing network, the system monitors the posture change sequences, proximity relationships, and time differences between the source and target devices to identify cross-device continuity patterns. When a phone's posture changes from being held with both hands or one hand to being placed flat on a table, and the TV is marked as very close by the proximity sensing network within 5 seconds before and after this change, while the TV's heartbeat display shows the screen is active, it is determined that the user is projecting phone content onto the TV, which is the phone → TV projection mode. When the TV detects the user leaving via camera, human body sensor, or a sudden drop in Bluetooth RSSI, and the phone is picked up within 10 seconds of leaving (accelerometer detects wrist raising or flipping), it is determined that the user is switching back to the phone from the TV, which is the TV → phone relay mode. When the smartwatch detects… When a wrist is raised, and the phone is picked up within 5 seconds, and the Bluetooth RSSI of both devices exceeds the strength threshold, the user is considered to have switched from the watch to the phone, which is the watch → phone relay mode. When the tablet screen is off for more than 15 seconds and enters standby mode, and the phone is woken up and unlocked within 3 seconds after standby, and the Bluetooth RSSI of both devices exceeds the strength threshold, the user is considered to have switched from the tablet to the phone, which is the tablet → phone switching mode. When the vehicle system detects that the vehicle is turned off (Bluetooth disconnected or the accelerometer is stationary for a long time), and the phone's proximity sensor reading changes from near (<5cm) to far (>15cm, i.e., taken out of a pocket) within 10 seconds after the engine is turned off, the driving is considered to have ended, and the advertising preferences in the vehicle environment need to be reversed to the phone, which is the vehicle → phone reverse migration mode.

[0038] S302, calculate contextual confidence and inheritance weight based on the cross-device continuity pattern, and determine whether to perform migration based on the contextual confidence and inheritance weight. If the determination result is yes, then implement the advertising strategy migration and perform advertising; otherwise, do not perform migration. Furthermore, the source device is the transmitting device across devices, and the target device is the receiving device across devices. Upon detecting a continuous pattern across devices, the current micro-context type (temporary operation or fixed entertainment) is extracted from the source device's memory. The context confidence is calculated based on the current micro-context type of the source device, using the following formula: ,in, The source device scenario confidence score represents the stability of the current micro-scenario (temporary operation or fixed entertainment) of the source device. Its value is equal to the duration of the current micro-scenario. `min` is a function that takes the smaller of 120 and the duration of the current micro-scenario. The formula for calculating the inheritance weight based on the scenario confidence score is: ,in, To inherit weights, the degree to which advertising strategy parameters from the source device are trusted and adopted when migrated to the target device. This represents the upper limit of the weight, meaning the maximum inherited weight is 1. The source device scenario confidence level is 30, and the baseline duration constant is 30 seconds. When the source device scenario duration reaches 30 seconds, the inheritance weight is 1.0. The method for determining whether to perform migration is: only when the source device scenario confidence level is greater than or equal to the time threshold (i.e., the inheritance weight is 1.0). ≥ The system only performs migration when the source device parameters are converted to a form applicable to the target device. After receiving the migration parameters from the source device, the system fuses these migration values ​​with the target device's own current real-time judgment results using fusion rules. The fusion rules are as follows: for discrete parameters, if the inheritance weight is greater than 0.6, the source device's value is directly used to overwrite the target device's value; otherwise, the target device's own judgment is used. For continuous parameters, the weighted average is fused according to the inheritance weight. For count parameters, the source device's value is proportionally added to the target device. After the parameter fusion is completed, the target device uses the updated advertising strategy parameters for advertising. If the source device decides not to migrate due to low context confidence, the target device will run the schemes of Embodiment 1 and Embodiment 2 independently, relying entirely on its own real-time sensor data, without introducing any strategy parameters from the source device.

[0039] The technical solutions in the above embodiments of this application have at least the following technical effects or advantages: by using multi-device Bluetooth proximity sensing and cross-device continuous pattern recognition, it is possible to more accurately determine when to trigger the migration of advertising strategies and the degree of inheritance of migration parameters by the target device, thereby achieving seamless and adaptive cross-device migration of advertising strategies, avoiding users from experiencing a cold start or inappropriate advertising experience again after switching devices, and improving the accuracy of advertising and user satisfaction.

[0040] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A personalized advertising delivery method based on user profiles, characterized in that, include: S101, collect basic data, calculate the total usage frequency of the software, and divide the delivery characteristic time period; wherein, the delivery characteristic time period includes inactive time period, general time period and active time period; the basic data includes device usage time, software usage type and software usage frequency; S102, Set a time window, use sensors to acquire posture data within the time window, calculate the screen rotation frequency, grip posture and device tilt angle change rate based on the posture data, and determine the micro-scene category; the micro-scene category includes temporary operation scenarios and fixed entertainment scenarios; S103, Generate ad delivery data based on the micro-scenario category determination results, and deliver ads based on the generated ad delivery data; wherein, the ad delivery data includes ad format, maximum display duration, initial volume, interaction method, display frequency limit and whether automatic video playback is allowed. S104. Calculate the time ratio of temporary operation scenarios and fixed entertainment scenarios based on the time period of the ad placement, and pre-set the ad placement strategy based on the time ratio.

2. The personalized advertising method based on user profiles according to claim 1, characterized in that, The formula for calculating the screen rotation frequency based on posture data is: ,in, Screen rotation frequency, representing the number of times the screen orientation changes per unit of time. This represents the number of times the screen orientation changes within a time window. Each change from the current state to a different state is counted as one change. Two consecutive changes in the same orientation are not counted. This represents the number of sampling points contained within the sliding window. The sampling interval is denoted as .

3. The personalized advertising method based on user profiles according to claim 2, characterized in that, The method for determining the micro-scenario category is as follows: when the holding posture is that the device is placed flat on the table, the tilt angle change rate is less than the minimum tilt angle change rate threshold, and the screen is continuously lit for more than the time threshold, it is determined to be a fixed entertainment scenario in the form of desktop viewing; when the holding posture is that the device is held with both hands, the screen rotation frequency is less than the minimum screen rotation frequency threshold, and the tilt angle change rate is less than the minimum tilt angle change rate threshold, it is determined to be a fixed entertainment scenario in the form of handheld immersion; when the holding posture is that the device is held with one hand, the screen rotation frequency is greater than or equal to the maximum screen rotation threshold, and the tilt angle change rate is greater than or equal to the maximum tilt angle change rate threshold, it is determined to be a temporary operation scenario.

4. The personalized advertising method based on user profiles according to claim 1, characterized in that, Advertising methods also include: S201, collect usage scenario data and operation data, determine the scenario type according to the usage judgment rules of the scenario data, and identify the operation comfort zone based on the operation data; wherein, the usage scenario data includes proximity sensor distance, ambient light data and screen brightness data; the operation data includes operation time, operation coordinates and operation duration; S202, cross-combine scenario types and operational comfort zones to form combined scenarios, adjust the advertising data generated in step S103 according to the combined scenarios, and perform advertising based on the adjusted advertising data.

5. The personalized advertising method based on user profiles according to claim 4, characterized in that, The screen brightness to ambient light ratio is calculated based on ambient light data and screen brightness data, using the following formula: ,in, This is the ratio of screen brightness to ambient light. This is the screen brightness value. The value is 0.1, representing the ambient light intensity, to prevent division by zero errors when the ambient light is zero.

6. The personalized advertising method based on user profiles according to claim 5, characterized in that, The steps for determining the scenario type using the judgment rules are as follows: set the face number threshold, ratio threshold range, and distance threshold range. The ratio threshold range includes a maximum ratio threshold and a minimum ratio threshold. The distance threshold range also has a minimum distance threshold and a maximum distance threshold. If the number of faces is greater than or equal to the preset face number threshold, it means that the front camera has detected at least two different face contours. The user must be in a state of sharing the screen with others. At this time, it is judged as a shared browsing in a multi-person scene. If the number of faces is less than the preset face number threshold, the proximity sensor distance is less than the minimum distance threshold, and the screen brightness to ambient light ratio is greater than the maximum ratio threshold, it is determined to be private browsing; if the proximity sensor distance is greater than the maximum distance threshold and the screen brightness to ambient light ratio is less than the minimum ratio threshold, it is determined to be shared browsing with the device placed horizontally at a distance; if the proximity sensor distance is greater than or equal to the minimum distance threshold and less than or equal to the maximum distance threshold, or the screen brightness to ambient light ratio is greater than or equal to the minimum ratio threshold and less than or equal to the maximum ratio threshold, it is determined to be a normal scenario.

7. The personalized advertising method based on user profiles according to claim 6, characterized in that, The steps for identifying the operating comfort zone based on operational data include calculating the center coordinates and discrete radius of the hot zone, determining the dominant hand tendency based on the center coordinates, identifying fatigue state, and the formula for calculating the discrete radius of the hot zone based on the center coordinates is as follows: ,in, Let M be the discrete radius of the hot zone, M be the total number of operation points in the current operation point queue, and i be the operation point index. , and are the horizontal and vertical coordinates of the screen for the i-th operation point, respectively.

8. The personalized advertising method based on user profiles according to claim 7, characterized in that, Based on the coordinates of the center of the hot zone and the screen width W, the dominant hand preference is determined. < This indicates that most touches occurred in the left third of the screen, suggesting left-hand operation; if > This indicates that most touches occurred in the right third of the screen, suggesting right-hand operation; if < < , The center position suggests that the operation was performed with both hands or the index finger.

9. A personalized advertising delivery method based on user profiles according to claim 4, characterized in that, The advertising delivery method also includes: S301 identifies devices under the same account, builds a proximity sensing network, and identifies continuous patterns across devices based on the proximity sensing network; S302: Calculate contextual confidence and inheritance weight based on the cross-device continuity pattern. Determine whether to perform migration based on contextual confidence and inheritance weight. If the determination result is yes, implement the advertising strategy migration and deliver the advertisement; otherwise, do not perform migration.

10. A personalized advertising delivery method based on user profiles according to claim 9, characterized in that, The method for constructing a proximity sensing network consists of three steps: local area network heartbeat broadcasting, Bluetooth proximity scanning, and spatial determination of proximity devices. The spatial determination of proximity devices is as follows: when device A and device B can detect each other in two consecutive scans, and the average Bluetooth signal strength of both devices is greater than the strength threshold, then A and B are considered to be in the same physical space.

Images