Intelligent advertisement position putting method

Abstract

The application discloses an intelligent advertisement position putting method, comprising the following steps: constructing each intelligent advertisement position as a distributed node, forming a trusted state consensus through periodic self-checking, digital signature declaration and neighbor node cross verification; when a fault consensus is reached, triggering an intelligent contract automatically, electing a backup node through multi-dimensional indexes, and realizing high-quality disaster recovery by combining the optimal playing strategy in the group strategy knowledge base; after each playing, analyzing the playing report and the strategy effectiveness through the intelligent contract, anonymizing and refining the efficient strategy, and updating the efficient strategy to the distributed knowledge base to form a closed-loop learning; the application not only ensures the high reliability and continuity of the service through the distributed consensus and the intelligent contract, but also realizes the self-optimization and intelligent improvement of the network playing quality by virtue of the continuous evolution of the group strategy.

Description

Technical Field

[0001] This invention relates to the field of advertising placement technology, and in particular to an intelligent method for placing advertising spaces. Background Technology

[0002] Current outdoor digital advertising networks typically employ centralized or hierarchical management systems for device monitoring and content distribution. This model heavily relies on the stable operation of the central server and network connectivity, posing a single point of failure risk. If a central node or critical network path fails, it can lead to widespread ad outages, and the fault location and recovery process is time-consuming, making it difficult to guarantee high-availability playback services. Furthermore, traditional heartbeat detection mechanisms are relatively simple, prone to misjudgments due to momentary network fluctuations, and ineffective in addressing security threats such as malicious control of devices to play illegal content, posing challenges to reliability and trustworthiness in complex and open environments.

[0003] Existing advertising delivery systems have limitations in playback quality optimization and adaptive learning. Ad content is typically played with fixed parameters, making dynamic optimization difficult based on real-time device performance, network conditions, and audience dynamics. The system lacks the ability to systematically collect, reliably store, and deeply analyze historical playback performance data, hindering the formation of reusable, high-quality playback strategy knowledge. The operational experience of different ad placement terminals is isolated, preventing the network as a whole from learning from successful playback practices or accumulating effective disaster recovery experience from fault recovery cases. This restricts the continuous and automated improvement of ad playback quality and network operational efficiency. Summary of the Invention

[0004] This application provides an intelligent method for ad placement, which enables self-healing from faults and promotes the intelligent improvement and collective evolution of the overall network playback quality, significantly enhancing the robustness of ad playback and service quality.

[0005] This application provides a method for intelligent placement of advertising slots, including: S1 treats each networked smart ad space as a node, periodically performs multi-dimensional state self-checks and generates a digitally signed health declaration. The node then floods the signed declaration to neighboring nodes, which cross-verify the authenticity of the signature with the consistency of the state. S2, the smart contract parses the fault and issues a disaster recovery task. At the same time, it requests the summary of the best playback strategy associated with the group strategy knowledge base. The best backup node is selected through multi-dimensional indicators. The backup node obtains the advertising source file and the group strategy summary, and starts playback after local adaptation. S3: After playback ends, the node generates and submits a detailed playback experience report containing multi-dimensional indicators. The smart contract calculates the overall benefit value of the playback and analyzes the correlation between strategy parameters and benefits to evaluate the effectiveness of the experience. S4 anonymizes efficient strategy parameters into experience fragments and submits them to a distributed knowledge base. An aggregation algorithm is then used to refine and update the optimal strategy summary for the scenario.

[0006] Preferably, the cross-verification of signature authenticity and state consistency by neighboring nodes specifically includes: verifying the authenticity of the digital signature of the signature declaration; comparing the state information recorded in the signature declaration with the state of the sending node monitored by the neighboring node itself; and verifying whether the playback content information recorded in the signature declaration is consistent with the preset schedule.

[0007] Preferably, the step of electing the optimal backup node through multi-dimensional indicators includes: after a node in the network hears the disaster recovery task, it participates in the election by submitting an election declaration; the smart contract elects a backup node based on the multi-dimensional indicators contained in each election declaration according to a publicly available election algorithm; and the results of the elected backup node are announced.

[0008] Preferably, the acquisition of the advertising source file and the group strategy summary specifically includes: S21, constructing a high-precision digital twin model for the target linear space, decomposing the advertisement provided by the advertiser into multiple narrative segments, each segment accompanied by metadata; S22, the advertising space utilizes its built-in anonymous sensors to work collaboratively, perceive the overall flow pattern of the crowd in the space, identify audience groups with similar movement patterns, and predict the movement trajectory of each group in the next n seconds; S23, when a group enters the target area, a dedicated narrative flow instance is created for that group, and based on its predicted trajectory, different segments in the narrative script are dynamically allocated to the advertising space that the trajectory will pass through; S24, the playback command is triggered based on the spatiotemporal event of the audience group arriving at the predicted position, controlling the advertising space before the group is expected to arrive. The allocated segment begins playing in seconds; S25, when multiple narrative streams compete for the same ad slot resource, dynamic arbitration is performed based on factors such as group size, narrative progress, and advertiser priority, and subsequent segments of the narrative stream that does not obtain the right to play are re-planned to nearby available nodes.

[0009] Preferably, predicting the movement trajectory of each group in the next n seconds specifically includes: based on the overall characteristics of each identified audience group, combined with the spatial structure information in the digital twin model, using a motion prediction model to predict the sequence of advertising spaces that each audience group is most likely to pass through in the future within a preset time period, so as to form a predicted trajectory.

[0010] Preferably, the step of dynamically allocating different segments in the narrative script to the ad slots that the trajectory will pass through includes: determining the sequence of ad slots that the predicted trajectory will pass through; matching the expected playback positions of the narrative segments in the narrative script with the actual positions of the ad slot sequence to filter out position-compatible segments; logically sorting and arranging the filtered segments on the timeline based on the temporal logical metadata and audience state assumption metadata of the narrative segments; and generating a playback plan for the ad slot sequence based on the arrangement result.

[0011] Preferably, the controlled ad placement is placed before the group is expected to arrive. The process of starting playback of the assigned segment at a given time also includes: continuously monitoring the real-time location of the audience group; calculating the estimated arrival time of the audience group to each relevant advertising space based on the real-time location and predicted trajectory; calculating the playback trigger time of each relevant advertising space based on the estimated arrival time and a preset lead time; generating a playback instruction containing the playback trigger time and sending it to the corresponding advertising space; and controlling the advertising space to start playing the assigned narrative segment at the playback trigger time.

[0012] Preferably, the step of dynamically allocating different segments of the narrative script to the advertising positions that the trajectory is about to pass through further includes: S321, using a multimodal sensor array deployed in each advertising position to anonymize and collect raw environmental signals, and performing preliminary processing at the edge to extract multi-dimensional feature data; S232, fusing multi-dimensional feature data with macro-level public data streams at regional edge nodes, and using an environmental sentiment computing model for real-time analysis to generate multi-dimensional environmental sentiment vectors; S233, using the real-time environmental sentiment vectors as the core input, a generative visual-sound effect model to synthesize corresponding visual and sound effect generation instruction sets in real time, and distributing them to relevant advertising positions; S234, each advertising position node, based on the received generation instructions and combined with its own screen attributes, performs real-time rendering locally, and seamlessly integrates the generated emotional content or the commercial advertising playback plan tuned by it into the playback control framework triggered by spatiotemporal events.

[0013] Preferably, the generation of the multi-dimensional environmental sentiment vector includes: standardizing the multi-dimensional feature data and macro-level publicly available data; concatenating the standardized data according to a predetermined order and dimensions to construct a fused feature vector; inputting the fused feature vector into a pre-trained environmental sentiment computing model for calculation; and outputting the multi-dimensional environmental sentiment vector by the environmental sentiment computing model.

[0014] Preferably, the real-time synthesis of the corresponding visual and sound effect generation instruction set specifically includes: constructing and maintaining a structured spatiotemporal data lake in the cloud for the geographical location corresponding to each advertising terminal, the data lake containing multiple logical layers; deploying a spatiotemporal weaving engine at the regional edge nodes, which includes a parsing module for anonymizing and calculating audience intent and a fusion decision module for making real-time decisions on presentation strategies based on the intent, environmental conditions, and spatiotemporal data; upgrading the hardware of the advertising terminal, including integrating a transparent reality overlay screen, a spatial projection and lighting system, a multimodal perception array, and a directional spatial audio system; controlling the terminal to operate in an adaptive multi-stage interactive loop, including a seamless wake-up, a gradual unfolding based on pausing, a deep narrative and personalized fusion, and a collaborative contribution stage; and encapsulating the playback instructions of the final presentation content generated by the engine decision into a standardized media task and connecting it to a playback control framework triggered by spatiotemporal events.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages: Through distributed node-based trusted state consensus and smart contract-driven automated disaster recovery, the high reliability and continuity of the advertising playback service are ensured. Simultaneously, leveraging a closed-loop learning and evolution mechanism based on a group strategy knowledge base using playback effect feedback, each playback and recovery process is continuously optimized. This not only achieves self-healing from faults but also drives the intelligent improvement and collective evolution of the entire network's playback quality, significantly enhancing the robustness and service quality of advertising playback.

[0016] By constructing a spatial digital twin and dynamic narrative flow, a qualitative leap has been achieved in advertising delivery, moving from single-point playback to immersive spatial storytelling. Its effects and advantages lie in: accurately and coherently mapping the advertising story onto physical movement lines based on the real-time flow of the audience, achieving synchronization between content and audience movement, greatly enhancing the attractiveness and narrative impact of the advertisement. Through dynamic arrangement and intelligent arbitration, the competition for screen resources among multiple narrative flows is efficiently resolved, ensuring the best experience. Simultaneously, the closed-loop data feedback throughout the process is incorporated into the collective intelligence evolutionary loop of Example 1, enabling continuous synergistic optimization of advertising delivery strategies and spatial narrative capabilities, resulting in significant improvements in the overall intelligence of the delivery method, resource utilization efficiency, and advertising effectiveness.

[0017] Through real-time creation by generative AI, advertising content seamlessly integrates with the pulse of the environment, constructing a new type of digital building facade and advertising media that is highly dynamic, expressive, and evolving. This significantly enhances the attractiveness and environmental adaptability of advertising. It achieves a leap from one-way display to two-way interaction with the environment and emotions, proactively generating or modulating content to reflect, adjust, or enhance the emotional atmosphere of public spaces, significantly improving the emotional resonance of advertising and the immersive experience for users. This elevates advertising spaces from traditional information playback nodes into intelligent interactive portals that carry and interpret multi-dimensional spatiotemporal memories of a location. Its core effect and advantage lie in its ability to integrate geological, historical, social events, collective memory, and real-time sensing data to present a progressive and personalized spatiotemporal narrative based on the audience's intentions. This transforms the advertising experience from one-way display to deep interaction, significantly enhancing the immersiveness, educational value, and emotional connection of the advertisement. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating an intelligent ad placement method according to an embodiment of the present invention. Detailed Implementation

[0019] 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.

[0020] 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 terminology used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1: Figure 1 This is a flowchart illustrating an intelligent ad placement method according to an embodiment of the present invention.

[0022] like Figure 1 As shown, a method for intelligent ad placement includes the following steps: S1 treats each networked smart ad space as a node, periodically performs multi-dimensional state self-checks and generates a digitally signed health declaration. The node then floods the signed declaration to neighboring nodes, which cross-verify the authenticity of the signature with the consistency of the state.

[0023] Specifically, each smart ad playback node in the network periodically performs a self-check. This check includes: its unique identity (ID), the hash value of the currently playing ad content and the frame rate strategy ID, CPU / memory usage, and the current network bandwidth sample value. The node then digitally signs this health status claim containing the above information using its private key.

[0024] Each node broadcasts its signed statement as a heartbeat to its pre-defined set of physically or network-topologically adjacent neighbor nodes. Upon receiving the heartbeat, each neighbor node performs the following verification: First, it verifies the authenticity of the signature using the sending node's public key; second, it compares the statement in the heartbeat with its observed state of the node (analyzing its screen content via a camera); finally, it checks whether the hash of the content being played matches the current schedule recorded on the consortium blockchain.

[0025] If a verification node detects an anomaly (missing heartbeat, inconsistent content, or abnormal resources), it does not make a direct judgment. Instead, it initiates a state challenge request to the other neighbors of the faulty node, attaching evidence. After independent verification, the other neighbors return a verification vote (normal, abnormal, or uncertain) within a specified time. When the number of abnormal votes exceeds a preset threshold (2 / 3), a local consensus is reached that the node's state is abnormal.

[0026] The aforementioned consensus conclusion, key evidence hashes, and all voting records are packaged together into a single transaction and submitted to the consortium blockchain, forming an immutable fault record block. The generation of this block automatically triggers a pre-built, transparent, collaborative recovery smart contract on the blockchain.

[0027] S2: The smart contract resolves the fault and issues a disaster recovery task. At the same time, it requests the optimal playback strategy summary associated with the group strategy knowledge base. The optimal backup node is selected through multi-dimensional indicators. The backup node obtains the advertising source file and the group strategy summary, and starts playback after local adaptation.

[0028] Specifically, the triggered smart contract parses the fault record to determine the faulty node ID and the ID of the advertisement content that should be played. Subsequently, the contract publishes a disaster recovery task on the chain. This task not only includes the identifier of the content to be played, but also includes a strategy request: querying the group strategy knowledge base for a verified summary of the optimal playback strategy (optimal frame rate curve characteristics) associated with the advertisement content.

[0029] Idle nodes in the network listen for on-chain tasks and participate in the election. Electing nodes submit a declaration to the smart contract, including: their physical distance from the faulty node, their current idle resource rate, and their local experience value (if applicable) for playing the advertisement content. The smart contract automatically selects the best 1-2 backup nodes based on a public algorithm (prioritizing nodes that are close, have sufficient resources, and have relevant experience) and announces the election results on the blockchain.

[0030] The selected backup node retrieves the ad source file from the decentralized storage network and simultaneously pulls a summary of the optimal playback strategy from the group strategy knowledge base. Based on its own hardware model, the node adapts this strategy summary into specific, executable parameters and immediately initiates high-quality playback.

[0031] S3: After playback ends, the node generates and submits a detailed playback experience report containing multi-dimensional indicators. The smart contract calculates the overall benefit value of the playback and analyzes the correlation between strategy parameters and benefits to evaluate the effectiveness of the experience.

[0032] Specifically, after each ad playback (whether normal or collaboratively resumed), the relevant nodes generate a detailed playback experience report, including: actual frame rate curve, bandwidth fluctuations, peak hardware resources, stuttering records, and estimated viewer dwell time. This report is then submitted to the blockchain.

[0033] The smart contract processes the report and calculates the overall quality-efficiency value (Q-score) for this playback (based on a weighted average of smoothness, image quality, and resource efficiency). The contract specifically analyzes the correlation between the specific strategy parameters used in this playback and the final Q-score, determining whether the high efficiency stems from recommendations in the group's strategy knowledge base or from more effective fine-tuning by the node itself.

[0034] Proven effective policy parameters (Q-value above the threshold) and their associated contextual information (ad content features, device type, network environment tags) are anonymized and submitted as policy experience fragments to the group policy knowledge base. This distributed knowledge base periodically runs an aggregation algorithm (cluster analysis). When the number of experience fragments for a certain scenario reaches a certain level, it automatically extracts the range of most robust recommendation policy parameters for that scenario and updates its optimal policy summary.

[0035] S4 anonymizes efficient strategy parameters into experience fragments and submits them to a distributed knowledge base. An aggregation algorithm is then used to refine and update the optimal strategy summary for the scenario.

[0036] Specifically, in S3, strategy experience fragments deemed valid by smart contracts (Q-value above a threshold) are encapsulated. The fragment includes: the verified strategy parameters, the resulting overall quality-benefit value (Q-value), and its associated contextual information (ad content characteristics, device hardware model, network environment tags). To protect privacy, all information identifying specific nodes (node ​​ID, geographical location) is stripped away, ensuring complete anonymity of the submitted data. This anonymized fragment is then submitted to a distributed, community-based strategy knowledge base.

[0037] The group strategy knowledge base is jointly maintained by network nodes. Specific data for experience fragments is stored in a decentralized storage network, while only the content addressing hash and key feature index of the corresponding fragment are generated and recorded on the consortium blockchain (advertisement type: sports drink; device type: 4K large screen; network: WiFi-5). This structure ensures data traceability, tamper-proof protection, and efficient retrieval.

[0038] The knowledge base automatically triggers an aggregation algorithm periodically (or when the accumulated experience fragments for a certain scenario reach a threshold). The algorithm performs cluster analysis and statistical modeling on all experience fragments under a specific scenario label, for example, calculating the strong correlation between various policy parameters and high Q-values. Through this process, more universal and robust patterns are extracted from massive amounts of individual experience, forming or updating the optimal policy summary for that type of scenario, which takes the form of a validated recommended parameter range (frame rate recommendation range: 25-30fps; preload buffer: 2-3 seconds).

[0039] The updated optimal strategy summary index is synchronized on the chain. Thereafter, whenever any node (whether playing regularly or as a backup node in S2) needs to play a relevant advertisement, it can query and apply this latest group knowledge.

[0040] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: Through distributed node-based trusted state consensus and smart contract-driven automated disaster recovery, the high reliability and continuity of the advertising playback service are ensured. Simultaneously, leveraging a closed-loop learning and evolution mechanism based on a group strategy knowledge base using playback effect feedback, each playback and recovery process is continuously optimized. This not only achieves self-healing from faults but also drives the intelligent improvement and collective evolution of the entire network's playback quality, significantly enhancing the robustness and service quality of advertising playback.

[0041] Example 2: In Example 1, distributed state consensus and smart contracts enabled self-healing of ad nodes and optimization of group strategies, ensuring the reliability and quality of single-point playback. However, this method primarily focuses on the health status of individual ad slots and isolated content playback, lacking an overall perception of the physical space where the ad slots are located, real-time response to viewer movement, and the ability to logically arrange ad content in spatial and temporal dimensions. This results in ad placement appearing as isolated point displays, lacking coherence between content, and failing to form a logical narrative flow based on the viewer's actual movement, thus limiting the immersive experience and overall dissemination effect of the ads.

[0042] In some embodiments, in obtaining the advertising source file and the group strategy summary, step S2 further includes: S21 constructs a high-precision digital twin model for the target linear space, decomposing the advertisement provided by the advertiser into multiple narrative fragments, each fragment accompanied by metadata.

[0043] The digital twin model includes the precise 3D coordinates, screen attributes, and relationship between each ad placement (i.e., the smart node in Implementation Example 1) and key spatial features (entrances / exits, corners). Each ad placement, as a smart node, registers its unique location identifier in the model. Metadata includes spatial anchors (the suggested physical location range for playback), temporal logic (the order and triggering relationship between segments), audience state assumptions (designed for states such as walking and standing), and optional interactive hooks (visual / auditory cues to guide the audience's attention to the next screen).

[0044] Specifically, a 3D mapping and modeling process is performed on the target linear space. The model precisely labels the 3D coordinates, screen size, and orientation angle of each ad placement (smart node), and identifies key spatial features (intersections, shop entrances). Each smart ad placement binds its digital identity to the corresponding location identifier in the twin model, achieving a precise mapping of the physical device in the virtual space. The spatial narrative script provided by the advertiser is deconstructed, breaking the complete advertising story down into multiple independent narrative segments (video / text / image units), and structured metadata is added to each narrative segment.

[0045] S22, the advertising space uses its built-in anonymous sensors to work together to sense the overall flow pattern of people in the space, identify audience groups with similar movement patterns, and predict the movement trajectory of each group in the next n seconds.

[0046] Specifically, each advertising space within the area simultaneously activates its built-in anonymous sensors (low-resolution wide-angle cameras, Wi-Fi probes, Bluetooth beacons) to collect raw signals of crowd movement within its coverage area (pixel-level optical flow, MAC address signal strength changes) in a manner that does not identify individuals. Central or edge computing nodes aggregate the data from all sensors and use algorithms to calculate in real time the overall main direction of pedestrian flow, average movement speed, and regional density heatmaps. All data processing is performed at the anonymous aggregation level, without any facial recognition or individual tracking.

[0047] Based on aggregated data, clustering algorithms are used to identify sets of individuals with highly similar movement patterns, dividing them into different audience groups. Each group is identified by an anonymous ID and its overall characteristics (centroid position, average velocity vector, size). For each identified audience group, the system uses a motion prediction model (physics-based linear extrapolation or a simple path-following algorithm) based on its current velocity vector, position, and spatial structure (path, inflection points) in the digital twin model to predict the sequence of ad placements that the group is most likely to pass through within a preset time period (30-60 seconds), forming a predicted trajectory.

[0048] S23, when a group enters the target area, create a dedicated narrative flow instance for the group, and dynamically allocate different segments of the narrative script to the ad slots that the trajectory will pass through based on its predicted trajectory.

[0049] Specifically, the output of step S22 continuously monitors each audience group and its predicted trajectory. When the starting point of a group's predicted trajectory enters the preset narrative coverage area in the digital twin model, a unique narrative flow instance is created for that group. This instance records its associated group ID, the spatial narrative script ID used, and its initial state.

[0050] Based on the group prediction trajectory associated with the narrative stream instance, the sequence of ad slots it is expected to pass through in the near future (60 seconds) is determined. Next, the spatial narrative script is read, traversing all narrative segments and their spatial anchor metadata. The expected playback position (spatial anchor) of each segment is matched with the actual positions of the ad slots traversed by the predicted trajectory, and positionally compatible segments are selected as candidate segments for that ad slot.

[0051] Based on the temporal logic metadata (sequential, parallel, and triggering relationships) defined in the narrative script, the selected candidate segments are logically sorted and arranged on the timeline. For example, the first suitable ad slot in the trajectory sequence is assigned an opening segment, and subsequent ad slots are assigned development, climax, and other segments in sequence. At the same time, considering the audience state assumption metadata, short, impactful segments are assigned to areas that are quickly passed through, while more detailed and immersive segments are assigned to areas where viewers are likely to linger.

[0052] Once orchestration is complete, a detailed, time-driven playback plan is generated for the current narrative stream instance. This plan specifies for each ad slot on the trajectory: the segment ID to be played, the planned trigger playback time calculated based on the group's predicted arrival time, and the required synchronization signal parameters. This plan is distributed to the relevant ad slots, which receive it as a high-level task and initiate the predictive content preloading and playback preparation described in step S3 accordingly.

[0053] S24, the playback command is triggered based on the spatiotemporal event of the audience group arriving at the predicted location, controlling the ad placement before the group's expected arrival. The pre-assigned segment will begin playing in seconds.

[0054] Specifically, the system continuously monitors the real-time locations of audience groups associated with each narrative stream instance. When a group's location is updated, the estimated time to reach each ad slot assigned to a narrative segment is calculated based on its latest predicted trajectory. .

[0055] For each relevant ad placement, calculate the exact time to trigger playback. ( (This is a preset lead time, such as 2 seconds). Then, a record containing the fragment ID and absolute trigger time is generated. The scheduled playback command is sent to the corresponding ad slot via a low-latency channel.

[0056] Upon receiving the instruction, the ad placement immediately preloads the specified narrative fragment content from the local cache or the network, and strictly adheres to... Playback will begin at any time. Simultaneously, if the segment needs to be synchronized with adjacent screens, the node will either establish a synchronization mechanism or wait for a synchronization signal based on the playback plan.

[0057] During playback, the actual location of the group is continuously monitored. If the group's movement deviates significantly from the prediction (stagnation), the node can receive instructions to pause or switch to a waiting screen, ensuring that the playback matches the actual spatiotemporal state of the audience. After playback is complete, the node generates a playback report and proceeds to the subsequent evaluation process.

[0058] S25: When multiple narrative streams compete for the same ad slot resource, dynamic arbitration is performed based on factors such as group size, narrative progress, and advertiser priority, and subsequent segments of the narrative stream that does not obtain the right to play are re-planned to nearby available nodes.

[0059] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: By constructing a spatial digital twin and dynamic narrative flow, a qualitative leap has been achieved in advertising delivery, moving from single-point playback to immersive spatial storytelling. Its effects and advantages lie in: accurately and coherently mapping the advertising story onto physical movement lines based on the real-time flow of the audience, achieving synchronization between content and audience movement, greatly enhancing the attractiveness and narrative impact of the advertisement. Through dynamic arrangement and intelligent arbitration, the competition for screen resources among multiple narrative flows is efficiently resolved, ensuring the best experience. Simultaneously, the closed-loop data feedback throughout the process is incorporated into the collective intelligence evolutionary loop of Example 1, enabling continuous synergistic optimization of advertising delivery strategies and spatial narrative capabilities, resulting in significant improvements in the overall intelligence of the delivery method, resource utilization efficiency, and advertising effectiveness.

[0060] Example 3: In Example 2, by constructing a digital twin space and dynamic narrative flow, intelligent arrangement and playback of advertising content along the audience's movement path were achieved, effectively improving the spatial coherence of the narrative. However, the core driving logic of this method mainly relies on the audience's physical location and movement trajectory, and its content generation and scheduling mechanism is essentially a projection of a predefined script in the spatial dimension. This method ignores the real-time changing emotional atmosphere and collective emotional state of the physical environment where the advertising space is located. Whether it is an exciting sports celebration, a quiet rainy night street, or a busy commuting rush hour, its unique atmospheric pulse is not taken into consideration in content creation, which may cause the played narrative content to be out of touch with the overall emotional tone of the current environment, thereby weakening the potential for deep emotional resonance between the advertisement and the audience, and failing to fully utilize the emotional resources of the public space itself to enhance the effectiveness of communication.

[0061] In some embodiments, step S23, which involves dynamically assigning different segments of the narrative script to the ad slots that the trajectory is about to pass through, further includes: S321 uses a multimodal sensor array deployed in various advertising spaces to anonymize and collect raw environmental signals, and performs preliminary processing at the edge to extract multi-dimensional feature data.

[0062] The raw environmental signals include auditory signals, visual signals, physical external air signals, and wireless signals. Multi-dimensional feature data includes auditory features such as average sound intensity (decibels), spectral distribution (high-frequency, mid-frequency, and low-frequency energy proportions), and specific sound recognition (proportion of human voices); visual features such as overall illumination intensity (lux), light color temperature (Kelvin, warm or cool), overall velocity vector of crowd movement, and regional density variation trends; physical features such as temperature, humidity, and air quality index; and pedestrian flow features such as instantaneous pedestrian flow, average movement direction, and gathering hotspots estimated based on anonymous signals.

[0063] S232 integrates multi-dimensional feature data with macro-level public data streams at regional edge nodes, and uses an environmental sentiment computing model for real-time analysis to generate multi-dimensional environmental sentiment vectors.

[0064] The macro-level public data stream includes real-time weather, traffic congestion index, local large-scale event status, and anonymized summaries of social media sentiment trends. The environmental sentiment computation model standardizes data from different sources and with varying dimensions, bringing them to the same magnitude. All standardized features within a time window are concatenated according to a predefined order and dimensions to construct a unified, high-dimensional fusion feature vector. This vector contains both micro-environmental features and macro-contextual information. The constructed fusion feature vector is then input into a pre-trained environmental sentiment computation model. This model is typically a deep neural network, whose core function is to uncover the complex mapping relationship between multi-dimensional data and human-perceived emotional dimensions. The model performs real-time computation on the input vector, outputting a multi-dimensional environmental sentiment vector. Each dimension of this vector corresponds to the intensity value of a basic emotional or atmospheric attribute, typically within a continuous interval of [0,1] or [-1,1].

[0065] S233 uses real-time environmental emotion vectors as its core input and a generative visual-sound effect model to synthesize corresponding visual and sound effect generation instruction sets in real time and distribute them to relevant advertising slots.

[0066] S234, each ad placement node performs real-time rendering locally based on the received generation instructions and its own screen attributes, seamlessly integrating the generated emotional content or the commercial advertising playback plan tuned by it into the playback control framework triggered by spatiotemporal events.

[0067] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: Through real-time creation by generative AI, advertising content seamlessly integrates with the pulse of the environment, constructing a new type of digital building skin and advertising media that is highly dynamic, expressive, and evolving. This greatly enhances the attractiveness and environmental adaptability of advertising. It realizes a leap from one-way display of advertising to two-way interaction with the environment and emotions, and can proactively generate or tune content to reflect, adjust, or enhance the emotional atmosphere of public spaces, significantly improving the emotional resonance of advertising and the immersive experience for users.

[0068] Example 4: In Example 3, by sensing environmental emotions in real time and generating corresponding dynamic content, a preliminary integration of advertising and spatial emotional atmosphere was achieved. However, the core driving logic of this method mainly relies on real-time, instantaneous environmental emotional vectors, and its content generation and presentation are essentially direct or inverse mappings of the current environmental state. This model lacks the ability to bear and invoke the long-term historical depth and multi-layered spatial memory of a location, resulting in generated content that, while possessing emotional real-timeity, lacks cultural depth, historical narrative, and personalized memory connections. The expression dimension of the content is relatively singular, failing to fully utilize the spatiotemporal data assets accumulated by the location to build a more attractive, educational, and user-engaging long-term experience.

[0069] In some embodiments, step S233, which involves real-time synthesis of the corresponding visual and sound effect generation instruction set, further includes: For each ad placement terminal, a structured spatiotemporal data lake is built and maintained in the cloud corresponding to its geographical location. This data lake comprises five logical layers: a geological epoch layer integrating geological and climate model data; a built environment layer based on digital reconstruction of historical data; a social event layer linking multimodal records of major events; a crowdsourced collective and individual memory layer with anonymized data; and a real-time perception layer from IoT sensors. Each layer's data carries precise timestamps and geographic coordinates, forming a full-time digital archive of the location.

[0070] A spatiotemporal weaving engine is deployed at the edge nodes of the region. The engine's audience intent parsing module processes terminal sensor data in real time, anonymizing and calculating the audience's spatial pose, visual focus, interaction posture, and attention level. The spatiotemporal fusion decision module, based on the parsed audience intent, current environmental conditions, and spatiotemporal data retrieved from the data lake, makes real-time decisions on the spatiotemporal layer combination, fusion method, and rendering detail level, generating a presentation strategy following the principles of minimal interference and in-depth rendering as needed.

[0071] Upgrade the hardware of the advertising space terminals to create an integrated perception-presentation portal. The core components include: a high-transmittance transparent reality overlay screen for virtual-real overlay display; a high-precision spatial projection and lighting system that can project content onto building facades or the ground; a multimodal sensing array for deep interactive perception; and a directional spatial audio system that enables targeted sound delivery.

[0072] The terminal operates as an adaptive four-stage interactive loop: Stage 1 is seamless wake-up. When there is no explicit interaction, the transparent overlay screen presents low-interference visualizations that correspond to real-time perceived data (such as environmental emotion vectors from Example 3), like the trajectory of light spots changing with the speed of the flow of people. Stage 2 is gradual unfolding. When the audience stops, the engine dynamically unfolds the spatiotemporal layer according to their attention. For example, a semi-transparent outline of a historical building is overlaid on the real scene first, accompanied by directional audio introduction; if the audience continues to stay, the history is further juxtaposed with the current scene, or the timeline is browsed through gesture interaction. Stage 3 is in-depth narrative and personalized integration. For highly interested audiences, the engine drives a deep narrative mode, which can play videos reconstructing historical events or display collages of collective memories. Under the premise of authorization and anonymity, the audience's personal memories (old photos) can be integrated to generate unique memory landmarks presented in the overlay scene. Stage 4 is collaborative contribution. At the end of the interaction, the audience can be invited to anonymously contribute their current moment's record (synthesized photo) to the collective memory layer, completing the transformation from experiencer to co-writer.

[0073] The final presentation content (whether it's a historical overlay, a landmark memory, or a dynamic narrative) generated by the spatiotemporal weaving engine has its playback instructions encapsulated as standardized media tasks.

[0074] The technical solutions described in the embodiments of this application have at least the following technical effects or advantages: This elevates advertising spaces from traditional information playback nodes into intelligent interactive portals that carry and interpret multi-dimensional spatiotemporal memories of a location. Its core effect and advantage lie in its ability to integrate geological, historical, social events, collective memory, and real-time sensing data to present a progressive and personalized spatiotemporal narrative based on the audience's intentions. This transforms the advertising experience from one-way display to deep interaction, significantly enhancing the immersiveness, educational value, and emotional connection of the advertisement.

[0075] 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. An intelligent method for placing an advertisement position, characterized in that, include: S1 treats each networked smart ad space as a node, periodically performs multi-dimensional state self-checks and generates a digitally signed health declaration. The node then floods the signed declaration to neighboring nodes, which cross-verify the authenticity of the signature with the consistency of the state. S2, the smart contract parses the fault and issues a disaster recovery task. At the same time, it requests the summary of the best playback strategy associated with the group strategy knowledge base. The best backup node is selected through multi-dimensional indicators. The backup node obtains the advertising source file and the group strategy summary, and starts playback after local adaptation. S3: After playback ends, the node generates and submits a detailed playback experience report containing multi-dimensional indicators. The smart contract calculates the overall benefit value of the playback and analyzes the correlation between strategy parameters and benefits to evaluate the effectiveness of the experience. S4 anonymizes efficient strategy parameters into experience fragments and submits them to a distributed knowledge base. An aggregation algorithm is then used to refine and update the optimal strategy summary for the scenario.

2. The intelligent ad placement method according to claim 1, characterized in that, The cross-verification of signature authenticity and state consistency by neighboring nodes specifically includes: verifying the authenticity of the digital signature of the signature declaration; comparing the state information recorded in the signature declaration with the state of the sending node monitored by the neighboring node itself; and verifying whether the playback content information recorded in the signature declaration is consistent with the preset schedule.

3. The intelligent placement method for advertising slots according to claim 1, characterized in that, The process of selecting the optimal backup node through multi-dimensional indicators includes: after a node in the network hears the disaster recovery task, it participates in the election by submitting an election declaration; the smart contract selects a backup node based on the multi-dimensional indicators contained in each election declaration according to a publicly available election algorithm; and the results of the selected backup node are announced.

4. The intelligent placement method for advertising slots according to claim 1, characterized in that, The acquisition of the advertising source file and the group strategy summary specifically includes: S21, constructing a high-precision digital twin model for the target linear space, decomposing the advertisement provided by the advertiser into multiple narrative segments, each segment accompanied by metadata; S22, the advertising space utilizes its built-in anonymous sensors to collaboratively perceive the overall flow pattern of the crowd in the space, identify audience groups with similar movement patterns, and predict the movement trajectory of each group in the next n seconds; S23, when a group enters the target area, a dedicated narrative flow instance is created for that group, and based on its predicted trajectory, different segments in the narrative script are dynamically allocated to the advertising space that the trajectory will pass through; S24, the playback command is triggered based on the spatiotemporal event of the audience group arriving at the predicted position, controlling the advertising space to play before the group is expected to arrive. The allocated segment begins playing in seconds; S25, when multiple narrative streams compete for the same ad slot resource, dynamic arbitration is performed based on factors such as group size, narrative progress, and advertiser priority, and subsequent segments of the narrative stream that does not obtain the right to play are re-planned to nearby available nodes.

5. The intelligent placement method for advertising slots according to claim 4, characterized in that, The prediction of the movement trajectory of each group in the next n seconds specifically includes: based on the overall characteristics of each identified audience group, combined with the spatial structure information in the digital twin model, using a motion prediction model to predict the sequence of advertising positions that each audience group is most likely to pass through in the future within a preset time period, so as to form a predicted trajectory.

6. The intelligent ad placement method according to claim 4, characterized in that, The step of dynamically allocating different segments in the narrative script to the ad positions that the trajectory will pass through includes: determining the sequence of ad positions that the predicted trajectory will pass through; matching the expected playback positions of the narrative segments in the narrative script with the actual positions of the ad position sequence to filter out position-compatible segments; logically sorting and arranging the filtered segments on the timeline based on the temporal logical metadata and audience state assumption metadata of the narrative segments; and generating a playback plan for the ad position sequence based on the arrangement results.

7. The intelligent placement method for advertising slots according to claim 4, characterized in that, The controlled ad placement is scheduled before the group is expected to arrive. The process of starting playback of the assigned segment at a given time also includes: continuously monitoring the real-time location of the audience group; calculating the estimated arrival time of the audience group to each relevant advertising space based on the real-time location and predicted trajectory; calculating the playback trigger time of each relevant advertising space based on the estimated arrival time and a preset lead time; generating a playback instruction containing the playback trigger time and sending it to the corresponding advertising space; and controlling the advertising space to start playing the assigned narrative segment at the playback trigger time.

8. The intelligent placement method for advertising slots according to claim 4, characterized in that, The method of dynamically allocating different segments of the narrative script to the ad positions that the trajectory is about to pass through also includes: S321, using a multimodal sensor array deployed in each ad position to anonymize and collect raw environmental signals, and perform preliminary processing at the edge to extract multi-dimensional feature data; S232, fusing multi-dimensional feature data with macro-level public data streams at the regional edge nodes, and using an environmental sentiment computing model for real-time analysis to generate multi-dimensional environmental sentiment vectors; S233, using the real-time environmental sentiment vectors as the core input, a generative visual-sound effect model to synthesize corresponding visual and sound effect generation instruction sets in real time, and distributing them to the relevant ad positions; S234, each ad position node, based on the received generation instructions and combined with its own screen attributes, performs real-time rendering locally, and seamlessly integrates the generated emotional content or the commercial advertising playback plan tuned by it into the playback control framework triggered by spatiotemporal events.

9. The intelligent placement method for advertising slots according to claim 8, characterized in that, The process of generating a multi-dimensional environmental sentiment vector includes: standardizing the multi-dimensional feature data and macro-level publicly available data; concatenating the standardized data according to a predetermined order and dimensions to construct a fused feature vector; inputting the fused feature vector into a pre-trained environmental sentiment computing model for calculation; and outputting the multi-dimensional environmental sentiment vector from the environmental sentiment computing model.

10. The intelligent placement method for advertising slots according to claim 9, characterized in that, The real-time synthesis of corresponding visual and sound effect generation instruction sets specifically includes: constructing and maintaining a structured spatiotemporal data lake in the cloud for the geographical location corresponding to each advertising terminal, the data lake containing multiple logical layers; deploying a spatiotemporal weaving engine at the regional edge nodes, which includes a parsing module for anonymizing and calculating audience intent and a fusion decision module for making real-time decisions on presentation strategies based on the intent, environmental conditions, and spatiotemporal data; upgrading the hardware of the advertising terminal, including integrating a transparent reality overlay screen, a spatial projection and lighting system, a multimodal perception array, and a directional spatial audio system; controlling the terminal to operate in an adaptive multi-stage interactive loop, including a seamless wake-up, a gradual unfolding based on pausing, a deep narrative and personalized fusion, and a collaborative contribution stage; and encapsulating the playback instructions of the final presentation content generated by the engine into standardized media tasks and connecting them to a playback control framework triggered by spatiotemporal events.

Images