Real-time scene advertisement publishing system based on intelligent waste recycling box

Abstract

The invention discloses a real-time scene advertisement publishing system based on an intelligent waste recycling box, which relates to the technical field of waste recycling and comprises an environment sensing module, a scene decision module, an advertisement management module, a user interaction module and a data management module. The modules are clear in responsibility and smooth in data interaction, the environment sensing module provides real-time data support, the scene decision-making module outputs accurate strategies, the advertisement management module efficiently completes content adaptation and putting, the user interaction module collects feedback, and the data management module achieves information precipitation and optimization. The closed-loop mechanism not only ensures the intelligentization of the whole advertisement release process, but also provides flexible space for subsequent function upgrading, and not only can realize the advertisement function by relying on the existing hardware resources of the intelligent waste recovery box without extra large-scale investment, but also can realize the advertisement function through the effect analysis report of the data management module. Clear delivery data feedback is provided for an advertiser, and the advertiser is assisted to optimize an advertisement strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of waste recycling, in particular to a real-time scene-based advertising system based on a waste intelligent recycling bin. BACKGROUND

[0002] A waste intelligent recycling bin is a modern waste recycling device integrating multiple advanced technologies. It breaks through the single function of traditional recycling bins as storage containers and realizes intelligent processing and paid recycling of recyclable waste by integrating intelligent recognition, automatic classification, weighing, human-computer interaction, and data management technologies. Such recycling bins usually have multiple independent recycling openings corresponding to different types of waste (such as paper, plastic, metal, glass, clothing, batteries, etc.). Users can complete operations through touch screens, scanning codes, swiping cards, or facial recognition, etc. After selecting the type of waste, the automatic door will open to facilitate the disposal. The bin is equipped with weighing sensors, image recognition modules, infrared sensors, and other components, which can accurately identify, classify, and weigh the waste. Some devices also integrate crushing functions to reduce waste volume and improve storage efficiency. At the same time, the recycling bin has a built-in main control module that automatically calculates the corresponding points or amount based on the type and weight of the waste and feeds back to the user through the touch screen interface. Users can transfer points or amount into personal accounts through scanning codes, etc. for gift redemption or cash withdrawal.

[0003] However, the advertising function of existing waste intelligent recycling bins mostly stays at the static display level and cannot flexibly adjust the advertising content according to dynamic changes in the environment. Traditional systems often use pre-set advertising delivery plans and cannot timely perceive actual scene conditions such as weather changes, time changes, and surrounding commercial activities. In rainy weather, community recycling bins still display daily goods advertisements according to the established plan, completely ignoring the urgent needs of residents for rain gear and waterproof products in such weather. During the morning rush hour in an office building, the advertising content pushed by the recycling bin has nothing to do with the commuting needs of office workers, not only failing to attract their attention, but also easily causing users to feel disgusted, resulting in the waste of advertising resources. This uniform advertising delivery mode makes the advertising content seriously disconnected from the actual needs of users in specific scenarios, failing to fully play the guiding and promotional role of advertising. With the passage of time, it easily leads to user "aesthetic fatigue" of advertising, causing the click rate and attractiveness of advertising to decline, making it difficult to meet the market demand for efficient and accurate advertising delivery.

[0004] Currently, there is no effective solution to the problems in the related art. SUMMARY

[0005] In view of the problems in the related art, the application provides a real-time scene-based advertisement publishing system based on a waste intelligent recycling bin to overcome the above technical problems existing in the prior art.

[0006] The technical scheme of the application is implemented as follows:

[0007] The real-time scene-based advertisement publishing system based on a waste intelligent recycling bin comprises an environment perception module, a scene decision module, an advertisement management module, a user interaction module and a data management module, wherein:

[0008] The environment perception module is configured to collect real-time environment parameters and transmit the real-time environment parameters to the scene decision module; the scene decision module is configured to analyze the environment parameters to determine a scene type and generate an advertisement matching strategy; the advertisement management module is configured to filter, adapt and display advertisement content according to the advertisement matching strategy; the user interaction module is configured to realize interaction between a user and the advertisement and collect feedback; and the data management module is configured to store and analyze system data and ensure data security.

[0009] Further, the environment perception module comprises a meteorological data acquisition submodule, a space-time information acquisition submodule and a surrounding information acquisition submodule, wherein:

[0010] The meteorological data acquisition submodule collects rainfall, temperature, humidity and light intensity through a rainfall sensor, a temperature and humidity sensor and a light sensor; the space-time information acquisition submodule obtains a geographic position through a GPS positioning unit and obtains time and time period information through a clock module; and the surrounding information acquisition submodule interfaces with a map service API through a communication module to obtain commercial facilities and activity information within a preset range.

[0011] Further, the scene decision module comprises a data preprocessing submodule, a scene feature extraction submodule, an AI decision submodule and a strategy generation submodule, wherein:

[0012] The data preprocessing submodule performs cleaning, standardization and feature discretization processing on environment data; the scene feature extraction submodule constructs a multi-dimensional feature vector comprising weather state, time period feature, geographic position type, surrounding activity type and historical data feature; the AI decision submodule calculates feature weights through an analytic hierarchy process and determines a scene type in combination with a cosine similarity; and the strategy generation submodule generates an advertisement matching strategy according to the scene type.

[0013] Further, the AI decision submodule comprises the following steps:

[0014] Pre-calculating feature weights comprises determining a weight vector ω of each feature dimension by using an analytic hierarchy process, and is expressed as:

[0015] ω = [ω1, ω2, ω3, ω4, ω5];

[0016] Wherein, the weather feature weight ω1=0.3, the time feature weight ω2=0.25, the location feature weight ω3=0.2, the surrounding activity weight ω4=0.15, the historical data weight ω5=0.1;

[0017] Calculate the scene matching degree, including calculating the cosine similarity of the current feature vector and the preset scene template, expressed as:

[0018] Sim(V, S k ) = (V · S k ) / (||V||·||S k ||);

[0019] Wherein, S k is the kth scene template vector; V·S k is the vector dot product, ||V||, ||S k || is the L2 norm of the vector respectively;

[0020] Select the highest similarity scene type as the determination result, expressed as scene type, expressed as:

[0021] S CJ = argmax(Sim(V, S k )), k = 1, 2,..., n

[0022] Wherein, n is the total number of preset scene types.

[0023] Further, the strategy generation submodule, including the following steps:

[0024] Pre-compute the advertisement category weight, including calculating the matching weight of each category of advertisement for each scene type S CJ , expressed as:

[0025] W c = α·R c + β·C c + γ·T c ;

[0026] Wherein, W c is the matching weight of the cth advertisement, R c is the historical conversion rate of the cth advertisement in the scene, C c is the content relevance of the cth advertisement and the scene, T c is the timeliness coefficient of the cth advertisement, α, β, γ are weight coefficients;

[0027] The advertisement sorting and screening, including: according to W cRanking the advertisement categories in descending order, and selecting the top 3 categories as the recommended advertisement types.

[0028] Further, the advertisement management module comprises a material library management submodule, an intelligent screening submodule, a content adaptation submodule, and a delivery control submodule.

[0029] The material library management submodule stores and manages pictures, videos, and audio; the intelligent screening submodule screens materials according to an advertisement matching strategy; the content adaptation submodule crops and scales materials according to touch screen parameters; and the delivery control submodule controls the display duration, frequency, and carousel of advertisements.

[0030] Further, the user interaction module comprises an interaction interface submodule, a behavior collection submodule, an incentive submodule, and a feedback processing submodule.

[0031] The interaction interface submodule displays advertisements and interactive buttons on the touch screen; the behavior collection submodule records the number of clicks and the viewing duration through a camera and a touch sensor; the incentive submodule gives additional waste recycling points according to user interaction behavior; and the feedback processing submodule analyzes user feedback to optimize advertisements.

[0032] Further, the data management module comprises a database management submodule, a data synchronization submodule, a data analysis submodule, and a security management submodule.

[0033] The database management submodule manages scene features, advertisement materials, user behavior, and model training databases; the data synchronization submodule synchronizes local and cloud data of recycling bins; the data analysis submodule generates advertisement exposure, click volume, and conversion rate; and the security management submodule protects privacy through data encryption and anonymization processing.

[0034] The present application has the following advantages:

[0035] 1. This invention achieves a dual improvement in functional scalability and operational efficiency through modular architecture design and technological collaboration. Each module of the system has clearly defined responsibilities and smooth data interaction. The environmental perception module provides real-time data support, the scenario decision-making module outputs precise strategies, the advertising management module efficiently completes content adaptation and delivery, the user interaction module collects feedback, and the data management module achieves information accumulation and optimization. This closed-loop mechanism not only ensures the intelligent operation of the entire advertising process but also provides flexible space for subsequent functional upgrades. For example, more scenario types can be expanded by updating the scenario decision-making model, or the advertising management module can be upgraded to support new multimedia advertising formats. For operators, the system can leverage the existing hardware resources of the smart waste recycling bins to implement advertising functions without requiring large-scale additional investment. Furthermore, the performance analysis reports from the data management module provide advertisers with clear data feedback on their placement, helping them optimize their advertising strategies. Simultaneously, advertising revenue can be used to support the operation of the recycling bins, reducing equipment maintenance costs and forming a virtuous cycle of "waste recycling and scenario-based advertising," promoting the coordinated development of environmental protection and commercial value.

[0036] Meanwhile, by comprehensively collecting real-time scene elements such as weather, time of day, and surrounding activities through the environmental perception module, and combining this with AI algorithm analysis from the scene decision-making module, advertising strategies highly matched to the current scene can be dynamically generated, completely changing the traditional static "one-size-fits-all" advertising delivery model. For example, on rainy days, community recycling bins automatically push rain gear-related advertisements; in the evening, office building areas focus on food delivery information; and when there are promotional activities in surrounding business districts, promotional content is linked simultaneously. This "time- and place-appropriate" delivery method can accurately reach users' actual needs in specific scenarios, improving the attractiveness and click-through rate of advertisements, while avoiding the interference of ineffective advertisements, maximizing the utilization of advertising resources. At the same time, the user interaction module's points incentive mechanism links advertisement viewing and clicking behaviors with waste recycling points, further enhancing users' active attention to advertisements, forming a dual effect of "scene adaptation and incentive guidance," significantly improving advertising conversion rates. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of a real-time contextualized advertising system based on a smart waste recycling bin, according to an embodiment of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0040] According to an embodiment of the present invention, a real-time contextualized advertising publishing system based on a smart waste recycling bin is provided.

[0041] like Figure 1 As shown, the real-time contextualized advertising system based on a smart waste recycling bin according to an embodiment of the present invention includes:

[0042] Environmental perception module 1 is used to obtain real-time environmental parameters, providing a data foundation for scene decision-making module 2. It includes:

[0043] The meteorological data acquisition submodule is used to collect meteorological parameters such as rainfall, temperature, humidity, and light intensity through rain sensors, temperature and humidity sensors, and light sensors, respectively.

[0044] The spatiotemporal information acquisition submodule is used to obtain the geographical location of the recycling bin through the GPS positioning unit and to obtain the current time and time information through the embedded clock module.

[0045] The surrounding information collection submodule is used to connect to the map service API through the embedded communication module to obtain information on commercial facilities and activities within a 500-meter radius, such as POI data for supermarket promotions and shopping mall anniversaries.

[0046] Scene decision module 2 is used to analyze and process the data collected by the environment perception module 1, determine the current scene type, and generate an advertising matching strategy. This includes:

[0047] The data preprocessing submodule is used to clean and standardize the collected environmental data and remove outliers.

[0048] The technical scheme, the outlier rejection, includes cleaning the original data collected by the environment perception module, such as rainfall, temperature, time period identifier, etc., and identifying outliers by adopting 3σ criterion; the data standardization includes uniformly mapping different dimension environmental parameters to the [0, 1] interval, and adopting min-max standardization formula, expressed as: x'=(x-min(x)) / (max(x)-min(x)), wherein x is the original data, min(x) and max(x) are the minimum value and the maximum value of the parameter respectively, and x' is the standardized value. In addition, feature discretization is performed, including discretizing continuous features (such as time), specifically as follows: time period division: morning (6:00-9:00), morning (9:00-12:00), afternoon (12:00-17:00), evening (17:00-20:00), night (20:00-6:00); rainfall classification: no rain (0mm), light rain (0-5mm / h), moderate rain (5-15mm / h), heavy rain (>15mm / h).

[0049] The scene feature extraction sub-module is used for extracting key scene features from the processed data, and constructing a multi-dimensional feature vector containing weather state, time period feature, geographical location type and surrounding activity type;

[0050] Specifically, the preprocessed data is constructed as a multi-dimensional feature vector V, expressed as:

[0051] V=[W,T,L,A,H];

[0052] Wherein, W represents a weather feature vector, used to represent rainfall, temperature, humidity and light intensity; T represents a time feature, used to represent the discretized time period code; L represents a location feature, used to represent the area type of the recycling bin: community / office building / commercial area / school; A represents a surrounding activity feature, used to represent the intensity of commercial activities and the activity type code; H represents a historical data feature, used to represent the response rate of the same period advertisement in the area.

[0053] The AI decision sub-module is used for analyzing the scene feature vector, determining the current scene type, including the rain community scene, the evening office building scene and the commercial activity scene; including the following steps:

[0054] The feature weight is calculated in advance, including determining the weight vector ω of each feature dimension by adopting analytic hierarchy process (AHP), expressed as:

[0055] ω=[ω1,ω2,ω3,ω4,ω5];

[0056] Wherein, the weather feature weight ω1=0.3, the time feature weight ω2=0.25, the location feature weight ω3=0.2, the surrounding activity weight ω4=0.15, the historical data weight ω5=0.1;

[0057] Calculate the scene matching degree, including calculating the cosine similarity of the current feature vector and the preset scene template, denoted as:

[0058] Sim(V, S k ) = (V·S k ) / (||V||·||S k ||);

[0059] Wherein, S k is the kth scene template vector, such as rain community scene, evening office building scene, etc. V·S k is the vector dot product, and ||V||, ||S k || are the L2 norms of the vectors respectively.

[0060] Select the scene type with the highest similarity as the determination result, denoted as scene type, denoted as:

[0061] S CJ = argmax(Sim(V, S k )), k = 1, 2,..., n

[0062] Wherein, n is the total number of preset scene types;

[0063] The strategy generation submodule is used to generate an advertisement matching strategy according to the scene type, such as rain scene preferentially matching rain gear type advertisement, and evening office building scene preferentially matching catering type advertisement. Including the following steps:

[0064] Pre-calculate the advertisement category weight, including calculating the matching weight of each category of advertisement for each scene type S CJ , denoted as:

[0065] W c = α·R c + β·C c + γ·T c ;

[0066] Wherein, W c is the matching weight of the cth category of advertisement, R c is the historical conversion rate of the cth category of advertisement in the scene, C c is the content relevance of the cth category of advertisement and the scene, T c is the timeliness coefficient of the cth category of advertisement, α, β, γ are weight coefficients, satisfying α+β+γ=1, respectively taking α=0.5, β=0.3, γ=0.2;

[0067] The advertisement sorting and screening includes: arranging the advertisements in descending order according to W c arranging the advertisement categories in descending order, selecting the top 3 categories as recommended advertisement types, and simultaneously satisfying: ΣP c ≤1, wherein P is the display probability of the cth category of advertisement, and the total probability sum is ensured to be 1.

[0068] In addition, dynamic adjustment is also included, which includes: when the advertisement delivery effect is lower than a threshold, which can be less than 0.05 for 3 consecutive times, triggering weight updating, which is represented as:

[0069] W' c =W c +λ·(R' c -R c )

[0070] wherein λ is the learning rate, taking a value of 0.1, and R' c is the latest conversion rate.

[0071] The technical solution can use a logistic regression model to predict the advertisement conversion rate, which is represented as:

[0072]

[0073] wherein θ is the scene feature coefficient vector, φ is the advertisement category feature coefficient, and θ0 is the bias term.

[0074] The advertisement management module 3 is used for management, screening and adaptation of advertisement materials, and pushes corresponding advertisement content according to the strategy generated by the scene decision module 2.

[0075] which includes:

[0076] The material library management sub-module is used for storing and classifiedly managing various advertisement materials, such as pictures, videos, audios and the like, and supports uploading, auditing, updating and unloading of the materials;

[0077] The intelligent screening sub-module is used for screening the most suitable advertisement materials from the material library according to the scene matching strategy;

[0078] The content adaptation sub-module is used for automatically cropping, scaling and the like of the advertisement materials according to the size and resolution of the recycling bin touch screen, so as to ensure the display effect;

[0079] The delivery control sub-module is used for controlling the display duration and frequency of the advertisement, and supports advertisement carousel and emergency content insertion.

[0080] The user interaction module 4 is used for realizing the interaction between the user and the advertisement content, collecting user feedback information and improving user participation.

[0081] which includes:

[0082] An interactive interface submodule is configured to display advertising content and interactive buttons on the recycling bin touch screen.

[0083] A behavior collection submodule is configured to collect user interaction behaviors on the advertising through a camera and a touch sensor, including the number of clicks, viewing time, and dwell time.

[0084] An incentive submodule is configured to give additional waste recycling points to users according to their interaction behaviors on the advertising, and the points are associated with the recycling accounts of the users.

[0085] A feedback processing submodule is configured to sort and analyze user feedback information to provide a basis for advertising optimization.

[0086] A data management module 5 is configured to store, update, analyze, and securely manage various types of data in the system.

[0087] A database management submodule is configured to manage a scene feature database, an advertising material library, a user behavior database, and a model training database.

[0088] A data synchronization submodule is configured to synchronize data between edge nodes, i.e., recycling bins locally and the cloud, to ensure data consistency.

[0089] A data analysis submodule is configured to statistically analyze advertising data, including exposure, clicks, and conversion rates, to generate an advertising effect report.

[0090] With the above scheme, the environmental perception module 1 collects meteorological data, spatiotemporal information, and surrounding activity information in real time and transmits the data to the scene decision module 2. The scene decision module 2 processes the received data, extracts scene features, determines the current scene type through the AI decision submodule, and generates an advertising matching strategy. The advertising management module 3 selects appropriate advertising materials from the material library according to the scene matching strategy and performs content adaptation. The advertising management module 3 displays advertising content through the recycling bin touch screen, and the user interaction module 4 provides an interactive interface. Users interact with the advertising through the touch screen, the behavior collection submodule records user behaviors, and the incentive submodule gives corresponding point rewards. The data management module 5 collects advertising data and user behavior data, analyzes and processes the data to generate an effect report, and provides a basis for subsequent advertising optimization.

[0091] Specifically, in the implementation process, the following steps are included:

[0092] The sensor hardware of the environment perception module 1 is integrated in the intelligent recycling bin body in advance, wherein the rain sensor and the temperature and humidity sensor are installed on the top of the recycling bin, and the GPS module and the communication module are built-in in the box body; the edge computing part of the scene decision module 2 and the data management module 5 is deployed on the embedded processor of the recycling bin, and the cloud part is deployed on the remote server; the software program of the advertisement management module 3 and the user interaction module 4 runs on the touch screen terminal of the recycling bin, and the advertisement display and the user interaction are realized through the touch screen.

[0093] The scene-based advertisement is put in, and the specific implementation is as follows:

[0094] In the rainy day scene, the meteorological data acquisition submodule of the environment perception module 1 detects that the rainfall is greater than 5 mm / h, and the spatio-temporal information acquisition submodule determines that the recycling bin is located in a residential area; the scene decision module 2 determines that it is a “rainy day community scene”, and generates a strategy of “pushing rain gear, waterproof products and housekeeping cleaning service advertisements”; the advertisement management module 3 selects related advertisements from the material library, and displays them after adaptation on the touch screen; the user can get a coupon and additional points by clicking the advertisement.

[0095] In the evening office building scene, the spatio-temporal information acquisition submodule detects that the current time is 18:00-19:00, and the recycling bin is located in a dense office building area; the scene decision module 2 determines that it is an “evening office building scene”, and generates a strategy of “pushing food delivery, fresh food delivery, and gym discount advertisements”; the advertisement management module 3 pushes the corresponding advertisements according to the strategy, and the user can get points reward by watching the advertisement for more than 30 seconds.

[0096] In the commercial circle activity scene, the surrounding information acquisition submodule obtains that the surrounding supermarket is holding an anniversary celebration; the scene decision module 2 determines that it is a “commercial circle promotion scene”, and generates a strategy of “pushing the supermarket promotion advertisement and coupon”; the advertisement management module 3 pushes the related advertisements, and the user can directly jump to the supermarket activity page by clicking the advertisement.

[0097] In addition, the data collected by the environment perception module 1 is preprocessed by the scene decision module 2, and only the feature vector is uploaded to the cloud, and the original data is automatically deleted after being stored for 24 hours; the user behavior data is anonymized, and the information that can identify the personal identity is removed; the advertisement material and the user data transmission adopt the SSL encryption protocol, so as to ensure the safety of data transmission.

[0098] According to the above technical scheme of the present application, the following effects can be achieved:

[0099] 1. This invention achieves a dual improvement in functional scalability and operational efficiency through modular architecture design and technological collaboration. Each module of the system has clearly defined responsibilities and smooth data interaction. The environmental perception module provides real-time data support, the scenario decision-making module outputs precise strategies, the advertising management module efficiently completes content adaptation and delivery, the user interaction module collects feedback, and the data management module achieves information accumulation and optimization. This closed-loop mechanism not only ensures the intelligent operation of the entire advertising process but also provides flexible space for subsequent functional upgrades. For example, more scenario types can be expanded by updating the scenario decision-making model, or the advertising management module can be upgraded to support new multimedia advertising formats. For operators, the system can leverage the existing hardware resources of the smart waste recycling bins to implement advertising functions without requiring large-scale additional investment. Furthermore, the performance analysis reports from the data management module provide advertisers with clear data feedback on their placement, helping them optimize their advertising strategies. Simultaneously, advertising revenue can be used to support the operation of the recycling bins, reducing equipment maintenance costs and forming a virtuous cycle of "waste recycling and scenario-based advertising," promoting the coordinated development of environmental protection and commercial value.

[0100] Meanwhile, by comprehensively collecting real-time scene elements such as weather, time of day, and surrounding activities through the environmental perception module, and combining this with AI algorithm analysis from the scene decision-making module, advertising strategies highly matched to the current scene can be dynamically generated, completely changing the traditional static "one-size-fits-all" advertising delivery model. For example, on rainy days, community recycling bins automatically push rain gear-related advertisements; in the evening, office building areas focus on food delivery information; and when there are promotional activities in surrounding business districts, promotional content is linked simultaneously. This "time- and place-appropriate" delivery method can accurately reach users' actual needs in specific scenarios, improving the attractiveness and click-through rate of advertisements, while avoiding the interference of ineffective advertisements, maximizing the utilization of advertising resources. At the same time, the user interaction module's points incentive mechanism links advertisement viewing and clicking behaviors with waste recycling points, further enhancing users' active attention to advertisements, forming a dual effect of "scene adaptation and incentive guidance," significantly improving advertising conversion rates.

[0101] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art, upon considering the disclosure in the specification and embodiments, will readily conceive of other embodiments of this disclosure. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0102] It should be understood that the present disclosure is not limited to the precise construction that has been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the present disclosure. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A real-time, contextualized advertising system based on a smart waste recycling bin, characterized in that, include: The system comprises an environmental perception module, a scene decision-making module, an advertising management module, a user interaction module, and a data management module, among which; The environmental perception module is used to collect real-time environmental parameters and transmit them to the scene decision module; the scene decision module is used to analyze the environmental parameters to determine the scene type and generate an advertising matching strategy. The advertising management module is used to filter, adapt, and display advertising content according to the advertising matching strategy; the user interaction module is used to enable user interaction with the advertisement and collect feedback; the data management module is used to store and analyze system data and ensure data security.

2. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 1, characterized in that, The environmental perception module includes a meteorological data acquisition submodule, a spatiotemporal information acquisition submodule, and a surrounding information acquisition submodule, wherein; The meteorological data acquisition submodule collects rainfall, temperature, humidity, and light intensity data through rain sensors, temperature and humidity sensors, and light sensors; the spatiotemporal information acquisition submodule obtains geographical location data through a GPS positioning unit and time and time period information through a clock module; the surrounding information acquisition submodule connects to a map service API through a communication module to obtain information on commercial facilities and activities within a preset surrounding area.

3. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 1, characterized in that, The scenario decision-making module includes a data preprocessing submodule, a scenario feature extraction submodule, an AI decision-making submodule, and a strategy generation submodule, wherein; The data preprocessing submodule cleans, standardizes, and discretizes the environmental data; the scene feature extraction submodule constructs a multi-dimensional feature vector that includes weather conditions, time period features, geographical location type, surrounding activity type, and historical data features. The AI ​​decision-making submodule calculates feature weights using the analytic hierarchy process and determines the scene type by combining cosine similarity; the strategy generation submodule generates an advertising matching strategy based on the scene type.

4. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 3, characterized in that, The AI ​​decision-making submodule includes the following steps: Pre-calculate the feature weights, including using the analytic hierarchy process (AHP) to determine the weight vector ω for each feature dimension, expressed as: ω = [ω1, ω2, ω3, ω4, ω5]; Among them, the weights for weather features are ω1 = 0.3, time features are ω2 = 0.25, location features are ω3 = 0.2, surrounding activities are ω4 = 0.15, and historical data are ω5 = 0.

1. Calculating scene matching degree includes calculating the cosine similarity between the current feature vector and the preset scene template, expressed as: Sim(V,S k )=(V·S k ) / (||V||·||S k ||); Among them, S k V·S is the template vector for the k-th scene class; k For vector dot products, ||V|| and ||S k || represents the L2 norm of the vector; The scene type with the highest similarity is selected as the judgment result, denoted as the scene type, and is represented as follows: S CJ =argmax(Sim(V,S k )),k=1,2,...,n Where n is the total number of preset scene types.

5. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 4, characterized in that, The strategy generation submodule includes the following steps: Pre-calculate ad category weights, including for each scenario type S CJ Calculate the matching weight for each category of advertisement, expressed as: W c =α·R c +β·C c +γ·T c ; Among them, W c R is the matching weight for ad class c. c C represents the historical conversion rate of category C ads in this scenario. c For the content relevance of category C ads to the context, T c α, β, and γ are the timeliness coefficients for Class C advertisements, and α, β, and γ are the weighting coefficients. Sorting and filtering ads, including: by W c The ad categories are sorted in descending order, and the top 3 categories are selected as the recommended ad types.

6. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 1, characterized in that, The advertising management module includes a material library management submodule, an intelligent filtering submodule, a content adaptation submodule, and a delivery control submodule, wherein; The media library management submodule stores and manages images, videos, and audio; the intelligent filtering submodule filters media according to advertising matching strategies; and the content adaptation submodule crops and scales media according to touch screen parameters. The delivery control submodule controls the duration, frequency, and rotation of ad displays.

7. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 1, characterized in that, The user interaction module includes an interactive interface submodule, a behavior collection submodule, an incentive submodule, and a feedback processing submodule, wherein; The interactive interface submodule displays advertisements and interactive buttons on the touch screen; the behavior collection submodule records the number of user clicks and viewing time through a camera and touch sensors. The incentive submodule awards extra points for waste recycling based on user interaction behavior; the feedback processing submodule analyzes user feedback to optimize advertisements.

8. The real-time contextualized advertising system based on intelligent waste recycling bins according to claim 1, characterized in that, The data management module includes a database management submodule, a data synchronization submodule, a data analysis submodule, and a security management submodule, wherein; The database management submodule manages the database of scene features, advertising materials, user behavior, and model training; the data synchronization submodule realizes the synchronization of local and cloud data in the recycling bin; and the data analysis submodule generates advertising impressions, clicks, and conversion rates. The security management submodule uses data encryption and anonymization to protect privacy.

Images