Label anti-counterfeiting query system

Abstract

The invention discloses a label anti-counterfeiting query system, relates to the technical field of intelligent query, and innovatively introduces a time sequence coding technology guided by space-time constraint on the basis of non-tampering evidence storage of a block chain network by constructing a distributed anti-counterfeiting code verification system. A discrete query log is converted into a continuous behavior mode with space-time semantic features, and the limitation that a collaborative counterfeiting behavior cannot be captured through traditional timestamp comparison is broken through. According to the method, an intelligent anti-counterfeiting mechanism with a dynamic risk perception capability can be established, and batch copy attacks and cross-regional collaborative counterfeiting behaviors are accurately identified by deeply mining the space-time correlation characteristics of the query behaviors, so that the anomaly detection capability in an anti-counterfeiting code verification process is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent query technology, and more specifically, to a label anti-counterfeiting query system. Background Art

[0002] In the field of product anti-counterfeiting and traceability, the anti-counterfeiting query system based on digital labels has become an important technical means to ensure product authenticity. The current mainstream technology generally uses a centralized database to store anti-counterfeiting code information and realizes the basic anti-counterfeiting function through the user-side code scanning verification. However, the existing system exposes significant defects when dealing with increasingly complex forgery attacks: the single verification mechanism of the anti-counterfeiting code cannot identify batch copying attacks. When the same anti-counterfeiting code encounters high-frequency cross-regional queries, the system only returns a simple "verified" status without the ability to analyze abnormal behaviors.

[0003] In addition, traditional time-series log analysis technology is often limited to basic timestamp comparison and fails to establish spatio-temporal semantic associations between query behaviors, resulting in the inability to effectively detect covert collaborative forgery behaviors. The existing solutions mostly adopt a centralized architecture for storing query logs, which has the risk of data tampering and lacks a trustworthy distributed evidence storage mechanism. These technical defects make it difficult for the existing anti-counterfeiting systems to provide dynamic risk warnings and trustworthy verification evidence chains when facing professional forgery industrial chains, seriously restricting the actual protection effectiveness of anti-counterfeiting technologies.

[0004] Therefore, an optimized label anti-counterfeiting query system is needed to solve the above technical problems. Summary of the Invention

[0005] In order to solve the above technical problems, this application is proposed.

[0006] According to one aspect of this application, a label anti-counterfeiting query system is provided, which includes: An anti-counterfeiting code verification module, which is used to request the blockchain database for verification after obtaining the anti-counterfeiting code of the label to be verified uploaded by the user; A query record acquisition module, which is used to acquire the query record of the anti-counterfeiting code of the label to be verified; A query status detection module, which is used to perform query status detection based on the query record of the anti-counterfeiting code of the label to be verified to obtain a query status detection result, and the query status detection result is used to indicate whether the query status is abnormal; An anti-counterfeiting code query module, which is used to query the blockchain database to verify whether the anti-counterfeiting code of the label to be verified exists to obtain a query result in response to the query status detection result indicating that the query status is not abnormal; A query log upload module, which is used to return the query result and upload the query log of the anti-counterfeiting code of the label to be verified to the blockchain network.

[0007] Beneficial effects: Compared with the prior art, a label anti-counterfeiting query system provided by the present application constructs a distributed anti-counterfeiting code verification system. Based on the immutable evidence storage of the blockchain network, it innovatively introduces a temporal coding technology guided by spatio-temporal constraints, converts discrete query logs into continuous behavior patterns with spatio-temporal semantic features, breaks through the limitation that traditional timestamp comparison cannot capture collaborative forging behaviors, can establish an intelligent anti-counterfeiting mechanism with dynamic risk perception ability, accurately identify batch replication attacks and cross-regional collaborative counterfeiting behaviors by deeply mining the spatio-temporal correlation features of query behaviors, and thus improves the anomaly detection ability in the process of anti-counterfeiting code verification. Description of the Drawings

[0008] The embodiments of the present application will be described in more detail by combining the accompanying drawings. The above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 It is a block diagram of a label anti-counterfeiting query system according to an embodiment of the present application; Figure 2 It is a schematic diagram of data flow of a label anti-counterfeiting query system according to an embodiment of the present application; Figure 3 It is a block diagram of a query status detection module in a label anti-counterfeiting query system according to an embodiment of the present application; Figure 4 It is a block diagram of a query behavior temporal coding unit in a label anti-counterfeiting query system according to an embodiment of the present application. Detailed Embodiments

[0010] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0011] As shown in the present application and the claims, unless the context clearly indicates an exception, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "including" and "comprising" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0012] Although this application makes various references to certain modules in the system according to embodiments of this application, any number of different modules can be used and run on a user terminal and / or a server. These modules are merely illustrative, and different aspects of the system and method can use different modules.

[0013] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the operations above or below do not necessarily have to be executed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Also, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0014] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all embodiments of this application. It should be understood that this application is not limited by the exemplary embodiments described herein.

[0015] It should be noted in advance that all acquisition and processing of information or data in this application are carried out on the premise of complying with the corresponding national data protection regulations and policies and obtaining authorization from the authority manager.

[0016] In the technical solution of this application, a label anti-counterfeiting query system is proposed. Figure 1 FIG. [Diagram number] is a block diagram of a label anti-counterfeiting query system according to an embodiment of this application. Figure 2 FIG. [Another diagram number] is a schematic diagram of data flow of a label anti-counterfeiting query system according to an embodiment of this application. As Figure 1 and Figure 2 shown, a label anti-counterfeiting query system 300 according to an embodiment of this application includes: an anti-counterfeiting code verification module 310, configured to request a blockchain database for verification after obtaining a label anti-counterfeiting code to be verified uploaded by a user; a query record acquisition module 320, configured to acquire a query record of the label anti-counterfeiting code to be verified; a query status detection module 330, configured to perform query status detection based on the query record of the label anti-counterfeiting code to be verified to obtain a query status detection result, where the query status detection result is used to indicate whether there is an abnormality in the query status; an anti-counterfeiting code query module 340, configured to query the blockchain database to verify whether the label anti-counterfeiting code to be verified exists to obtain a query result in response to the query status detection result indicating that there is no abnormality in the query status; and a query log upload module 350, configured to return the query result and upload the query log of the label anti-counterfeiting code to be verified to a blockchain network.

[0017] Specifically, the anti-counterfeiting code verification module 310 is used to request the blockchain database for verification after obtaining the anti-counterfeiting code of the label to be verified uploaded by the user. It should be understood that traditional anti-counterfeiting systems generally rely on centralized databases to store anti-counterfeiting information. This method is vulnerable to data tampering attacks, resulting in the replication or forgery of anti-counterfeiting codes. By using blockchain technology, its immutable feature can be utilized to provide a secure and reliable evidence storage environment for each anti-counterfeiting code. The purpose of this is to ensure the authenticity of each anti-counterfeiting code queried and prevent forgers from deceiving the system by tampering with the records in the database. By requesting the blockchain database for verification after obtaining the anti-counterfeiting code of the label to be verified uploaded by the user, a trust mechanism can be established, enabling both consumers and merchants to rely on this blockchain-based verification method, thereby enhancing market transparency and integrity. During this process, due to the introduction of blockchain technology, each anti-counterfeiting code verification request will obtain a highly credible result, greatly reducing the possibility of successful forgery. At the same time, this also means that even in the face of complex and professional forgery methods, such as batch replication attacks or cross-regional collaborative counterfeiting behaviors, the system can still maintain its effectiveness. Specifically, when the user uploads the anti-counterfeiting code of the label to be verified, the system will first call the anti-counterfeiting code verification module. This module is responsible for establishing a connection with the blockchain database and sending the received anti-counterfeiting code as a query condition to the blockchain network. Next, the blockchain network will retrieve this anti-counterfeiting code to find out whether there are corresponding records and the status of these records (such as whether they have been verified).

[0018] Specifically, the query record acquisition module 320 is used to obtain the query records of the anti-counterfeiting code of the label to be verified. In the current field of product anti-counterfeiting and traceability, forgery attacks have become increasingly complex, and traditional single-time verification mechanisms can no longer meet the needs of identifying batch replication attacks or cross-regional collaborative counterfeiting behaviors. To effectively address these challenges, it is necessary to analyze the historical query records of each anti-counterfeiting code to be verified. This is because the verification result at a single point in time may not be sufficient to reveal potential abnormal behaviors, and by reviewing the historical query records, those imperceptible patterns and rules can be discovered, thereby more accurately determining whether there is forgery behavior. In one example, after the user uploads the anti-counterfeiting code of the label to be verified, the query record acquisition module in the system will be activated. First, this module will use the received anti-counterfeiting code as a keyword to search for all related query records in the blockchain database. During this process, the system not only has to retrieve the initial creation record but also find each query request for this anti-counterfeiting code and its result. Then, these records will be sorted into an ordered time series so that the query status detection module in the subsequent steps can conduct in-depth analysis on them.

[0019] Specifically, the query status detection module 330 is configured to perform query status detection based on the query records of the anti-counterfeiting codes of the tags to be verified, so as to obtain a query status detection result, and the query status detection result is used to indicate whether there is an abnormality in the query status. In a specific example of the present application, as Figure 3 shown, the query status detection module 330 includes: a query behavior temporal encoding unit 331, configured to perform semantic-level temporal context encoding with spatio-temporal constraint guidance on each query log in the query record based on a time stamp to obtain a query behavior temporal pattern encoding feature; a detection result generation unit 332, configured to obtain a query status detection result based on the query behavior temporal pattern encoding feature.

[0020] Specifically, the query behavior temporal encoding unit 331 is configured to perform semantic-level temporal context encoding with spatio-temporal constraint guidance on each query log in the query record based on a time stamp to obtain a query behavior temporal pattern encoding feature. In a specific example of the present application, as Figure 4 shown, the query behavior temporal encoding unit 331 includes: an arrangement subunit 3311, configured to arrange each query log in the query record in the order of the time stamp to obtain a time queue of query logs; a query behavior semantic encoding subunit 3312, configured to perform context semantic encoding on each query log in the time queue of query logs to obtain a time queue of query behavior semantic encoding features; a query behavior temporal transfer aggregation analysis subunit 3313, configured to perform query behavior temporal transfer aggregation analysis on the time queue of query behavior semantic encoding features to obtain a query behavior temporal pattern encoding feature.

[0021] More specifically, the arrangement subunit 3311 is configured to arrange each query log in the query record in the order of the time stamp to obtain a time queue of query logs. It should be understood that due to the neglect of the time series characteristics of query behaviors in existing systems, it is impossible to identify collaborative forgery behaviors across devices and regions within a short period of time. By constructing a strictly time-ordered log queue, the system can accurately restore the true historical trajectory of the anti-counterfeiting code being queried, providing basic data support for subsequent spatio-temporal semantic analysis. By arranging each query log in the query record in the order of the time stamp, the system can construct a causal query behavior evolution chain, providing high-quality input for subsequent temporal models such as bidirectional LSTM to capture potential associations between query events.

[0022] More specifically, the query behavior semantic encoding subunit 3312 is used to perform context semantic encoding on each query log in the time queue of the query log to obtain a time queue of query behavior semantic encoding features. That is, in an embodiment of the present application, first, semantic embedding encoding is performed on each query log in the time queue of the query log to obtain a time queue of query behavior semantic embedding encoding vectors. It should be understood that the traditional verification system can only identify explicit time interval anomalies, but cannot parse deep semantic features such as geographic location offset and device fingerprint mutation implied in the query behavior. The current counterfeit industry chain often uses distributed device clusters to simulate real user behavior. Its query logs seem to be compliant in the time dimension, but the semantic relevance of fields such as device model and IP address location often exposes traces of counterfeiting. Through semantic embedding coding, the system can map discrete log fields (such as geographic location coordinates, device hash values, network operator codes) into continuous high-dimensional vector spaces, and convert the originally fragmented field information into computable feature expressions. In this way, the system can construct a spatiotemporal semantic feature substrate that can be parsed by a deep learning model, so that the subsequent bidirectional LSTM model can capture potential correlation patterns across fields and provide high-density semantic feature input for subsequent spatiotemporal constraint analysis. In a specific example of the present application, each query log in the time queue of the query log can be semantically embedded using a query behavior embedding matrix to obtain a time queue of query behavior semantic embedding encoding vectors.

[0023] Then, the time queue of the query behavior semantic embedding encoding vector is contextually encoded based on the bidirectional LSTM model to obtain the time queue of the query behavior semantic encoding vector. It should be understood that modern collaborative counterfeiting behaviors often simulate normal query rhythms through distributed device groups. The semantic features of a single log may meet the specifications, but the contextual correlation across logs will expose anomalies. For example, when counterfeiters manipulate device clusters located in different geographical areas to initiate queries for the same anti-counterfeiting code, although the device fingerprint and IP address of each query log are independent and legal, the device type switching frequency (such as alternating queries between mobile and PC terminals) and the geographic location transition speed (such as crossing three time zones within 10 minutes) form a specific correlation pattern in the time series dimension. By contextually encoding the time queue of the semantic embedding vector with a bidirectional LSTM model, the system can simultaneously capture the potential dependencies in the forward and backward directions, breaking through the limitation that traditional unidirectional recurrent neural networks can only capture historical information. In this way, the system can construct spatiotemporal correlation features with global perception capabilities, so that the model can not only identify the isolated features of the current query log, but also gain insight into its role positioning in the entire query history. Specifically, forward propagation captures the gradual change pattern of device fingerprints, backward propagation analyzes the mutation trend of geographic location, and the encoding vector formed by the fusion of bidirectional information flows can quantitatively represent the rationality of the query behavior chain.

[0024] More specifically, the query behavior temporal transfer aggregation analysis subunit 3313 is configured to perform query behavior temporal transfer aggregation analysis on the time queue of query behavior semantic encoding features to obtain query behavior temporal pattern encoding features. It should be understood that traditional semantic encoding can only capture the features of isolated query events, but cannot analyze the spatio-temporal correlation across logs, resulting in the fact that forgery behaviors can be disguised as normal query sequences. For example, a forger may manipulate a group of devices distributed in different geographical regions to initiate queries at intervals that conform to statistical laws, making the spatio-temporal features of a single log seem legal, but the combined pattern thereof implies deep anomalies such as device fingerprint drift and IP home location jump. Due to the lack of the ability to dynamically model the joint features in the spatio-temporal dimension, existing systems are difficult to identify such collaborative attacks with the characteristics of "legal single point, abnormal association". Therefore, in the technical solution of this application, query behavior temporal transfer aggregation analysis is performed on the time queue of query behavior semantic encoding features to obtain query behavior temporal pattern encoding features. That is, by inputting the semantic encoding vector sequence of query logs into a recurrent neural network for preliminary temporal modeling, and on this basis, constructing spatio-temporal collaborative constraint factors to establish a global spatio-temporal association model of query behavior chains. In this process, by calculating the query behavior temporal confidence constraint factor to quantify the behavior continuity in the time dimension, and combining the query behavior spatial confidence constraint factor to evaluate the physical rationality of geographical location transitions, a comprehensive index reflecting the spatio-temporal collaborative anomalies of query behaviors is finally generated by fusion. Specifically, the temporal confidence factor captures the compliance of query intervals through an attention mechanism (such as the physical inaccessibility across three time zones within 10 minutes), and the spatial confidence factor quantifies the credibility of geographical jumps based on the topological relationship of IP addresses (such as the matching degree between the reachable distance of the high-speed rail from Beijing to Shanghai and the query time difference). In particular, by combining the spatio-temporal curvature compensation optimization mechanism, the system can eliminate the negative curvature interference generated when single-dimensional constraint factors are fused, ensuring that the spatio-temporal collaborative constraint factors accurately reflect the essential features of spatio-temporal contradictions in forgery behaviors. Through the message passing structure modulation technology, the semantic features extracted by bidirectional LSTM are dynamically fused with the spatio-temporal constraint factors, enabling the model to automatically amplify the influence of abnormal nodes (such as sudden intensive queries across time zones), while suppressing the noise interference of normal queries. This aggregation mechanism refracts the originally discrete query events into a continuous behavior spectrum, improving the accuracy of query state detection.

[0025] Specifically, first, the time queue of query behavior semantic encoding vectors is input into a sequence encoder based on a recurrent neural network to obtain a time series of query behavior sequences passing the initial encoding vectors. Among them, the recurrent neural network can establish an information transmission channel across time steps through a recurrent connection structure, breaking through the limitation that traditional feedforward neural networks can only process static features. By inputting the time queue of query behavior semantic encoding vectors into the sequence encoder based on the recurrent neural network, the system can extract potential temporal dependence features in the query behavior chain and construct an intermediate representation with dynamic evolution characteristics for subsequent spatio-temporal collaborative constraint analysis. In a specific implementation, the hidden state transmission mechanism of the RNN can capture temporal-sensitive features such as the gradual change of device types (such as the switching frequency from mobile to PC) and the drift speed of IP addresses (such as the continuous transition rate of geographical coordinates), convert discrete semantic encoding vectors into sequences containing time evolution rules to pass the initial encoding vectors, and obtain the time series of query behavior sequences passing the initial encoding vectors. In a specific example of the present application, the time queue of query behavior semantic encoding vectors is input into the sequence encoder based on the recurrent neural network with the following encoding formula to obtain the time series of query behavior sequences passing the initial encoding vectors; where the encoding formula is: Among them, among them, is the time queue of the query behavior semantic encoding vectors, are respectively the 1st, the 2nd, the th, and the th query behavior semantic encoding vectors in the time queue of the query behavior semantic encoding vectors, recurrent neural network, are respectively the 1st, the 2nd, the th, and the th query behavior sequence passing the initial encoding vectors in the time series of query behavior sequences passing the initial encoding vectors.

[0026] Next, calculate the query behavior time-series confidence constraint factor for each query behavior sequence that passes the initial encoding vector in the time series of query behavior sequences. It should be understood that existing forgery attack behaviors have shown deep camouflage characteristics in the time dimension. For example, forged queries are evenly distributed in the time gaps of legitimate queries, or false time continuity is constructed using historical normal query records, making it difficult for traditional fixed-time-window detection mechanisms to identify such time-series forgeries. Therefore, in the technical solution of this application, to establish a time-dimensional credibility evaluation system for query behavior chains, calculate the query behavior time-series confidence constraint factor for each query behavior sequence that passes the initial encoding vector in the time series of query behavior sequences, so as to dynamically evaluate the credibility fluctuations of query behaviors on the time axis through the attention mechanism. For example, when a large number of high-frequency queries occur for a certain anti-counterfeiting code in the early morning period, the system automatically increases the confidence weight of the encoding vector in that period, making the characteristics of abnormal time clusters significantly enhanced in subsequent analyses. In specific implementation, due to the subtle differences between forged events and real events in terms of time correlation intensity, interval rules, etc., the spatio-temporal confidence constraint factor makes the abnormal time nodes produce a feature distortion effect during aggregated encoding by dynamically adjusting the feature contribution degree, and based on this, constructs a query behavior abnormal feature screening mechanism with time cognitive intelligence. In this way, the originally evenly distributed time noise is transformed into a feature map with significant gradient changes, so that even when the query timestamps are completely compliant, deep camouflaged collaborative forgery attacks can still be identified through the lack of temporal coherence in the behavior pattern. In a specific example of this application, the query behavior time-series confidence constraint factor for each query behavior sequence that passes the initial encoding vector in the time series of query behavior sequences is calculated using the following formula; where the formula is: Among them, function, and respectively represent trainable weighted hyperparameters, and are weight matrices, represents vector multiplication, is the key node weight vector of the query behavior, is the time queue of the query behavior time-series confidence metric factor, function, is the query behavior time-series confidence constraint factor.

[0027] Subsequently, calculate the query behavior space confidence constraint factor of each query behavior sequence in the time series of the query behavior sequence transmitting the initial encoding vector. It should be understood that existing forgery attacks have developed highly collaborative spatial strategies. Attackers simulate the geographical distribution characteristics of real users through a distributed device network, making the geographical location information of a single query log seem compliant, but the spatial association pattern of the device cluster implies anomalies. For example, forgers may manipulate device groups distributed in different cities to initiate queries in a specific spatial topology (such as equidistant circular distribution), using the natural dispersion of geographical locations to conceal the intention of collaborative attacks. Due to the lack of in-depth modeling of spatial structure features, traditional detection methods are difficult to identify such spatial forgery behaviors with the characteristics of "local compliance and global anomaly". Therefore, in order to construct a spatial topology-sensitive feature evaluation system, in the technical solution of this application, calculate the query behavior space confidence constraint factor of each query behavior sequence in the time series of the query behavior sequence transmitting the initial encoding vector, so as to accurately capture the spatial collaboration anomalies of the device cluster in the forgery attack by quantifying the differences in the information propagation influence of different spatial nodes. In a specific implementation, this process analyzes two types of spatial features: one is the physical rationality of geographical location transitions (such as the possibility of crossing different climate zones in a short time), and the other is the spatial topology density of the device network (such as the abnormal aggregation degree of query devices in a specific area). By dynamically adjusting the weight distribution of spatial nodes, the system can suppress the influence of query noise in low-risk areas and at the same time strengthen the expression of high-risk spatial association features, enabling subsequent spatio-temporal collaboration constraint analysis to accurately identify the forgery attack spatial patterns disguised as natural distributions. In a specific example of this application, the query behavior space confidence constraint factor of each query behavior sequence in the time series of the query behavior sequence transmitting the initial encoding vector is calculated by the following formula; where the formula is: Wherein, Exponential operation, Represents the square of the norm, Represents The corresponding spatial confidence score value, Is the query behavior space confidence constraint factor.

[0028] Then, based on the query behavior space confidence constraint factor and the query behavior temporal confidence constraint factor for transmitting the initial encoding vector in each query behavior sequence, a query behavior transmission spatio-temporal collaborative constraint factor for transmitting the initial encoding vector in each query behavior sequence is constructed. That is, in the embodiments of the present application, first, the query behavior space confidence constraint factor and the query behavior temporal confidence constraint factor are fused to obtain an initial query behavior transmission spatio-temporal collaborative constraint factor. Considering that it is difficult to identify spatio-temporal coupled forgery patterns in a single time or space dimension. For example, constructing a query trajectory with a reasonable time distribution but a spatial migration that violates the logistics network topology (such as a commodity directly appearing in the terminal market without passing through the regional distribution center), which makes the single-dimension detection mechanism completely ineffective. Therefore, to break through the limitations of single-dimension analysis and construct a comprehensive index that can characterize spatio-temporal coupled query behavior anomalies, in the technical solution of the present application, the query behavior space confidence constraint factor and the query behavior temporal confidence constraint factor are fused to obtain an initial query behavior transmission spatio-temporal collaborative constraint factor. Specifically, the system uses a gating mechanism to dynamically adjust the weight allocation of spatio-temporal factors. When it detects a high-frequency query accompanied by a discontinuous jump in geographical coordinates, it automatically enhances the decision weight of the spatial constraint factor; while in the scenario of low-frequency queries but abnormal drift of device fingerprints, it emphasizes the contribution degree of the time constraint factor. This fusion mechanism reconstructs the isolated time flow and spatial points into a spatio-temporal manifold with conduction logic, elevating the feature expression dimension of forgery attacks from two-dimensional decoupling to three-dimensional coupling. By capturing the non-linear correlation features of spatio-temporal factors, the system can identify high-level forgery attacks of the type "legal time series + legal spatial distribution" that cannot be detected by traditional methods, providing a credible decision basis with spatio-temporal joint semantics for the blockchain evidence storage module and forming a multi-dimensional and three-dimensional combat ability against professional forgery industrial chains. In a specific example of the present application, the following fusion formula is used to fuse the query behavior space confidence constraint factor and the query behavior temporal confidence constraint factor to obtain an initial query behavior transmission spatio-temporal collaborative constraint factor; wherein, the fusion formula is: Wherein, function, and are fusion weight parameters, is the initial query behavior transmission spatio-temporal collaborative constraint factor.

[0029] Furthermore, a spatio-temporal curvature compensation optimization is performed on the initial query behavior by transmitting a spatio-temporal collaborative constraint factor to obtain a query behavior transmitted spatio-temporal collaborative constraint factor. In particular, during this process, due to the possible spatio-temporal joint distribution in forgery attacks (such as a non-linear positive correlation between time intervals and geographical transition speeds), the spatio-temporal factor fusion in the traditional Euclidean space causes representation distortion. For example, forgers may manipulate a cluster of devices to simulate the normal logistics rhythm in the time dimension (such as initiating a query every 6 hours), and at the same time distribute the device locations according to the reachable distance of high-speed trains in the space dimension, resulting in the initial spatio-temporal constraint factor being misjudged as a compliant behavior during planar fusion. In fact, there is a coupling relationship of super-physical laws between the time growth rate and the spatial displacement rate (such as crossing 2000 kilometers within 10 minutes). Therefore, in a preferred example of this application, a spatio-temporal curvature compensation optimization is performed on the initial query behavior by transmitting a spatio-temporal collaborative constraint factor to obtain a query behavior transmitted spatio-temporal collaborative constraint factor. That is, by constructing a constant curvature space representation model, the initial constraint factor is mapped from the planar Euclidean space to a manifold space adapted to non-linear spatio-temporal correlations. For example, when the query behavior of a certain anti-counterfeiting code shows a reasonable linear distribution in the time dimension, but the spatial migration violates the curvature characteristics of the connection of the logistics network hub, the system needs to automatically correct this internal contradiction of the spatio-temporal manifold. Specifically, the system uses a hyperbolic space to model the exponential decay characteristics of time factors, combines a spherical space to capture the periodic law of geographical location transitions, and dynamically adjusts the geometric distribution form of spatio-temporal factors through a curvature compensation mechanism, so that the "super-linear spatio-temporal correlation" (such as time compression accompanied by space expansion) implicit in forgery attacks is highlighted as an identifiable abnormal feature in the optimized constraint factor. At the technical effect level, this optimization mechanism enables the spatio-temporal collaborative constraint factor to have the ability to resist geometric forgery. By capturing the non-Euclidean geometric relationship between time acceleration and space expansion, the system can identify forgery attack patterns that cannot be detected by traditional planar fusion models, build a physically reasonable decision basis for the blockchain evidence storage module, and form a dimensionality reduction attack ability against spatio-temporal coupling forgery attacks.

[0030] In this example, the following optimization formula is used to perform spatio-temporal curvature compensation optimization on the initial query behavior by transmitting a spatio-temporal collaborative constraint factor to obtain a query behavior transmitted spatio-temporal collaborative constraint factor; where, the optimization formula is: Where, is the constant curvature space representation, is the spherical coordinate approximation representation, is the query behavior transmitted spatio-temporal collaborative optimization factor, represents the square of the query behavior transmitted spatio-temporal collaborative optimization factor, Indicates the spatio-temporal collaborative constraint factor for query behavior transmission.

[0031] Specifically, first, based on each query behavior sequence, an initial encoded vector is transmitted The corresponding temporal confidence constraint factor of the query behavior And the spatial confidence constraint factor of the query behavior To construct a constant curvature space representation And a spherical coordinate approach representation : Then, solve the spatio-temporal collaborative optimization factor for query behavior transmission And use the spatio-temporal collaborative optimization factor for query behavior transmission As the fused space metric benchmark representation to optimize the initial spatio-temporal collaborative constraint factor for query behavior transmission : That is, when It can make the unimodal correlation representations And In the time and space dimensions approach a flat spatio-temporal coupling attraction, thus reflecting the Euclidean-like property of the initial spatio-temporal collaborative constraint factor for query behavior transmission in the fused space. That is, by compensating for the generation of unimodal negative curvature to achieve plane preservation in the fused space, thereby enhancing the Fusion expression effect of the initial spatio-temporal collaborative constraint factor for query behavior transmission.

[0032] Further, based on the query behavior transmission spatiotemporal synergistic constraint factor, the query behavior transmission structure modulation is performed on the initial encoding vector of each query behavior sequence to obtain the time series of the query behavior sequence transmission structural modulation encoding vector. It should be understood that the traditional static feature weighting method is difficult to capture the dynamic evolution law in the multi-dimensional dynamic coupling forgery attack mode. Therefore, in the technical solution of the present application, the contribution of each time step feature is dynamically modulated by the spatiotemporal synergistic constraint factor to construct a feature enhancement mechanism with adaptive capabilities, so that the abnormal spatiotemporal correlation mode is amplified step by step during the feature propagation process. Specifically, the system uses a nonlinear gating mechanism to parse the deep semantics of the spatiotemporal constraint factor. When a time-intensive query is detected with a spatial topological mutation, the modulation coefficient of the corresponding time node is automatically enhanced, so that the implicit abnormal features such as the device fingerprint drift trajectory and the IP attribution transition rate form a resonance enhancement effect in the encoding vector. The modulation mechanism enables the system to have the ability to resist dynamic camouflage. Through the spatiotemporal constraint optimization of the feature propagation path, the abnormal signal that appears briefly in the forgery attack is converted into a continuously traceable feature identifier, and a decision input with dynamic adaptability is constructed for the blockchain verification module, forming a continuous suppression capability against evolutionary forgery attacks.

[0033] Subsequently, the positional sum of the time series of the structural modulation coding vector transmitted by the query behavior sequence is calculated to obtain the query behavior time series pattern feature coding vector. It should be understood that the query behavior sequence of a single time node is difficult to expose anomalies. In order to extract the abnormal correlation pattern across time steps, the weak abnormal signals originally scattered in the time series are integrated into recognizable feature identifiers. In the technical solution of the present application, the positional sum of the time series of the structural modulation coding vector transmitted by the query behavior sequence is calculated to obtain the query behavior time series pattern feature coding vector. In this process, the system accumulates the features dimension by dimension, so that the implicit features such as the continuity anomaly of the device fingerprint drift trajectory and the periodic law of the geographical location transition form a superposition resonance effect in the aggregated vector, while suppressing accidental noise interference. This aggregation mechanism enables the system to capture persistent anomalies from long-term query behaviors, build lightweight but complete information decision inputs for the blockchain verification module, and support the global attack on distributed camouflage attacks. In a specific example of the present application, the query behavior sequence transfer structural modulation coding vector is transferred by adding the positional sum of the time series to obtain the query behavior temporal pattern feature coding vector; wherein the summation formula is: in, the scale of the time series of structural modulation code vectors conveyed for the sequence, A temporal pattern feature encoding vector for the query behavior.

[0034] Specifically, the detection result generation unit 332 is configured to obtain a query status detection result based on the query behavior temporal pattern encoding feature. That is, in the technical solution of the present application, the query behavior temporal pattern feature encoding vector is passed through a query status detector based on a classifier to obtain a query status detection result, and the query status detection result is used to indicate whether there is an abnormality in the query status. That is, a machine learning classifier (such as a deep neural network or a gradient boosting decision tree) is used to perform a non-linear decision boundary division on the high-dimensional feature space of the query behavior temporal pattern to establish an abnormal pattern recognition engine with adaptive learning ability. Among them, the classifier learns the difference patterns between normal queries and forged attacks in dimensions such as spatio-temporal collaborative constraints, device fingerprint evolution, and query density distribution from the training data, and can identify implicit correlation features that are difficult to describe by artificial rules, such as the non-physical positive correlation between the device type switching frequency and the geographical location transition rate. In this way, the system has the ability to resist dynamically evolving attacks. Through the abstract mapping of the feature space and the adaptive optimization of the decision boundary, scattered spatio-temporal abnormal signals are transformed into global risk judgments, providing a decision output with both interpretability and reliability for the blockchain verification module, and forming the ability to continuously learn and defend against new forged attack patterns.

[0035] In particular, the anti-counterfeiting code query module 340 is configured to query the blockchain database to verify the existence of the anti-counterfeiting code of the label to be verified and obtain a query result in response to the detection result of the query status indicating that there is no abnormality in the query status. Since directly accessing the blockchain database in the presence of potential risks is not only inefficient but also may increase the burden on the system and expose more sensitive information. Therefore, in the technical solution of this application, in order to ensure the authenticity of the commodity and prevent counterfeiting, it is first necessary to perform a preliminary security assessment on the anti-counterfeiting code of the label to be verified uploaded by the user. This includes obtaining the historical query records of the anti-counterfeiting code and analyzing whether there are abnormalities in its query status based on these records (such as batch replication attacks or cross-regional coordinated counterfeiting behaviors). Specifically, the system first screens out suspicious anti-counterfeiting codes by performing spatio-temporal semantic analysis on the behavior patterns of the query logs, thereby reducing unnecessary blockchain query times. Secondly, after determining that there is no abnormality, the final blockchain verification is performed to ensure the authenticity and uniqueness of each anti-counterfeiting code. This strategy can not only effectively improve the response speed of the system but also enhance the defense ability against counterfeiting behaviors, protecting consumers from being affected by fake goods. In one example, in response to the detection result of the query status showing that there is no abnormality in the query status, the system considers that the anti-counterfeiting code is trustworthy and can proceed to the next step. Conversely, if abnormal behavior is detected, the user will be immediately notified or corresponding measures will be taken to prevent further operations; for the anti-counterfeiting codes considered to be normal, the system will call the anti-counterfeiting code query module to initiate a query request to the blockchain database. Here, the system will use the anti-counterfeiting code as a keyword to search for relevant information on the blockchain to verify the existence and validity of the anti-counterfeiting code; finally, based on the data on the blockchain, the system will generate a query result and return it to the user. If the anti-counterfeiting code is confirmed to be valid, the system will provide positive feedback to the user; if the anti-counterfeiting code is invalid or has been marked as counterfeit, the user will be warned to pay attention to the risks.

[0036] In particular, the query log uploading module 350 is configured to return the query result and upload the query log of the anti-counterfeiting code of the label to be verified to the blockchain network. It should be understood that the traditional centralized storage method is vulnerable to tampering attacks, while blockchain technology provides an immutable data storage and proof environment. By uploading the result of each query and the relevant query log records to the blockchain, the authenticity and integrity of this information can be ensured, preventing any party from making unauthorized modifications to it. This provides reliable data support for subsequent possible audits and helps to track the occurrence and development trend of forgery behavior. In one example, after the anti-counterfeiting code verification in the blockchain database is completed, the system generates a detailed query result report according to the verification result, which includes the validity status of the anti-counterfeiting code and other relevant information (such as the verification time, etc.); at the same time, the system also needs to prepare the relevant log information of this query, including but not limited to the query timestamp, the user's IP address, device fingerprint, etc. These information will serve as important vouchers for this query activity; then, the system packs the above-generated query result and query log into a transaction and sends it to the blockchain network for confirmation. During this process, all data will be encrypted to ensure its security during transmission; once the transaction is accepted by the blockchain network and reaches a consensus, this data will be permanently recorded on the blockchain and become an immutable part. At this time, the user will also receive the final query result feedback, informing the specific situation of this query; finally, for the convenience of subsequent queries and audits, the system provides corresponding interfaces or tools to allow users or other interested parties to view the query records stored on the blockchain. Through such a process, not only the security and integrity of the data are guaranteed, but also the transparency and reliability of the entire system are improved, providing strong technical support for combating forgery behavior.

[0037] As described above, the label anti-counterfeiting query system 300 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a label anti-counterfeiting query algorithm, etc. In one possible implementation manner, the label anti-counterfeiting query system 300 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the label anti-counterfeiting query system 300 can be a software module in the operating system of the wireless terminal, or can be an application program developed for the wireless terminal; of course, the label anti-counterfeiting query system 300 can also be one of the many hardware modules of the wireless terminal.

[0038] Alternatively, in another example, the label anti-counterfeiting query system 300 and the wireless terminal can also be separate devices, and the label anti-counterfeiting query system 300 can be connected to the wireless terminal through a wired and / or wireless network and transmit interaction information according to a predefined data format.

[0039] The embodiments of the present disclosure have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to technologies in the market, or to enable other ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A label anti-counterfeiting query system, characterized in that Including: An anti-counterfeiting code verification module, which is used to request the blockchain database for verification after obtaining the anti-counterfeiting code of the label to be verified uploaded by the user; A query record acquisition module, which is used to acquire the query record of the anti-counterfeiting code of the label to be verified; A query status detection module, which is used to perform query status detection based on the query record of the anti-counterfeiting code of the label to be verified to obtain a query status detection result, and the query status detection result is used to indicate whether there is an abnormality in the query status, including: a query behavior temporal encoding unit, which is used to perform semantic-level temporal context encoding guided by spatio-temporal constraints on each query log in the query record based on timestamps to obtain query behavior temporal pattern encoding features; a detection result generation unit, which is used to obtain a query status detection result based on the query behavior temporal pattern encoding features; An anti-counterfeiting code query module, which is used to query the blockchain database to verify whether the anti-counterfeiting code of the label to be verified exists to obtain a query result in response to the query status detection result indicating that there is no abnormality in the query status; A query log upload module, which is used to return the query result and upload the query log of the anti-counterfeiting code of the label to be verified to the blockchain network.

2. The label anti-counterfeiting query system according to claim 1, characterized in that The query behavior temporal encoding unit includes: An arrangement subunit, which is used to arrange each query log in the query record in the order of timestamps to obtain a time queue of query logs; A query behavior semantic encoding subunit, which is used to perform context semantic encoding on each query log in the time queue of query logs to obtain a time queue of query behavior semantic encoding features; A query behavior temporal transfer aggregation analysis subunit, which is used to perform query behavior temporal transfer aggregation analysis on the time queue of query behavior semantic encoding features to obtain query behavior temporal pattern encoding features.

3. The label anti-counterfeiting query system according to claim 2, wherein The query behavior semantic encoding subunit is used to: Perform semantic embedding encoding on each query log in the time queue of query logs to obtain a time queue of query behavior semantic embedding encoding vectors; Perform context semantic encoding based on a bidirectional LSTM model on the time queue of query behavior semantic embedding encoding vectors to obtain a time queue of query behavior semantic encoding vectors.

4. The label anti-counterfeiting query system according to claim 3, wherein The query behavior temporal transfer aggregation analysis subunit includes: A spatio-temporal constraint factor calculation secondary subunit, which is used to calculate the query behavior spatio-temporal collaborative confidence constraint factor of the time queue of query behavior semantic encoding vectors to construct a query behavior transfer spatio-temporal collaborative constraint factor; A query behavior sequence transfer aggregation encoding secondary subunit, which is used to perform query behavior sequence transfer aggregation encoding on the time queue of query behavior semantic encoding vectors based on the query behavior transfer spatio-temporal collaborative constraint factor to obtain a query behavior temporal pattern feature encoding vector as the actual process execution path semantic encoding feature as the query behavior temporal pattern encoding feature.

5. The label anti-counterfeiting query system according to claim 4, characterized in that, The spatio-temporal constraint factor calculation secondary subunit includes: A query behavior sequence encoding tertiary subunit, which is used to input the time queue of query behavior semantic encoding vectors into a sequence encoder based on a recurrent neural network to obtain a time series of query behavior sequence transfer initial encoding vectors; The three - level sub - unit for calculating the temporal confidence constraint factor is used to calculate the query behavior temporal confidence constraint factor of each query behavior sequence transmitting the initial encoding vector in the time series of query behavior sequences transmitting the initial encoding vector; The three - level sub - unit for calculating the spatial confidence constraint factor is used to calculate the query behavior spatial confidence constraint factor of each query behavior sequence transmitting the initial encoding vector in the time series of query behavior sequences transmitting the initial encoding vector; The three - level sub - unit for constructing the spatio - temporal collaborative constraint factor is used to construct the query behavior transmission spatio - temporal collaborative constraint factor of each query behavior sequence transmitting the initial encoding vector based on the query behavior spatial confidence constraint factor and the query behavior temporal confidence constraint factor of each query behavior sequence transmitting the initial encoding vector.

6. The label anti-counterfeiting query system according to claim 5, wherein, The three - level sub - unit for constructing the spatio - temporal collaborative constraint factor is used for: Fusing the query behavior spatial confidence constraint factor and the query behavior temporal confidence constraint factor to obtain the initial query behavior transmission spatio - temporal collaborative constraint factor; Performing spatio - temporal curvature compensation optimization on the initial query behavior transmission spatio - temporal collaborative constraint factor to obtain the query behavior transmission spatio - temporal collaborative constraint factor.

7. The label anti-counterfeiting query system according to claim 6, characterized in that, The two - level sub - unit for aggregating the encoded query behavior sequence transmission is used for: Based on the query behavior transmission spatio - temporal collaborative constraint factor, performing query behavior transmission structure modulation on each query behavior sequence transmitting the initial encoding vector to obtain a time series of query behavior sequence transmission structural modulation encoded vectors; Calculating the position - wise sum of the time series of query behavior sequence transmission structural modulation encoded vectors to obtain the query behavior temporal pattern feature encoded vector.

8. The label anti-counterfeiting query system according to claim 6, wherein The detection result generation unit is used for: Passing the query behavior temporal pattern feature encoded vector through the query state detector based on the classifier to obtain the query state detection result, and the query state detection result is used to indicate whether there is an abnormality in the query state.

Images