A digital economy label updating system and method based on big data analysis

Abstract

The application provides a digital economy label updating system and method based on big data analysis, relates to the technical field of label management, constructs a label graph reflecting the behavior semantic features of a target object based on the cross mapping relationship between explicit behavior features and implicit behavior features in the behavior semantic features; performs context scene associated semantic modeling on label nodes in the label graph, extracts the semantic expression weight of each label; detects the matching deviation degree between the current label and the latest behavior semantic features of the target object, and determines that there is label semantic deviation when the matching deviation degree exceeds the preset deviation threshold; when label semantic deviation is detected, the label updating strategy is determined according to the semantic expression weight of each label under different context scenes and the semantic deviation features of the current label, and then the label of the target object behavior features is updated by the label updating strategy. The application can realize adaptive updating management of user behavior labels in a digital economic environment.

Description

Technical Field

[0001] The present application relates to the field of label management technology, and more specifically, to a digital economy label update system and method based on big data analysis. Background Art

[0002] Against the backdrop of the rapid development of the digital economy, user behavior data is becoming more diverse and frequently updated. Traditional tag management methods are no longer able to meet the needs of refined operations and personalized services. As a bridge connecting users and digital platforms, behavior tags play a key role in user portrait construction, recommendation system optimization, and risk control. Therefore, dynamic management of behavior tags has become an important direction for improving the intelligence level of the platform.

[0003] However, existing label update technologies generally rely on fixed rules or static models to periodically adjust user labels, lacking a deep understanding of the semantic evolution of user behavior. Specifically, when faced with the ambiguity and cross-domain drift of user behavior in different contextual scenarios, traditional methods have difficulty identifying the semantic deviation between labels and actual behavior, which can lead to label lag, distortion, or misleading downstream decision-making systems. In addition, because they ignore the changes in the expression weight of labels in different semantic environments, existing methods have large deviations when dealing with label semantic consistency and cannot effectively support accurate label updates. Therefore, how to achieve adaptive update management of user behavior labels in the digital economy environment has become a difficult problem facing the industry. Summary of the Invention

[0004] This application provides a digital economy label update system and method based on big data analysis, which can realize adaptive update management of user behavior labels in the digital economy environment.

[0005] In a first aspect, the present application provides a method for updating digital economy labels based on big data analysis, comprising the following steps:

[0006] Collect structured and unstructured behavioral data of target users on e-commerce trading platforms, and construct user behavior data including operation records, transaction behaviors, and content generation dimensions;

[0007] Extracting behavioral semantic features of the target object in different contextual scenarios from the user behavior data, and then constructing a label map reflecting the behavioral semantic features of the target object based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features;

[0008] Performing context-related semantic modeling on the tag nodes in the tag graph, and then extracting the semantic expression weights of each tag in different contexts;

[0009] Detecting the matching deviation between the current tag and the latest behavior semantic features of the target object. When the matching deviation exceeds a preset deviation threshold, it is determined that there is a tag semantic deviation;

[0010] When a label semantic deviation is detected, a label update strategy is determined based on the semantic expression weight of each label in different contextual scenarios and the semantic deviation characteristics of the current label, and then the label of the target object behavior feature is updated by the label update strategy.

[0011] In this embodiment, extracting the behavior semantic features of the target object in different context scenarios from the user behavior data specifically includes:

[0012] Dividing the user behavior data into multiple user behavior data segments according to a preset time window, and labeling each data segment with a context scenario type;

[0013] For each data segment, extract the semantic entities and behavioral intentions of the unstructured behavioral data in the data segment, and extract the operation frequency and behavioral association path of the structured behavioral data in the data segment;

[0014] By fusing the semantic entity, the behavioral intention, the operation frequency and the behavioral association path, behavioral semantic features reflecting the target object in the corresponding context scene type are generated, and then the behavioral semantic features of the target object in different context scene types are obtained.

[0015] In this embodiment, constructing a label map reflecting the target object's behavior semantic features based on the cross-mapping relationship between the explicit behavior features and the implicit behavior features in the behavior semantic features specifically includes:

[0016] Mapping the explicit behavior feature to a root node of a label graph;

[0017] Mining the cross-dependency relationship between the implicit behavior features and the root node through association rules;

[0018] The child nodes connected to the root node are determined according to the strength of the cross-dependency relationship, and directed edges between the nodes are constructed with the behavioral semantic association as the weight, so as to obtain a label map reflecting the behavioral semantic characteristics of the target object.

[0019] In this embodiment, the context-related semantic modeling of the tag nodes in the tag graph is performed, and then the semantic expression weight of each tag in different context scenarios is extracted, which specifically includes:

[0020] Cluster historical context scenarios to generate multiple scenario classification sets;

[0021] Counting the occurrence frequency and associated node density of each label node in the label map under different scene classification sets;

[0022] Determine the correlation coefficient of each tag node in different context scenarios based on the occurrence frequency and the associated node density;

[0023] The semantic expression weight of each tag in different context scenarios is determined through all the correlation coefficients.

[0024] In this embodiment, the associated node density is an indicator that measures the degree of connection between a tag node and surrounding nodes in a specific scenario.

[0025] In this embodiment, detecting the matching deviation between the current tag and the latest behavior semantic feature of the target object specifically includes:

[0026] Map the latest behavioral semantic features of the target object to the label map to generate a sequence of labels to be matched;

[0027] Extracting the matching degree between each tag in the tag sequence to be matched and the current tag;

[0028] The matching deviation between the current label and the latest behavior semantic features of the target object is determined through all matching degrees.

[0029] In this embodiment, determining the label update strategy based on the semantic expression weight of each label in different context scenarios and the semantic deviation characteristics of the current label specifically includes:

[0030] Obtain matching deviation from the semantic deviation feature of the current label;

[0031] Based on the current context, retrieve the semantic expression weight of each tag;

[0032] Calculating a semantic consistency index of the tag based on the matching deviation and all semantic expression weights;

[0033] By presetting threshold rules, the label update operation type is determined based on the semantic consistency index, and the operation type includes adding a label, adjusting the node weight or deleting a semantically invalid node, and then a label update strategy including the operation type, target label and update parameters is generated.

[0034] In this embodiment, updating the target object behavior feature label by the label update strategy specifically includes:

[0035] Locating the label nodes and associated edges to be operated in the label graph according to the label update strategy;

[0036] The label addition operation, node weight dynamic adjustment, and semantic invalid node deletion are performed according to the priority order in the update rules, thereby generating a versioned label snapshot and updating the behavior label of the target object.

[0037] In this embodiment, the unstructured behavior data refers to user behavior content that cannot be directly expressed in a fixed format.

[0038] In a second aspect, the present application provides a digital economy label update system based on big data analysis, which is used to execute a digital economy label update method based on big data analysis. The digital economy label update system includes:

[0039] The collection module is used to collect structured and unstructured behavioral data of the target object in the e-commerce trading platform, and construct user behavior data including operation records, transaction behavior, and content generation dimensions;

[0040] A feature processing module is used to extract the behavioral semantic features of the target object in different contextual scenarios from the user behavior data, and then construct a label map reflecting the behavioral semantic features of the target object based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features;

[0041] The feature processing module is further used to perform context-related semantic modeling on the tag nodes in the tag graph, and then extract the semantic expression weight of each tag in different contextual scenarios;

[0042] The feature processing module is further configured to detect a matching deviation between the current label and the latest behavior semantic feature of the target object, and to determine that a label semantic deviation exists when the matching deviation exceeds a preset deviation threshold;

[0043] The update module is used to determine the label update strategy based on the semantic expression weight of each label in different context scenarios and the semantic deviation characteristics of the current label when a label semantic deviation is detected, and then update the label of the target object behavior feature according to the label update strategy.

[0044] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0045] First, the structured and unstructured behavioral data of the target object in the e-commerce transaction platform are collected to construct user behavior data including operation records, transaction behaviors, and content generation dimensions; the behavioral semantic features of the target object in different contextual scenarios are extracted from the user behavior data, and then a label graph reflecting the behavioral semantic features of the target object is constructed based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features; the label nodes in the label graph are modeled with the associated semantics of the contextual scenarios, and then the semantic expression weights of each label in different contextual scenarios are extracted; the matching deviation between the current label and the latest behavioral semantic features of the target object is detected, and when the matching deviation exceeds the preset deviation threshold, it is determined that there is a label semantic deviation; when the label semantic deviation is detected, the label update strategy is determined according to the semantic expression weights of each label in different contextual scenarios and the semantic deviation characteristics of the current label, and then the label of the target object's behavioral features is updated by the label update strategy.

[0046] It can be seen that the present application determines the label update strategy based on the semantic expression weight of each label in different contextual scenarios and the semantic deviation characteristics of the current label, and then updates the label of the target object's behavioral characteristics by the label update strategy; first, the present application scheme is based on structured and unstructured behavioral data to construct a user behavior data system covering multi-dimensional information such as operation records, transaction behaviors and content generation, providing high-quality data support for subsequent semantic feature extraction; secondly, by extracting the behavioral semantic characteristics of users in different contextual scenarios and further modeling the cross-mapping relationship between explicit behavioral characteristics and implicit behavioral characteristics, the label map has stronger semantic expression capabilities, breaking through the limitations of the existing method of static mapping between labels and behaviors; then, the scheme introduces a contextual scenario association semantic modeling mechanism in the label map, and by modeling and quantifying the semantic expression weight of each label node in multiple scenarios, it improves the adaptability and discrimination of label semantics in complex scenarios; finally, by comprehensively analyzing the semantic expression weight and deviation characteristics of each label in different contextual scenarios, the label update strategy is intelligently determined, thereby achieving accurate label updates.

[0047] To summarize, this application builds a closed-loop feedback mechanism from behavioral semantic perception, semantic offset identification to label adaptive update, so that user behavior labels can evolve dynamically with behavior changes, ensuring the timeliness and accuracy of user portraits in digital economic applications, and significantly improving the accuracy and robustness of the platform's intelligent decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the accompanying drawings in the following description only relate to the part of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0049] Figure 1 is a flowchart of a digital economic label updating method based on big data analysis provided by the present application;

[0050] Figure 2 is an exemplary flowchart for determining a label graph provided by the present application;

[0051] Figure 3 is an exemplary flowchart for determining a semantic expression weight provided by the present application;

[0052] Figure 4 is a module structure diagram of a digital economic label updating system based on big data analysis provided by the present application. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments only relate to part of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort fall within the scope of the present application.

[0054] The present application provides a digital economic label updating system and method based on big data analysis, which is based on the cross-mapping relationship between explicit behavior features and implicit behavior features in behavior semantic features to construct a label graph reflecting the behavior semantic features of a target object. The label nodes in the label graph are associated with the semantic modeling of the context scene, and the semantic expression weight of each label is extracted. The matching deviation between the current label and the latest behavior semantic features of the target object is detected, and when the matching deviation exceeds the preset deviation threshold, it is determined that there is a label semantic deviation. When the label semantic deviation is detected, the label updating strategy is determined according to the semantic expression weight of each label in different context scenes and the semantic deviation features of the current label, and then the label of the target object behavior features is updated by the label updating strategy. The present application can realize the adaptive update management of user behavior labels in the digital economic environment.

[0055] Embodiment one, in order to better understand the above technical solutions, the above technical solutions will be described in detail in the following with reference to the description of the drawings and specific embodiments, referring to Figure 1As shown in FIG, this figure is an exemplary flow chart of a digital economy label update method based on big data analysis according to this embodiment of the present application, and the digital economy label update method based on big data analysis includes the following steps:

[0056] In step S1, the structured and unstructured behavior data of the target object in the e-commerce transaction platform are collected to construct user behavior data including operation records, transaction behaviors, and content generation dimensions.

[0057] It should be noted that the structured behavioral data in this application refers to user behavior records that can be directly stored and parsed by predefined fields, such as number of clicks, purchase amount, visit duration, and other data with a clear format; the unstructured behavioral data in this application refers to user behavior content that cannot be directly expressed in a fixed format, such as user comments, search keywords, image content, and other data that require semantic parsing.

[0058] In step S2, the behavioral semantic features of the target object in different contextual scenarios are extracted from the user behavior data, and then a label map reflecting the behavioral semantic features of the target object is constructed based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features.

[0059] In this embodiment, extracting the behavioral semantic features of the target object in different contextual scenarios from the user behavior data can be achieved by using the following steps:

[0060] Dividing the user behavior data into multiple user behavior data segments according to a preset time window, and labeling each data segment with a context scenario type;

[0061] For each data segment, extract the semantic entities and behavioral intentions of the unstructured behavioral data in the data segment, and extract the operation frequency and behavioral association path of the structured behavioral data in the data segment;

[0062] By fusing the semantic entity, the behavioral intention, the operation frequency and the behavioral association path, behavioral semantic features reflecting the target object in the corresponding context scene type are generated, and then the behavioral semantic features of the target object in different context scene types are obtained.

[0063] It should be noted that the context scenario type in this application refers to the behavioral situation category composed of the user's behavioral environment and operation intention in a specific time period; the behavioral association path in this application refers to the sequential relationship and transfer logic path between the user's behavioral nodes in a continuous operation; the behavioral semantic feature in this application is a feature representation that reflects the user's behavioral elements in a specific scenario.

[0064] In specific implementation, first, the user behavior log can be time-sliced ​​based on a fixed time window to obtain multiple time-continuous behavior sequence fragments, and the context scenario type of each data fragment can be annotated using a rule template in combination with the page type, activity status or system prompt information when the behavior occurs. The context scenario types in this application specifically include: promotion browsing, order payment, and after-sales interaction; secondly, for the unstructured behavior data, named entity recognition (such as using HanLP) can be used to extract key semantic entities (including product names, brand words, and emotional words) and combined with dependency syntax analysis to infer the user's behavior intentions (including consultation, price comparison, and exclusion). For the structured behavior data, the frequency of various operations is counted, and the behavior transfer path is calculated. The behavior transfer relationship can be modeled with the help of Markov chain construction; then, the feature splicing strategy is used to combine the above-extracted semantic entities, behavior intentions, operation frequencies, and behaviors The paths are vectorized and uniformly embedded in the same contextual semantic space, and a graph neural network is used for fusion modeling to output a behavioral semantic feature vector reflecting the current contextual scenario. Finally, the results of each fragment are summarized to obtain a set of semantic features of the target object in various contextual scenarios. It should be further explained that the graph neural network in this application uses semantic entities, behavioral intentions, operation frequencies and behavioral paths in user behavior data as nodes in fusion modeling to construct a graph structure containing semantic relationships and behavioral transfer logic. The edge relationships between nodes are represented by an adjacency matrix, and after initializing the feature vectors of each node, a graph convolutional network model is used to perform multiple rounds of information transmission and feature aggregation in the graph. In each round of propagation, the node receives the feature information of its neighboring nodes and performs weighted updates based on the edge weights, thereby encoding the deep structural relationship between contextual dependencies and behavioral semantics, and finally outputting a scene semantic embedding representation that integrates multi-source behavioral features.

[0065] In this embodiment, reference Figure 2 As shown in FIG, this figure is an exemplary flow chart for determining a label map in an embodiment of the present application. In this embodiment, the label map reflecting the target object's behavioral semantic features is constructed based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features. The following steps can be used to implement it:

[0066] In step S21, the explicit behavior feature is mapped to the root node of the tag map;

[0067] In step S22, the cross-dependency relationship between the implicit behavior feature and the root node is mined through association rules;

[0068] In step S23, the child nodes connected to the root node are determined according to the strength of the cross-dependency relationship, and directed edges between the nodes are constructed with the behavioral semantic association as the weight to obtain a label map reflecting the behavioral semantic characteristics of the target object.

[0069] It should be noted that the explicit behavioral characteristics in this application refer to observable operational characteristics that users directly demonstrate on the platform and have clear behavioral intentions; the implicit behavioral characteristics in this application refer to behavioral tendencies or preference characteristics that users indirectly reflect in continuous operations; the cross-dependency relationship in this application is a measure of the mutual influence and correlation strength between explicit behavioral characteristics and implicit behavioral characteristics; the label graph in this application refers to a graph structure composed of label nodes and weighted directed edges, which is a knowledge representation graph structure used to express semantic associations between labels.

[0070] In specific implementation, first, the behavior event encoder can be used to convert the explicit behavior features into label nodes and initialize them as the root nodes of the label graph; then, the Apriori algorithm is used to construct a feature set of explicit-implicit behavior pairs in the historical behavior logs of multiple users, and the corresponding support is calculated. The dependency relationship between the explicit behavior features and the implicit behavior features can be used as the support to quantify the cross-dependency, that is, the strong association rules with explicit behavior features as the antecedent and implicit behavior features as the consequent are retained as the basis for composition after meeting the support threshold and confidence requirements; finally, based on the confirmed explicit-implicit behavior pairs, the implicit behavior is connected to the corresponding root node as a child node in the graph structure, and the normalized support is used as the edge weight to construct a weighted directed edge. In the final label graph structure, the root node represents the high-confidence operation intention, the child node captures the complex potential behavior semantics, and the semantic connection structure reflects the overall semantic feature network of the target object behavior.

[0071] It should be noted that this application overcomes the problems of isolated analysis and semantic fragmentation of user behavior in the existing technology by extracting explicit and implicit behavioral features in different contextual scenarios and constructing a label map based on their cross-mapping relationship, realizes deep correlation and semantic fusion of behavioral features, and improves the expression ability and dynamic adaptability of the label map to the semantics of user behavior.

[0072] In step S3, context-related semantic modeling is performed on the tag nodes in the tag graph, and the semantic expression weights of each tag in different context scenarios are extracted.

[0073] In this embodiment, reference Figure 3 As shown in FIG, this figure is an exemplary flow chart for determining the semantic expression weight in an embodiment of the present application. In this embodiment, the context-related semantic modeling of the tag nodes in the tag graph is performed, and then the semantic expression weight of each tag in different context scenarios is extracted. The following steps can be used to achieve this:

[0074] In step S31, the historical context scenes are clustered to generate multiple scene classification sets;

[0075] In step S32, the occurrence frequency and associated node density of each tag node in the tag graph under different scene classification sets are counted;

[0076] In step S33, the correlation coefficient of each tag node in different context scenarios is determined based on the occurrence frequency and the associated node density;

[0077] In step S34, the semantic expression weight of each tag in different context scenarios is determined through all the correlation coefficients.

[0078] It should be noted that the frequency of occurrence in this application is an indicator that reflects the number of times a label node is associated in a specific context scenario; the associated node density in this application is an indicator that measures the degree of connection between a label node and surrounding nodes in a specific scenario; the semantic expression weight in this application is an indicator used to quantify the semantic importance and expression strength of a label node in a specific context scenario.

[0079] In specific implementation, first, based on the contextual information annotated in the historical user behavior data, the clustering algorithm K-means can be used to divide the contextual scenes and generate several scene classification sets; secondly, for each label node in the label graph, its occurrence frequency in different scene classification sets is counted, that is, the number of times the label node is activated or associated in the corresponding scene data segment, and the density of associated nodes between the node and other nodes in the graph in the same scene is calculated. The node degree or local clustering coefficient indicator in graph theory can be used to reflect the connection density; then, based on the occurrence frequency and associated node density, the correlation coefficient of each label node in each scene classification set is calculated by the weighted summation method to characterize the semantic importance of the label in the scene; finally, all correlation coefficients are summarized, and the weighted fusion or maximum value strategy is used to determine the final semantic expression weight of the label in different contextual scenes, so as to realize the dynamic context adaptive expression of label semantics.

[0080] It should be noted that this application performs contextual scene association semantic modeling on the label graph nodes and extracts semantic expression weights, which makes up for the lack of multi-scene adaptability of the traditional label system and enhances the semantic distinction and weight adjustment capabilities of the labels in different application environments, thereby improving the flexibility and accuracy of the label system.

[0081] In step S4, the matching deviation between the current tag and the latest behavior semantic feature of the target object is detected. When the matching deviation exceeds a preset deviation threshold, it is determined that there is a tag semantic deviation.

[0082] In this embodiment, detecting the matching deviation between the current tag and the latest behavior semantic feature of the target object can be achieved by using the following steps:

[0083] mapping the latest behavior semantic feature of the target object to a label graph to generate a to-be-matched label sequence;

[0084] extracting a matching degree of each label in the to-be-matched label sequence with a current label;

[0085] determining a matching deviation degree between the current label and the latest behavior semantic feature of the target object through all the matching degrees.

[0086] It should be noted that the matching degree in the present application is an index for measuring the semantic similarity between two labels; the matching deviation degree in the present application is an index for measuring the difference degree between the current label and the latest behavior semantic feature of the target object.

[0087] In specific implementation, firstly, the latest behavior semantic feature vector of the target object is mapped to a label graph node through a label mapping function to generate a corresponding to-be-matched label sequence, and the mapping process can be completed based on cosine similarity matching retrieval technology to ensure the effective corresponding relationship between the semantic features and the label nodes; secondly, for each label node in the to-be-matched label sequence, the matching degree between the label node and the current label node is calculated, the matching degree index can be measured by cosine similarity, and the semantic expression weight of the node in the label graph is combined for weighted adjustment to reflect the semantic similarity and weight contribution between the labels, that is, for each label in the to-be-matched label sequence, the to-be-matched label and the current label are respectively mapped into semantic vectors with consistent dimensions through a pre-trained semantic vector model (such as Word2Vec), and then the cosine similarity between the semantic vectors of the to-be-matched label and the current label is taken as the initial matching degree, and the initial matching degree is further weighted and corrected in combination with the semantic expression weight of the to-be-matched label in the current context scenario, that is, the calculation formula of the matching degree is: matching degree = initial matching degree x semantic expression weight of to-be-matched label; then, all single label matching degrees are aggregated by weighted average to obtain the matching deviation degree of the current label and the latest behavior semantic feature of the target object as a whole.

[0088] It should be noted that when the matching deviation degree exceeds a preset threshold, it indicates that the current label and the latest behavior semantic feature have significant differences, and the label needs to be updated to maintain semantic accuracy; it should also be noted that the deviation threshold can be set according to historical experience data, which is not limited here.

[0089] In step S5, when the label semantic deviation is detected, a label update strategy is determined according to the semantic expression weight of each label in different context scenarios and the semantic deviation feature of the current label, and then the label of the behavior feature of the target object is updated by the label update strategy.

[0090] In this embodiment, the tag update strategy is determined based on the semantic expression weight of each tag in different context scenarios and the semantic deviation characteristics of the current tag. The following steps can be used to implement it:

[0091] Obtain matching deviation from the semantic deviation feature of the current label;

[0092] Based on the current context, retrieve the semantic expression weight of each tag;

[0093] Calculating a semantic consistency index of the tag based on the matching deviation and all semantic expression weights;

[0094] By presetting threshold rules, the label update operation type is determined based on the semantic consistency index, and the operation type includes adding a label, adjusting the node weight or deleting a semantically invalid node, and then a label update strategy including the operation type, target label and update parameters is generated.

[0095] It should be noted that the semantic consistency index in this application is an indicator that measures the degree of match between the current label state and the semantic characteristics of the target behavior; the preset threshold in this application can be determined through statistical analysis of historical data or empirical tuning methods based on model performance indicators, and is not limited here.

[0096] In the specific implementation, first, based on the current context scenario type, the label semantic expression weight library is called to retrieve the weight information of all relevant labels in the scenario. The weight reflects the semantic importance of the label in a specific scenario and can be quickly accessed through the index structure or cache mechanism; secondly, the matching deviation and the semantic expression weight are combined to calculate the semantic consistency index of the label through fusion, that is, first calculate the average value of all semantic expression weights, and then calculate the semantic consistency index through the following formula: semantic consistency index = (1-matching deviation) × the average value of all semantic expression weights, where 1-matching deviation can reflect the current label and the latest behavior. The fit of the features can represent the overall semantic importance of the relevant labels in the current scenario through the average value of all semantic expression weights; then, the semantic consistency indicators are classified and determined according to the preset threshold rules, and the update operation type of the label is clarified. For example, when the consistency is lower than the deletion threshold, the semantic invalid node is deleted; when it is higher than the new addition threshold, the label is added; when it is between the two, the node weight is dynamically adjusted; finally, the system generates a label update strategy including the operation type, target label node and update parameters (including new weight value, deletion flag), providing specific instructions for subsequent execution modules to realize dynamic maintenance and optimization of the label system.

[0097] In this embodiment, updating the target object behavior feature label by the label update strategy can be achieved by the following steps:

[0098] Locate the label node and associated edge to be operated in the label graph according to the label update strategy;

[0099] Perform the label addition operation, node weight dynamic adjustment and semantic invalid node deletion in the priority order in the update rule, and then generate a versioned label snapshot and update the behavior label of the target object.

[0100] In specific implementation, first, according to the operation instruction in the label update strategy, the label node and its associated edge to be operated are quickly located in the label graph by using the graph database, ensuring that the positioning process has efficient spatial query ability and accurate node identification; then, according to the preset update rule priority order, the label addition operation is performed first, the new label node is inserted into the graph and connected with the related nodes, then the dynamic adjustment of node weight is performed, the change of label importance is reflected by modifying the node weight attribute and edge weight value, finally the deletion operation is performed for the semantic invalid label node, including removing the node and all associated edges, ensuring the cleanliness and effectiveness of the label graph, after each update, the system automatically generates a versioned label snapshot, uses the version number to record the current label graph state, supports subsequent history version backtracking and change comparison, finally the updated label information is synchronized to the user behavior label library, ensuring the timeliness and consistency of the label system.

[0101] It should be noted that in the present application, the label update strategy is determined based on the semantic expression weight and deviation characteristics, and the label addition, weight change or deletion operation is dynamically adjusted combined with the preset rule, which breaks through the limitation of single and lag of traditional label update mechanism, realizes the automatic and fine management of label system, and significantly improves the intelligent level and response ability of label update.

[0102] It can be seen that the present application determines the label update strategy based on the semantic expression weight of each label in different contextual scenarios and the semantic deviation characteristics of the current label, and then updates the label of the target object's behavioral characteristics by the label update strategy; first, the present application scheme is based on structured and unstructured behavioral data to construct a user behavior data system covering multi-dimensional information such as operation records, transaction behaviors and content generation, providing high-quality data support for subsequent semantic feature extraction; secondly, by extracting the behavioral semantic characteristics of users in different contextual scenarios and further modeling the cross-mapping relationship between explicit behavioral characteristics and implicit behavioral characteristics, the label map has stronger semantic expression capabilities, breaking through the limitations of the existing method of static mapping between labels and behaviors; then, the scheme introduces a contextual scenario association semantic modeling mechanism in the label map, and by modeling and quantifying the semantic expression weight of each label node in multiple scenarios, it improves the adaptability and discrimination of label semantics in complex scenarios; finally, by comprehensively analyzing the semantic expression weight and deviation characteristics of each label in different contextual scenarios, the label update strategy is intelligently determined, thereby achieving accurate label updates.

[0103] To summarize, this application builds a closed-loop feedback mechanism from behavioral semantic perception, semantic offset identification to label adaptive update, so that user behavior labels can evolve dynamically with behavior changes, ensuring the timeliness and accuracy of user portraits in digital economic applications, and significantly improving the accuracy and robustness of the platform's intelligent decision-making.

[0104] In the second embodiment, this application provides a digital economy label update system based on big data analysis, referring to Figure 4 As shown, this figure is a schematic diagram of a digital economy label update system based on big data analysis according to this embodiment of the present application. The digital economy label update system based on big data analysis includes:

[0105] The collection module 100 is used to collect structured and unstructured behavior data of the target object in the e-commerce transaction platform, and construct user behavior data including operation records, transaction behaviors, and content generation dimensions;

[0106] A feature processing module 200 is configured to extract the behavior semantic features of the target object in different contextual scenarios from the user behavior data, and then construct a label map reflecting the behavior semantic features of the target object based on the cross-mapping relationship between the explicit behavior features and the implicit behavior features in the behavior semantic features;

[0107] The feature processing module 200 is further used to perform context-related semantic modeling on the tag nodes in the tag graph, and then extract the semantic expression weight of each tag in different contextual scenarios;

[0108] The feature processing module 200 is further configured to detect a matching deviation degree between the current label and the latest behavior semantic feature of the target object, and determine that there is a label semantic deviation when the matching deviation degree exceeds a preset deviation threshold.

[0109] The updating module 300 is configured to determine a label updating strategy according to the semantic expression weight of each label in different context scenarios and the semantic deviation feature of the current label when the label semantic deviation is detected, and update the label of the behavior feature of the target object according to the label updating strategy.

[0110] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system) and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate a device for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The device for implementing the functions specified in one flow or multiple flows and / or one block or multiple blocks. Figure 1 The device for implementing the functions specified in one flow or multiple flows and / or one block or multiple blocks.

[0111] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiments can be completed by programs instructing the related hardware, and the programs can be stored in a computer readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, a magnetic disk storage, a magnetic tape storage, or any other medium capable of carrying or storing data which can be read by a computer.

[0112] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

Claims

1. A digital economy label updating method based on big data analysis, characterized in that: The steps include: Collect structured and unstructured behavioral data of target users on e-commerce trading platforms, and construct user behavior data including operation records, transaction behaviors, and content generation dimensions; Extracting behavioral semantic features of the target object in different contextual scenarios from the user behavior data, and then constructing a label map reflecting the behavioral semantic features of the target object based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features; Performing context-related semantic modeling on the tag nodes in the tag graph, and then extracting the semantic expression weights of each tag in different contexts; Detecting the matching deviation between the current tag and the latest behavior semantic features of the target object. When the matching deviation exceeds a preset deviation threshold, it is determined that there is a tag semantic deviation; When a label semantic deviation is detected, a label update strategy is determined based on the semantic expression weight of each label in different contextual scenarios and the semantic deviation characteristics of the current label, and then the label of the target object's behavioral characteristics is updated by the label update strategy; The process of performing context-related semantic modeling on the tag nodes in the tag graph and then extracting the semantic expression weights of each tag in different contexts specifically includes: Cluster historical context scenarios to generate multiple scenario classification sets; Counting the occurrence frequency and associated node density of each label node in the label map under different scene classification sets; Determine the correlation coefficient of each tag node in different context scenarios based on the occurrence frequency and the associated node density; Determine the semantic expression weight of each tag in different context scenarios through all correlation coefficients; The label update strategy is determined based on the semantic expression weight of each label in different contextual scenarios and the semantic deviation characteristics of the current label, including: Obtain matching deviation from the semantic deviation feature of the current label; Based on the current context, retrieve the semantic expression weight of each tag; Calculating a semantic consistency index of the tag based on the matching deviation and all semantic expression weights; By presetting threshold rules, the label update operation type is determined based on the semantic consistency index, and the operation type includes adding a label, adjusting the node weight or deleting a semantically invalid node, and then a label update strategy including the operation type, target label and update parameters is generated.

2. The method according to claim 1, wherein Extracting the behavioral semantic features of the target object in different context scenarios from the user behavior data specifically includes: Dividing the user behavior data into multiple user behavior data segments according to a preset time window, and labeling each data segment with a context scenario type; For each data segment, extract the semantic entities and behavioral intentions of the unstructured behavioral data in the data segment, and extract the operation frequency and behavioral association path of the structured behavioral data in the data segment; By fusing the semantic entity, the behavioral intention, the operation frequency and the behavioral association path, behavioral semantic features reflecting the target object in the corresponding context scene type are generated, and then the behavioral semantic features of the target object in different context scene types are obtained.

3. The method according to claim 1, wherein The label graph reflecting the target object's behavioral semantic features is constructed based on the cross-mapping relationship between explicit behavioral features and implicit behavioral features in the behavioral semantic features. Specifically, the following are included: Mapping the explicit behavior feature to a root node of a label graph; Mining the cross-dependency relationship between the implicit behavior features and the root node through association rules; The child nodes connected to the root node are determined according to the strength of the cross-dependency relationship, and directed edges between the nodes are constructed with the behavioral semantic association as the weight, so as to obtain a label map reflecting the behavioral semantic characteristics of the target object.

4. The method according to claim 1, wherein The associated node density is an indicator that measures how tightly a label node is connected to surrounding nodes in a specific scenario.

5. The method according to claim 1, wherein Detecting the matching deviation between the current label and the latest behavior semantic features of the target object specifically includes: Map the latest behavioral semantic features of the target object to the label map to generate a sequence of labels to be matched; Extracting the matching degree between each tag in the tag sequence to be matched and the current tag; The matching deviation between the current label and the latest behavior semantic features of the target object is determined through all matching degrees.

6. The method according to claim 1, wherein The labels of the target object's behavioral characteristics updated by the label update strategy specifically include: Locating the label nodes and associated edges to be operated in the label graph according to the label update strategy; The label addition operation, node weight dynamic adjustment, and semantic invalid node deletion are performed according to the priority order in the update rules, thereby generating a versioned label snapshot and updating the behavior label of the target object.

7. The method according to claim 1, wherein The unstructured behavior data refers to user behavior content that cannot be directly expressed in a fixed format.

8. A digital economy label update system based on big data analysis, used to execute the digital economy label update method based on big data analysis according to any one of claims 1 to 7, characterized in that: The digital economy label update system includes: The collection module is used to collect structured and unstructured behavioral data of the target object in the e-commerce trading platform, and construct user behavior data including operation records, transaction behavior, and content generation dimensions; A feature processing module is used to extract the behavioral semantic features of the target object in different contextual scenarios from the user behavior data, and then construct a label map reflecting the behavioral semantic features of the target object based on the cross-mapping relationship between the explicit behavioral features and the implicit behavioral features in the behavioral semantic features; The feature processing module is further used to perform context-related semantic modeling on the tag nodes in the tag graph, and then extract the semantic expression weight of each tag in different contextual scenarios; The feature processing module is further configured to detect a matching deviation between the current label and the latest behavior semantic feature of the target object, and to determine that a label semantic deviation exists when the matching deviation exceeds a preset deviation threshold; The update module is used to determine the label update strategy based on the semantic expression weight of each label in different context scenarios and the semantic deviation characteristics of the current label when a label semantic deviation is detected, and then update the label of the target object behavior feature according to the label update strategy.

Images