User tag generation method and system based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent label generation, in particular to a user label generation method and system based on artificial intelligence, and the method comprises the steps: based on user behavior data, through difference preserving function mapping and behavior label generation, outputting a user behavior label set; and performing label fusion based on the user behavior label set and preset user historical interaction data to generate a comprehensive behavior label set. According to the method and the device, emergent behaviors of the user in different time periods, places and scenes are reserved, and information loss caused by averaging is avoided. The difference preserving mapping does not need to depend on complex weight training, is logically transparent, and can perform online rapid calculation; and meanwhile, traceable input with fine granularity is provided for subsequent steps, so that the real-time interest and intention of the user can be accurately reflected by tag generation, and the response speed and personalized precision of downstream recommendation, risk control and other systems are remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent tag generation, and in particular to a method and system for generating user tags based on artificial intelligence. Background Art

[0002] User portraits are virtual character models built based on real data. They abstract key characteristics of target users, such as demographic attributes, behavioral habits, and demand pain points, and are used to accurately guide product design and marketing strategies.

[0003] In massive internet scenarios, user profiling systems typically rely on analyzing multi-dimensional behavioral data such as clicks, stays, and purchases to categorize users into groups with different interests or identities. Current mainstream solutions mostly use statistical aggregation or deep embedding models to generate user labels. However, statistical aggregation can easily mask key behaviors that change over time, while deep embedding often requires extensive weight training, suffers from poor interpretability, and has high cold start costs.

[0004] In addition, in existing technologies, user behavior features are often extracted through methods such as sliding average, daily and weekly statistics, or simple vector embedding. These methods are not sensitive to capturing fine-grained differences such as short-term fluctuations and cross-regional jumps. Once key changes are diluted, the subsequent label granularity will become too coarse and lack distinction, which will lead to delayed user portrait updates and inaccurate recommendation strategies.

[0005] Therefore, there is an urgent need for a user tag generation method and system based on artificial intelligence. Summary of the Invention

[0006] Based on this, it is necessary to provide an artificial intelligence-based user label generation method and system that addresses the above technical issues by setting a difference-preserving function mapping to divide the original multi-dimensional behavior streams such as clicks, stays, and jumps into time and space type subdomains with business significance, and then determine whether the behavior difference exceeds the preset threshold in each subdomain. If it exceeds, the segment will be explicitly marked as a "significant behavior point". This will produce a set of sparse and highly recognizable discrete features, which not only retains the user's sudden behavior in different time periods, places and scenarios, but also avoids the information loss caused by averaging.

[0007] The technical solutions of the present invention are as follows: A method for generating user tags based on artificial intelligence, the method comprising: Based on user behavior data, a set of user behavior labels is output through difference-preserving function mapping and behavior label generation. Perform tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set; Performing semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set; The final user tag set is subjected to specificity screening to output an accurate user tag set.

[0008] Specifically, based on user behavior data, through difference-preserving function mapping and behavior label generation, a set of user behavior labels is output, including: Based on user behavior data, a user behavior feature set is generated through difference-preserving function mapping; Based on the user behavior feature set, a user behavior tag set is output through behavior tag generation.

[0009] Specifically, based on user behavior data, a user behavior feature set is generated through difference-preserving function mapping, including: Dividing the user behavior data into a plurality of behavior subdomains with preset differentiation; For each of the behavior subdomains, calculating a change degree parameter of its local behavior sequence; Comparing the change degree parameter with a preset feature threshold through a difference preservation function, and encoding the local behavior sequence as a discrete feature if the change degree parameter is greater than the feature threshold; Aggregate the discrete features of all the behavior subdomains to generate a user behavior feature set.

[0010] Specifically, based on the user behavior feature set, a user behavior tag set is output through behavior tag generation, including: Based on the user behavior feature set, obtaining a behavior tag to be activated, and constructing a feature response channel according to the behavior tag to be activated, wherein each feature response channel corresponds to one behavior tag to be activated; Calculating each of the feature response channels to obtain a feature matching cumulative value; Comparing the feature matching cumulative value with a preset activation threshold, if the feature matching cumulative value is greater than the activation threshold, activating the corresponding behavior tag to be activated and generating a user behavior tag; Aggregate all the user behavior tags and output a user behavior tag set.

[0011] Specifically, based on the user behavior tag set and the preset user historical interaction data, a tag fusion is performed to generate a comprehensive behavior tag set, including: Constructing a perturbation function based on the user behavior tag set, wherein each of the perturbation functions corresponds to a behavior tag in the user behavior tag set; Calculate and obtain a response disturbance value based on the disturbance function and preset user historical interaction data; Based on the response disturbance value, the user behavior feature set is subjected to label fusion to generate a comprehensive behavior label set.

[0012] Specifically, semantic mapping modeling is performed on the comprehensive behavior tag set to obtain a final user tag set, including: Perform semantic mapping modeling based on the comprehensive behavior tag set and the preset candidate user tag library to calculate and obtain the overall semantic score; The candidate tag library is screened based on the overall semantic score to generate a final user tag set.

[0013] Specifically, the final user tag set is subjected to a specificity screening to output an accurate user tag set, including: Calculating each of the end-user tags in the end-user tag set to generate a specificity index; Comparing the specificity index with a preset screening threshold, if the specificity index is greater than the screening threshold, marking the corresponding final user tag as a precise user tag; Aggregate the precise user tags and output a precise user tag set.

[0014] Specifically, a user tag generation system based on artificial intelligence is also provided, the system comprising: A user behavior label generation module is used to output a user behavior label set based on user behavior data through difference-preserving function mapping and behavior label generation; A comprehensive behavior tag generation module is used to generate a comprehensive behavior tag set based on the user behavior tag set and preset user historical interaction data by tag fusion; An end-user tag generation module is used to perform semantic mapping modeling on the comprehensive behavior tag set to obtain an end-user tag set; The precise user tag generation module is used to perform specificity screening on the final user tag set and output a precise user tag set.

[0015] Specifically, the user behavior label generation module is further configured to: generate a user behavior feature set based on user behavior data through difference-preserving function mapping; and output a user behavior label set through behavior label generation based on the user behavior feature set.

[0016] Specifically, the user behavior label generation module is also used to: divide the user behavior data into several behavior subdomains with preset distinctions; for each of the behavior subdomains, calculate the change degree parameter of its local behavior sequence; compare the change degree parameter with a preset feature threshold through a difference preservation function, and if the change degree parameter is greater than the feature threshold, encode the local behavior sequence as a discrete feature; aggregate the discrete features of all several of the behavior subdomains to generate a user behavior feature set.

[0017] Specifically, the user behavior label generation module is also used to: obtain the behavior label to be activated based on the user behavior feature set, and construct a feature response channel according to the behavior label to be activated, wherein each feature response channel corresponds to one behavior label to be activated; calculate each feature response channel to obtain a feature matching cumulative value; compare the feature matching cumulative value with a preset activation threshold, if the feature matching cumulative value is greater than the activation threshold, activate the corresponding behavior label to be activated and generate a user behavior label; aggregate all the user behavior labels and output a user behavior label set.

[0018] Specifically, the comprehensive behavior label generation module is also used to: construct a perturbation function based on the user behavior label set, wherein each perturbation function corresponds to a behavior label in the user behavior label set; calculate and obtain a response disturbance value based on the perturbation function and preset user historical interaction data; and perform label fusion on the user behavior feature set based on the response disturbance value to generate a comprehensive behavior label set.

[0019] Specifically, the final user tag generation module is also used to: perform semantic mapping modeling based on the comprehensive behavior tag set and the preset candidate user tag library, and calculate the overall semantic score; screen the candidate tag library based on the overall semantic score to generate the final user tag set.

[0020] Specifically, the precise user tag generation module is also used to: calculate and generate a specificity index for each of the final user tags in the final user tag set; compare the specificity index with a preset screening threshold, and if the specificity index is greater than the screening threshold, mark the corresponding final user tag as a precise user tag; aggregate the precise user tags and output a precise user tag set.

[0021] Optionally, a computer device is also provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps described in the above-mentioned artificial intelligence-based user tag generation method when executing the computer program.

[0022] Optionally, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, the steps described in the above-mentioned artificial intelligence-based user tag generation method are implemented.

[0023] The present invention relates to machine learning and deep learning technologies, and the technical effects achieved are as follows: The above-mentioned AI-based user tag generation method and system sequentially outputs a user behavior tag set based on user behavior data through difference-preserving function mapping and behavior tag generation; performs tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set; performs semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set; and performs attribute-specific screening on the final user tag set to output a precise user tag set. Furthermore, by setting a difference-preserving function mapping, the original multi-dimensional behavior streams such as clicks, stays, and jumps are divided into time and space type subdomains with business significance. Then, within each subdomain, it is determined whether the behavior difference exceeds a preset threshold. If so, the segment is explicitly marked as a "significant behavior point" processing strategy, thereby generating a set of sparse and highly recognizable discrete features, which not only retains the user's sudden behavior in different time periods, places, and scenarios, but also avoids the information loss caused by averaging. Difference-preserving mapping does not require complex weight training, has transparent logic, and can be quickly calculated online. It also provides fine-grained, traceable input for subsequent steps, enabling label generation to accurately reflect users' real-time interests and intentions, significantly improving the response speed and personalization accuracy of downstream recommendation, risk control, and other systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Schematic diagram of a flow chart of a method for generating user tags based on artificial intelligence in one embodiment; Figure 2 A structural block diagram of an artificial intelligence-based user tag generation system in one embodiment. DETAILED DESCRIPTION

[0025] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0026] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0027] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0028] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0029] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0030] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0031] In one embodiment, a terminal is provided, which is used to: output a user behavior label set based on user behavior data through difference-preserving function mapping and behavior label generation; generate a comprehensive behavior label set based on label fusion of the user behavior label set and preset user historical interaction data; perform semantic mapping modeling on the comprehensive behavior label set to obtain a final user label set; perform specificity screening on the final user label set to output a precise user label set.

[0032] The terminal may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices.

[0033] In one embodiment, Figure 1 As shown, a method for generating user tags based on artificial intelligence is provided, the method comprising: Step S100: Based on user behavior data, output a user behavior label set through difference-preserving function mapping and behavior label generation; Step S200: performing tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set; Step S300: performing semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set; Step S400: performing a specificity screening on the final user tag set and outputting an accurate user tag set.

[0034] This application is based on the user tag generation method and system based on artificial intelligence, which outputs a user behavior tag set in sequence through difference-preserving function mapping and behavior tag generation based on user behavior data; performs tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set; performs semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set; performs specificity screening on the final user tag set to output an accurate user tag set. First, the original behavior stream is sub-domain sliced ​​and threshold-judged to produce a sparse and highly recognizable user behavior feature set, and then these difference features are activated into an interpretable user behavior tag set according to the preset channel. The behavior tags are then placed in a long-term interaction sequence for perturbation statistics, retaining a true and stable comprehensive behavior tag set, and then according to the tag-semantic mapping rules, the comprehensive tags are aggregated into a final user tag for the portrait. Finally, a specificity scoring threshold is introduced to eliminate edge tags to form a concise and accurate user tag set, realizing a layer-by-layer abstraction and refined closed loop from the original behavior to the high-quality portrait tag. Therefore, this application divides the original multi-dimensional behavior streams such as clicks, stays, and jumps into time and space type subdomains with business significance by setting a difference-preserving function mapping, and then determines whether the behavior difference exceeds the preset threshold in each subdomain. If it exceeds, the segment is explicitly marked as a "significant behavior point" processing strategy, thereby generating a set of sparse and highly recognizable discrete features, which not only retains the user's sudden behavior in different time periods, places and scenarios, but also avoids the information loss caused by averaging. Difference-preserving mapping does not need to rely on complex weight training, has transparent logic, and can be quickly calculated online; at the same time, it provides fine-grained and traceable input for subsequent steps, so that label generation can accurately reflect the user's real-time interests and intentions, significantly improving the response speed and personalization accuracy of downstream recommendation, risk control and other systems.

[0035] In one embodiment, step S100: Based on the user behavior data, outputting a user behavior tag set through difference-preserving function mapping and behavior tag generation includes: Step S110: generating a user behavior feature set based on the user behavior data through difference-preserving function mapping; Step S120: Based on the user behavior feature set, generate behavior tags and output a user behavior tag set.

[0036] In this embodiment, a structured user behavior feature set that can be used for subsequent user tag generation is extracted from the original user behavior data, reflecting the significant difference patterns of users in multiple behavior dimensions, and further tags are generated for user behaviors by constructing a behavior tag generation function to generate a user behavior tag set.

[0037] In one embodiment, step S110: generating a user behavior feature set based on user behavior data through difference-preserving function mapping includes: Step S111: Divide the user behavior data into a number of behavior subdomains with preset differentiation; Step S112: for each of the behavior subdomains, calculating a change degree parameter of its local behavior sequence; Step S113: comparing the change degree parameter with a preset feature threshold using a difference preservation function; if the change degree parameter is greater than the feature threshold, encoding the local behavior sequence as a discrete feature; Step S114: Aggregate the discrete features of all the behavior subdomains to generate a user behavior feature set.

[0038] In this example, the original user behavior data is split into subdomains, and the degree of change parameter is calculated. A difference-preserving function is then used to determine whether the local behavior exceeds the feature threshold. All local behavior sequences judged to be "significant" are encoded as discrete features, and ultimately all discrete features are concatenated to form a user behavior feature set.

[0039] Furthermore, the user behavior feature set is set as follows:

[0040] in, For users The behavioral feature set, The number of dimensions for dividing the set behavioral feature subdomains, For users The local behavior sequence of For local behavior sequence In the The difference operator mapping is performed within the partition interval. For the The local behavior feature threshold of the interval, is the difference preserving function, .

[0041] The user Local behavior sequence Contains multi-dimensional attributes, such as timestamp, location information and behavior category, The local behavior feature threshold of the interval Used to control the sensitivity of feature extraction. This threshold parameter is set by aggregate statistics after comprehensively considering the frequency of user historical behavior, the distribution density of behavior tags within the time window, and the stability of behavior patterns. The system first counts the number of occurrences of each type of behavior tag in multiple time intervals, and calculates the minimum stable occurrence frequency of each type of tag based on these statistical values; then, combining this frequency with the degree of fluctuation of user behavior in different time periods, it constructs a dynamically adaptable interval local behavior feature threshold. , used to determine whether the behavior label has sufficient historical support, the difference preservation function The difference is mapped into a discrete feature representation to determine whether the behavior deviates significantly from the expected value within the local range.

[0042] Furthermore, the difference preserving function The setting method is as follows: ; The local behavior sequence In the Difference operator mapping within a partition interval Reflects the degree of change in behavioral activities within this interval.

[0043] Furthermore, the difference operator mapping The setting method is as follows:

[0044] in, is the number of sub-time slices used to count the behavior interval, For users In the interval No. The number of behavioral events in a sub-time slice, For interval The average number of behaviors under

[0045] Furthermore, the interval The average number of behaviors under The calculation method is as follows:

[0046] It's important to note that existing behavioral feature extraction generally relies on two mature approaches: statistical aggregation (e.g., daily average clicks, weekly conversion rates), and low-level weight embedding (e.g., RNN / Transformer behavioral encoding). The former performs averaging or summing operations within a window, which can obscure short-term, drastic changes; the latter requires pre-training a large number of learnable weights, resulting in poor model interpretability and poor cold-start user friendliness.

[0047] Different from the prior art, this application uses an explicit feature threshold Make a hard distinction between "abnormal / non-abnormal" to ensure that mutation behavior is accurately captured; use interpretable discrete features , eliminating black box weights and facilitating direct tracing back to specific time periods or geographical areas; processing the differences in time, space, and type dimensions simultaneously within the same formula framework, allowing subsequent steps to directly perform label clustering under unsupervised conditions, reducing the model's cold start dependency.

[0048] In one embodiment, step S120: generating a user behavior tag set based on the user behavior feature set by behavior tag generation includes: Step S121: Based on the user behavior feature set, obtain the behavior tags to be activated, and construct feature response channels according to the behavior tags to be activated, wherein each feature response channel corresponds to one behavior tag to be activated; Step S122: Calculate each of the feature response channels to obtain a feature matching cumulative value; Step S123: comparing the feature matching cumulative value with a preset activation threshold; if the feature matching cumulative value is greater than the activation threshold, activating the corresponding behavior tag to be activated and generating a user behavior tag; Step S124: Aggregate all the user behavior tags and output a user behavior tag set.

[0049] In this embodiment, the system is developed around a one-time mapping of "differential features-semantic labels". First, a discretized user behavior feature set is loaded, and a feature response channel is preset for each user behavior label to be activated to clarify the type of differential feature trigger required for the label. Then, the user behavior feature set is scanned in turn for each feature response channel, a Boolean response is made to the hit feature bit, and the feature matching value is accumulated. Subsequently, the feature matching accumulated value is compared with the activation threshold corresponding to the label to determine whether the user truly meets the semantic pattern of the label to be activated. Finally, a one-time collection is performed on all activated user behavior labels to generate a user behavior label set for the current user, thereby achieving efficient, zero-weight merging from high-dimensional differential features to explainable behavior labels.

[0050] Furthermore, the user behavior tag set generation function is set as follows:

[0051] in, For users The set of behavior tags, is the number of user behavior label categories, To obtain the User behavior characteristics, For the The feature response channel that the user behavior label function focuses on, For the The activation threshold of the user behavior label function, Based on user behavior characteristics The mapping generated User behavior tags, Activation function for behavior labels.

[0052] The user behavior label function represents a behavior label semantic structure, and the feature response channel Only responding to behavioral features of a specific dimension, the output value is {0,1}, and the activation threshold It is determined based on the product relationship between the maximum number of responses of the user behavior label function in the user behavior feature response and the preset activation threshold, and is obtained by multiplying the maximum number of responses with a proportional factor representing the required response strength and rounding it off. The proportional factor is a preset label response coverage standard.

[0053] Furthermore, the feature pair The mapping generated Behavior tags The settings are as follows:

[0054] Furthermore, the behavior tag activation function The details are as follows:

[0055] in, That is , is the The activation threshold that a behavior tag needs to meet, Cumulative value for feature matching.

[0056] The activation threshold Specifies the minimum number of features that must be met for a tag to be "activated", and the feature matching cumulative value Indicates the number of behavioral characteristics that the user meets under a certain behavior tag definition.

[0057] Furthermore, the feature matching cumulative value The setup is as follows:

[0058] in, Indicates the Does the behavioral characteristic meet the The characteristic channel response conditions defined by the behavior tags.

[0059] It's important to note that traditional behavior label generation typically uses bag-of-words, TFIDF, K-means clustering, or deep sentence embedding with learnable weights / contrastive learning. The former applies discrete features to unsupervised clustering, resulting in semantically random labels that require manual naming. While the latter can output label embeddings end-to-end, it relies heavily on large-scale annotated data and parameter training, and the output is difficult to interpret manually.

[0060] Different from the prior art, this application first uses the characteristic response channel First define "what features the label needs to respond to", then count to avoid clustering "deviation"; secondly, through the activation threshold It uses a business-configurable threshold to determine whether to generate a label, and controls the label granularity and sparsity, rather than letting the model learn opaque weights on its own. Finally, it achieves zero weight and zero iteration. The entire set of formulas only involves Boolean discrimination and addition, which is easier to implement than deep models or K-Means in cold start and edge device scenarios. It also facilitates the subsequent steps to directly perform historical perturbation correction and semantic aggregation on the labels. In summary, this application avoids the drawbacks of traditional clustering "uncontrollable semantics" and the defects of deep weight models "heavy training-difficult to explain", and realizes an efficient, transparent and easy-to-maintain user behavior label generation mechanism.

[0061] In one embodiment, step S200: performing tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set includes: Step S210: constructing a perturbation function based on the user behavior tag set, wherein each perturbation function corresponds to a behavior tag in the user behavior tag set; Step S220: Calculate and obtain a response disturbance value based on the disturbance function and preset user historical interaction data; Step S230: Based on the response disturbance value, perform label fusion on the user behavior feature set to generate a comprehensive behavior label set.

[0062] In this embodiment, the user's historical interaction data is first introduced, and a corresponding historical hit counter is established for each user behavior label through a perturbation function; then the interaction data is traversed in chronological order, and the number of hits is accumulated when a difference feature that matches the semantics of the user behavior label is encountered, forming a response perturbation value for the user behavior label. Next, the response perturbation value of each user behavior label is compared with the preset perturbation threshold. User behavior labels above the threshold are identified as stable behavior patterns and enter the "retention-refinement" label fusion, with additional contextual markers such as the scene or time period; user behavior labels below the threshold are classified into the "elimination" branch and are removed as occasional noise. After completing the threshold diversion of all labels, the retention and refinement results are subjected to label fusion to form a comprehensive behavior label set.

[0063] Furthermore, the comprehensive tag set is set as follows:

[0064] in, For users A comprehensive set of behavioral tags, is a set of user behavior tags, For the User behavior tags, To integrate the comprehensive behavior labels after the historical interaction features, For users Labeling in user historical interaction data The response disturbance value, is the label perturbation fusion function.

[0065] The user Labeling in user historical interaction data The response disturbance value To measure users Labeling user behavior in historical interactions The actual response strength, the label perturbation fusion function Used to adjust the expression form of the label or retain / remove status according to the response disturbance value Is it above the disturbance threshold? , decide whether to keep and semantically enhance the tag, or remove the tag.

[0066] Furthermore, the user Labeling user behavior in historical interaction data The response disturbance value Here’s how to set it up:

[0067] in, For users In the User behavior features extracted from interaction behaviors, is the total number of interaction behaviors in the user's historical interaction data, User behavior labels The behavior label activation function of the input user behavior feature is used to determine the semantic match between the user behavior feature and the user behavior label.

[0068] Furthermore, the label perturbation function The specific settings are as follows:

[0069] in, For label The disturbance threshold represents the critical value of the label's "effective" response in historical behavior.

[0070] It is important to note that mainstream practices often use the following methods to correct comprehensive behavioral labels: The statistical weighted method directly uses the frequency of occurrence of behavioral labels as the accumulated weight, resulting in the unconditional retention of high-frequency but low-discrimination labels. The temporal deep network uses RNN / Transformer to learn label sequences, but this requires a large number of parameters and a long training cycle.

[0071] Different from the existing technology, this application has the characteristics of nonlinear but interpretable, zero-weight training, and business configurability. and disturbance threshold It performs hard gating and allows Refine to embed scene modifiers while retaining labels, achieving "retention + semantic gain" rather than simple weighting to achieve nonlinear but interpretable results; all decision points are driven by explicit thresholds and do not rely on learnable weights, avoiding cold start and overfitting of deep models and thus achieving zero-weight training; due to the perturbation threshold Refine rules can be adjusted instantly by the operations side based on business strategies, offering far greater flexibility than deep end-to-end models for achieving business configurability. Through these three key points, this application ensures the stability of comprehensive behavioral labels while retaining the lightweight, easy-to-interpret, and maintainable nature of the method, forming the innovative "historical authenticity verification" component of this application.

[0072] In one embodiment, step S300: performing semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set includes: Step S310: performing semantic mapping modeling based on the comprehensive behavior tag set and the preset candidate user tag library, and calculating and obtaining an overall semantic score; Step S320: screening the candidate tag library based on the overall semantic score to generate a final user tag set.

[0073] In this embodiment, the diverse comprehensive behavioral tag set from the behavioral level is converted into a structured, categorizable, and final user tag set that can be used to identify user attributes, which plays a role in classification and induction and provides clear input for subsequent screening.

[0074] Furthermore, the final user tag set generation method is set as follows:

[0075] in, For users The collection of end-user tags for is the candidate user label, Comprehensive behavioral label and candidate user tags The overall semantic score between Induce the mapping function for the labels.

[0076] The candidate user tags Derived from the candidate user tag library, the comprehensive behavior tag and candidate user tags The overall semantic score between The label induction mapping function is calculated by the word vector cosine similarity. Infer the end user label set based on the input label set.

[0077] Furthermore, the overall semantic score The setup is as follows:

[0078] in, Comprehensive behavioral label The semantic vector representation of Candidate user labels Semantic vector representation of .

[0079] In one embodiment, step S400: performing a specificity screening on the final user tag set to output an accurate user tag set includes: Step S410: Calculating each of the end-user tags in the end-user tag set to generate a specificity index; Step S420: comparing the specificity index with a preset screening threshold, and if the specificity index is greater than the screening threshold, marking the corresponding final user tag as a precise user tag; Step S430: Aggregate the precise user tags and output a precise user tag set.

[0080] In this embodiment, the precise user tag set generation function is set as follows:

[0081] in, For precise user tag sets, Labels for end users The specificity index is the PageRank weight of the tag in the user tag network. Labels for preset end-user tags The minimum screening threshold.

[0082] In one embodiment, based on the above-mentioned AI-based user tag generation method, with e-commerce platforms as the application direction, the process of generating a precise user tag set for the e-commerce platform includes: 1. E-commerce platforms collect user browsing, searching, purchasing, and review behavior data to extract user behavior characteristics. Based on these characteristics, they construct a set of user behavior labels, such as "stable purchasing users," "frequent purchasing users," "price-sensitive users," and "promotion-preferring users." 2. Integrate user behavior tag collections with historical user interaction data, including promotional response status and activity participation status, to dynamically adjust the tag weight and activation level; 3. Generate a collection of end-user tags for use in recommendation systems and marketing activities; 4. Generate an accurate user tag set by screening and eliminating unstable or misjudged tags to ensure the accuracy and applicability of the tags.

[0083] In one embodiment, Figure 2 As shown, a user tag generation system based on artificial intelligence is also provided, and the system includes: A user behavior label generation module is used to output a user behavior label set based on user behavior data through difference-preserving function mapping and behavior label generation; A comprehensive behavior tag generation module is used to generate a comprehensive behavior tag set based on the user behavior tag set and preset user historical interaction data by tag fusion; An end-user tag generation module is used to perform semantic mapping modeling on the comprehensive behavior tag set to obtain an end-user tag set; The precise user tag generation module is used to perform specificity screening on the final user tag set and output a precise user tag set.

[0084] In another embodiment, the user behavior label generation module is further configured to: generate a user behavior feature set based on user behavior data through difference preserving function mapping; and output a user behavior label set through behavior label generation based on the user behavior feature set.

[0085] In another embodiment, the user behavior label generation module is further used to: divide the user behavior data into several behavior subdomains with preset distinctions; for each of the behavior subdomains, calculate the change degree parameter of its local behavior sequence; compare the change degree parameter with a preset feature threshold through a difference preservation function, and if the change degree parameter is greater than the feature threshold, encode the local behavior sequence as a discrete feature; aggregate the discrete features of all several of the behavior subdomains to generate a user behavior feature set.

[0086] In another embodiment, the user behavior label generation module is further used to: obtain the behavior label to be activated based on the user behavior feature set, and construct a feature response channel according to the behavior label to be activated, wherein each feature response channel corresponds to one behavior label to be activated; calculate each feature response channel to obtain a feature matching cumulative value; compare the feature matching cumulative value with a preset activation threshold, and if the feature matching cumulative value is greater than the activation threshold, activate the corresponding behavior label to be activated and generate a user behavior label; aggregate all the user behavior labels and output a user behavior label set.

[0087] In another embodiment, the comprehensive behavior label generation module is further used to: construct a perturbation function based on the user behavior label set, wherein each of the perturbation functions corresponds to a behavior label in the user behavior label set; calculate and obtain a response disturbance value based on the perturbation function and preset user historical interaction data; and perform label fusion on the user behavior feature set based on the response disturbance value to generate a comprehensive behavior label set.

[0088] In another embodiment, the final user tag generation module is further used to: perform semantic mapping modeling based on the comprehensive behavior tag set and the preset candidate user tag library, and calculate and obtain the overall semantic score; screen the candidate tag library based on the overall semantic score to generate the final user tag set.

[0089] In another embodiment, the precise user tag generation module is further used to: calculate and generate a specificity index for each of the end user tags in the end user tag set; compare the specificity index with a preset screening threshold, and if the specificity index is greater than the screening threshold, mark the corresponding end user tag as a precise user tag; aggregate the precise user tags and output a precise user tag set.

[0090] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps described in the above-mentioned artificial intelligence-based user tag generation method when executing the computer program.

[0091] In one embodiment, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based user tag generation method are implemented.

[0092] It should be noted that the information interaction, execution process and other contents between the above modules are based on the same concept as the method embodiment of this application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0093] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0094] It should be noted that the information interaction, execution process and other contents between the above modules are based on the same concept as the method embodiment of this application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0095] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0096] An embodiment of the present application also provides a network device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps of any of the above-mentioned method embodiments when executing the computer program.

[0097] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0098] An embodiment of the present application provides a computer program product. When the computer program product is run on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned various method embodiments when executing the computer program product.

[0099] If the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a camera / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, removable hard drives, magnetic disks, or optical disks. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals or telecommunication signals.

[0100] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0101] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0102] In the embodiments provided in this application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0103] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0104] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

[0105] An embodiment of the present application also provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any embodiment of the above method when executing the computer program.

[0106] The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above description is an example of a computer device and does not limit the computer device. The computer device may include more or fewer components than described above, or a combination of certain components, or different components. For example, the computer device may also include input / output devices, network access devices, etc.

[0107] The processor may be a central processing unit (CPU), and the processor 0 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0108] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped with the computer device. Furthermore, the memory may include both an internal storage unit of the computer device and an external storage device. The memory is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is about to be output.

[0109] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0110] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and such modifications and improvements are intended to fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for generating user tags based on artificial intelligence, characterized in that: The method comprises: Based on user behavior data, a set of user behavior labels is output through difference-preserving function mapping and behavior label generation. Perform tag fusion based on the user behavior tag set and preset user historical interaction data to generate a comprehensive behavior tag set; Performing semantic mapping modeling on the comprehensive behavior tag set to obtain a final user tag set; The final user tag set is subjected to specificity screening to output an accurate user tag set.

2. The method for generating user tags based on artificial intelligence according to claim 1, characterized in that: Based on the user behavior data, the user behavior label set is output through difference-preserving function mapping and behavior label generation, including: Based on user behavior data, a user behavior feature set is generated through difference-preserving function mapping; Based on the user behavior feature set, a user behavior tag set is output through behavior tag generation.

3. The method for generating user tags based on artificial intelligence according to claim 2, characterized in that: The method of generating a user behavior feature set based on user behavior data by mapping a difference-preserving function includes: Dividing the user behavior data into a plurality of behavior subdomains with preset differentiation; For each of the behavior subdomains, calculating a change degree parameter of its local behavior sequence; Comparing the change degree parameter with a preset feature threshold through a difference preservation function, and encoding the local behavior sequence as a discrete feature if the change degree parameter is greater than the feature threshold; Aggregate the discrete features of all the behavior subdomains to generate a user behavior feature set.

4. The method for generating user tags based on artificial intelligence according to claim 2, characterized in that: The step of generating a user behavior tag set based on the user behavior feature set includes: Based on the user behavior feature set, obtaining a behavior tag to be activated, and constructing a feature response channel according to the behavior tag to be activated, wherein each feature response channel corresponds to one behavior tag to be activated; Calculating each of the feature response channels to obtain a feature matching cumulative value; Comparing the feature matching cumulative value with a preset activation threshold, if the feature matching cumulative value is greater than the activation threshold, activating the corresponding behavior tag to be activated and generating a user behavior tag; Aggregate all the user behavior tags and output a user behavior tag set.

5. The method for generating user tags based on artificial intelligence according to claim 1, characterized in that: The step of fusing the user behavior tag set with preset user historical interaction data to generate a comprehensive behavior tag set includes: Constructing a perturbation function based on the user behavior tag set, wherein each of the perturbation functions corresponds to a behavior tag in the user behavior tag set; Calculate and obtain a response disturbance value based on the disturbance function and preset user historical interaction data; Based on the response disturbance value, the user behavior feature set is subjected to label fusion to generate a comprehensive behavior label set.

6. The method for generating user tags based on artificial intelligence according to claim 1, characterized in that: The semantic mapping modeling is performed on the comprehensive behavior tag set to obtain the final user tag set, including: Perform semantic mapping modeling based on the comprehensive behavior tag set and the preset candidate user tag library to calculate and obtain the overall semantic score; The candidate tag library is screened based on the overall semantic score to generate a final user tag set.

7. The method for generating user tags based on artificial intelligence according to claim 1, characterized in that: The specificity screening of the final user tag set to output an accurate user tag set includes: Calculating each of the end-user tags in the end-user tag set to generate a specificity index; Comparing the specificity index with a preset screening threshold, if the specificity index is greater than the screening threshold, marking the corresponding final user tag as a precise user tag; Aggregate the precise user tags and output a precise user tag set.

8. A user tag generation system based on artificial intelligence, characterized in that: The system comprises: A user behavior label generation module is used to output a user behavior label set based on user behavior data through difference-preserving function mapping and behavior label generation; A comprehensive behavior tag generation module is used to generate a comprehensive behavior tag set based on the user behavior tag set and preset user historical interaction data by tag fusion; An end-user tag generation module is used to perform semantic mapping modeling on the comprehensive behavior tag set to obtain an end-user tag set; The precise user tag generation module is used to perform specificity screening on the final user tag set and output a precise user tag set.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images