Consumption coupon issuing control method and system based on big data

Abstract

The invention discloses a consumption coupon issuing control method and system based on big data, and the method comprises the steps: collecting the time sequence information of the consumption behavior data of a target user after obtaining the authorization of the target user; a data processing and semantic comprehension algorithm based on artificial intelligence and natural language analysis is introduced at the rear end to analyze time sequence information of the consumption behavior data of the target users, so that feature dynamic aggregation information of the consumption behavior data semantics of the target users in the time sequence is captured; therefore, consumption coupon type recommendation is performed for the target user, and a detailed user portrait can be constructed based on the preference and consumption habit of the consumer so as to predict the consumption behavior, thereby realizing personalized consumption coupon recommendation and issuing.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to a consumption voucher issuance control method and system based on big data. Background Art

[0002] Consumer vouchers are an economic tool used to stimulate consumption and boost economic growth. They are usually in electronic or paper form, and consumers can use them to offset part of the cost at specific merchants or services. The original intention of consumer vouchers was to promote short-term consumption growth and stimulate market demand by providing consumers with additional purchasing power.

[0003] However, in traditional consumption voucher distribution strategies, a "one-size-fits-all" approach is often adopted, that is, the same type of consumption vouchers are distributed indiscriminately to all users. This approach not only ignores the huge differences between consumers and makes it difficult to meet the personalized needs of different users, but may also lead to waste of resources and discounted marketing effects, thereby reducing the actual effectiveness of the consumption vouchers.

[0004] Therefore, an optimized consumption voucher issuance control method and system are needed to solve the above technical problems. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a consumption voucher issuance control method and system based on big data, which collects the time series information of the consumption behavior data of the target user after obtaining the authorization of the target user, and introduces data processing and semantic understanding algorithms based on artificial intelligence and natural language analysis in the back end to analyze the time series information of the consumption behavior data of these target users, so as to capture the dynamic aggregation information of the semantic characteristics of the consumption behavior data of the target user in the time series, so as to recommend the type of consumption vouchers for the target user, and build a detailed user portrait based on the consumer's preferences and consumption habits to predict consumption behavior, thereby realizing personalized consumption voucher recommendation and issuance.

[0006] According to one aspect of the present application, a consumption voucher issuance control system based on big data is provided, which includes: A user consumption behavior data collection module is used to obtain a data set of consumption behavior data of a target user after obtaining authorization from the target user; A user consumption behavior data semantic coding module, used to perform time sequence regularization and semantic embedding coding on the target user's consumption behavior data set to obtain a time queue of the consumption behavior data semantic embedding coding features; A consumption behavior semantic temporal association coding module is used to perform consumption behavior temporal association coding on the time queue of the semantic embedding coding feature of the consumption behavior data to obtain a time queue of the semantic temporal association coding feature of the consumption behavior data; A consumer behavior semantic temporal information aggregation module is used to perform dynamic propagation aggregation processing of consumer behavior temporal features on the time queue of semantic temporal associated coding features of the consumer behavior data to obtain a significant propagation aggregation representation of consumer behavior temporal features; The consumption voucher recommendation module is used to recommend consumption vouchers based on the consumption behavior time series significant propagation aggregation representation to obtain a recommended type label of the consumption voucher.

[0007] According to another aspect of the present application, a consumption voucher issuance control method based on big data is provided, which includes: After obtaining authorization from the target user, obtain a data set of consumption behavior data of the target user; Performing time sequence regularization and semantic embedding coding on the data set of the consumption behavior data of the target user to obtain a time queue of semantic embedding coding features of the consumption behavior data; Performing consumer behavior temporal association coding on the time queue of the semantic embedding coding feature of the consumer behavior data to obtain a time queue of the semantic temporal association coding feature of the consumer behavior data; Performing dynamic propagation aggregation processing of consumer behavior temporal features on the time queue of semantic temporal correlation coding features of the consumer behavior data to obtain a significant propagation aggregation representation of consumer behavior temporal features; Based on the time series significant propagation aggregation representation of the consumption behavior, consumption voucher recommendations are made to obtain a recommended type label of the consumption voucher.

[0008] The present application has at least the following technical effects: Compared with the prior art, the present application provides a consumption voucher issuance control method and system based on big data, which, after obtaining the authorization of the target user, collects the time series information of the target user's consumption behavior data, and introduces data processing and semantic understanding algorithms based on artificial intelligence and natural language analysis at the back end to analyze the time series information of the consumption behavior data of these target users, so as to capture the dynamic aggregation information of the semantic features of the target user's consumption behavior data in the time series, thereby recommending the type of consumption vouchers for the target user, and being able to build a detailed user portrait based on the consumer's preferences and consumption habits, so as to predict consumption behavior, thereby realizing personalized consumption voucher recommendations and issuance. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 It is a block diagram of a consumption voucher issuance control system based on big data according to an embodiment of the present application; Figure 2 A schematic diagram of data flow of a consumption voucher issuance control system based on big data according to an embodiment of the present application; Figure 3 It is a flowchart of sub-step S4 of the consumption voucher issuance control system based on big data according to an embodiment of the present application; Figure 4 This is a flowchart of a consumption voucher issuance control method based on big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0011] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0012] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0013] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0014] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0015] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0016] In the traditional consumption voucher issuance strategy, a "one-size-fits-all" approach is often adopted, that is, the same type of consumption vouchers are issued to all users without distinction. This approach not only ignores the huge differences between consumers and is difficult to meet the personalized needs of different users, but also may lead to waste of resources and discount of marketing effect, thereby reducing the actual utility of consumption vouchers. Therefore, an optimized consumption voucher issuance control system is desired.

[0017] With the rapid development of information technology, big data analysis has become an important tool for enterprises to optimize operations and improve user experience. Especially in consumer fields such as retail and e-commerce, by analyzing massive consumer behavior data, we can deeply understand consumers' purchasing preferences, consumption habits and other information, and then provide users with more personalized and accurate services.

[0018] Based on this, in the technical solution of this application, a consumption voucher issuance control system based on big data is proposed, which can use big data technology and artificial intelligence technology to integrate consumers' behavioral data and consumption habits, thereby constructing a detailed user portrait, and providing each user with more targeted activity voucher amount and issuance quantity management, so that consumption vouchers can be combined with consumers' actual needs, thereby improving the utilization rate of consumption vouchers and consumer satisfaction.

[0019] Specifically, in the technical solution of the present application, a consumption voucher issuance control system based on big data is proposed. Figure 1 It is a block diagram of a consumption voucher issuance control system based on big data according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a consumption voucher issuance control system based on big data according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, a consumption voucher issuance control system 300 based on big data includes: a user consumption behavior data collection module 310, which is used to obtain a data set of consumption behavior data of the target user after obtaining the authorization of the target user; a user consumption behavior data semantic coding module 320, which is used to perform time sequence regularization and semantic embedding coding on the data set of the consumption behavior data of the target user to obtain a time queue of semantic embedding coding features of the consumption behavior data; a consumption behavior semantic temporal association coding module 330, which is used to perform consumption behavior temporal association coding on the time queue of the semantic embedding coding features of the consumption behavior data to obtain a time queue of semantic temporal association coding features of the consumption behavior data; a consumption behavior semantic temporal information aggregation module 340, which is used to perform dynamic propagation aggregation processing of consumption behavior temporal features on the time queue of the semantic temporal association coding features of the consumption behavior data to obtain a consumption behavior temporal significant propagation aggregation representation; a consumption voucher recommendation module 350, which is used to recommend consumption vouchers based on the consumption behavior temporal significant propagation aggregation representation to obtain a recommended type label of the consumption voucher.

[0020] In particular, the user consumption behavior data collection module 310 is used to obtain a data set of consumption behavior data of the target user after obtaining authorization from the target user. It should be understood that the data set of consumption behavior data includes the user's historical consumption records, consumption frequency, browsing history and search history, etc. By collecting the consumption behavior data of the target user, the preferences and habits of each user can be better understood. By constructing a detailed user portrait based on the consumer's preferences and consumption habits, consumption behavior can be predicted, thereby realizing personalized consumption voucher recommendation and issuance.

[0021] In particular, the user consumption behavior data semantic encoding module 320 is used to perform time-series regularization and semantic embedding encoding on the data set of the consumption behavior data of the target user to obtain a time queue of the semantic embedding encoding features of the consumption behavior data. In an embodiment of the present application, first, the data set of the consumption behavior data of the target user is sorted according to the time dimension to obtain a time queue of the consumption behavior data; here, it is considered that the consumption behavior of the target user has a strong time correlation. The user's purchase decision is often affected by a variety of time factors such as past purchase experience, seasonal factors, and promotional activities. In addition, the user's consumption preferences and behavior patterns are not static, but change over time. Therefore, in order to more clearly observe these time patterns and trends of the target user's consumption behavior, so as to better understand the user's consumption habits and preferences, and identify the user's long-term and short-term behavior changes, in the technical solution of the present application, the data set of the target user's consumption behavior data is further sorted according to the time dimension to obtain a time queue of the consumption behavior data, which is crucial for personalized consumption voucher type recommendation and precision marketing. Next, the consumer behavior semantic embedding coding matrix is ​​used to perform semantic embedding coding on each consumer behavior data in the time queue of the consumer behavior data to obtain a time queue of the consumer behavior data semantic embedding coding vector as the time queue of the consumer behavior data semantic embedding coding features. Here, it is considered that the consumer behavior data usually contains a large amount of unstructured information, such as the name, description, category, etc. of the purchased goods. It is unrealistic to use these text information directly for model training and semantic feature extraction, because machine learning and deep learning models usually require numerical input. Therefore, in the technical solution of the present application, the consumer behavior semantic embedding coding matrix is ​​further used to perform semantic embedding coding on each consumer behavior data in the time queue of the consumer behavior data to obtain a time queue of the consumer behavior data semantic embedding coding vector. Through semantic embedding coding, these unstructured text data can be converted into high-dimensional vectors so that they can be effectively processed by the model. Among them, through the semantic embedding coding method, not only can each consumer behavior data be embedded and mapped into a common space for subsequent processing and analysis, but also the semantic embedding coding feature information in each consumer behavior data can be captured, providing a data basis for the subsequent consumer voucher recommendation task.

[0022] In particular, the consumer behavior semantic temporal association encoding module 330 is used to perform consumer behavior temporal association encoding on the time queue of the semantic embedded coding features of the consumer behavior data to obtain a time queue of the semantic temporal association coding features of the consumer behavior data. Specifically, the time queue of the semantic embedded coding vector of the consumer behavior data is input into the consumer behavior temporal context encoder based on the converter structure to obtain the time queue of the context semantic coding vector of the consumer behavior data as the time queue of the semantic temporal association coding features of the consumer behavior data. Specifically, by inputting the time queue of the semantic embedded coding vector of the consumer behavior data into the consumer behavior temporal context encoder based on the converter structure for encoding, the temporal context long-distance dependency association feature information between the embedded semantics of each consumer behavior of the target user is extracted, thereby obtaining the time queue of the context semantic coding vector of the consumer behavior data, which helps to more accurately understand the consumption pattern of the user and more accurately predict the user's future purchasing behavior.

[0023] In particular, the consumer behavior semantic temporal information aggregation module 340 is used to perform dynamic propagation aggregation processing of consumer behavior temporal features on the time queue of the semantic temporal association coding features of the consumer behavior data to obtain a significant propagation aggregation representation of consumer behavior temporal. It should be understood that each consumer behavior data context semantic coding vector in the time queue of the consumer behavior data context semantic coding vector contains context semantic association feature information about each consumer behavior semantic of the target user in time sequence. Moreover, in the consumer behavior data, the user's consumer behavior semantic features at different time points have different effects on the current consumer behavior. For example, recent purchases may be more influential than purchases a few months ago. However, traditional static methods are difficult to capture this type of dynamic consumer behavior semantic dynamic propagation information, resulting in a decrease in the accuracy of subsequent consumer behavior prediction and consumer voucher type recommendation tasks. Based on this, in the technical solution of the present application, the time queue of the semantic temporal association coding features of the consumer behavior data is further subjected to dynamic propagation aggregation processing of consumer behavior temporal features to obtain a significant propagation aggregation representation of consumer behavior temporal. In particular, the dynamic propagation and aggregation processing of the consumption behavior time series features can better adapt to the time queue distribution changes of the semantic time series associated coding features of the consumption behavior data, wherein the semantic coding vector of the consumption behavior data context of each node (when used as the current node feature) not only represents its own information, but also acts as a feature prompt gate to adjust the information propagation of different consumption behavior data semantic time series associated coding feature information between different nodes in the node feature sequence. In this way, the importance of each node feature can be dynamically adjusted according to the semantic time series context of the current node's consumption behavior during the dynamic propagation of feature information. In this way, the key time series features in the user's consumption data can be captured more accurately, and the prediction accuracy of the model for the target user's consumption behavior can be improved. Specifically, the dynamic propagation and aggregation processing of the consumption behavior time series features enhances the model's perception of the importance of node features. By calculating the feature association strength prompt value and using it as the gating standard, the network can automatically identify the semantic node features of the consumption behavior data context that are more important for the current target user consumption behavior prediction and consumption voucher type recommendation tasks, thereby improving the feature expression ability and subsequent task accuracy.

[0024] Specifically, the dynamic aggregation processing process based on the current node feature includes the following steps: first, after extracting the current consumer behavior data semantic temporal association coding feature from the time queue of the consumer behavior data semantic temporal association coding feature, calculate the feature association strength prompt value of the time series of the consumer behavior data semantic temporal association coding feature relative to the current consumer behavior data semantic temporal association coding feature to obtain a sequence of consumer behavior data context semantic feature association strength prompt values. Specifically, first extract the current consumer behavior data context semantic coding vector from the time queue of the consumer behavior data context semantic coding vector as the current consumer behavior data semantic temporal association coding feature; specifically, extract the current consumer behavior data context semantic coding vector from the time queue of the consumer behavior data context semantic coding vector as the current node feature. This operation is essentially preparing for subsequent feature propagation and updating. In the feature sequence distribution, each node can have different attributes or features. These different node feature vectors (i.e., consumer behavior data context semantic coding vectors) represent the information of each node itself. Extracting the semantic encoding vector of the consumer behavior data context of the current node means locking a specific reference point, and using this node as the center to evaluate the relative importance of other nodes, which is similar to selecting a focus in the feature sequence distribution, and then building a sequence attention perspective around this focus, providing a basis for the subsequent dynamic aggregation process of the semantic temporal association encoding feature of the consumer behavior data. Next, using the current consumer behavior data context semantic encoding vector as the feature prompt gate, the feature association strength prompt value of each consumer behavior data context semantic encoding vector in the time queue of the consumer behavior data context semantic encoding vector relative to the current consumer behavior data context semantic encoding vector is calculated to obtain a sequence of consumer behavior data context semantic feature association strength prompt values. This step is intended to measure the relationship strength between nodes and provide a basis for determining which node features should be propagated more strongly in the next step. It can essentially be regarded as an application of the attention mechanism, in which the attention score is determined by the feature similarity between nodes. In this way, neighbor nodes that have a greater impact on the current node (i.e., the current consumer behavior data semantic temporal association encoding feature) can be identified, and a higher weight can be given to subsequent feature propagation.

[0025] Among them, taking the current consumer behavior data context semantic coding vector as the feature prompt gate, the specific process of calculating the feature association strength prompt value of each consumer behavior data context semantic coding vector in the time queue of the consumer behavior data context semantic coding vector relative to the current consumer behavior data context semantic coding vector includes: calculating the positional division between the consumer behavior data context semantic coding vector and the current consumer behavior data context semantic coding vector to obtain the consumer behavior data context semantic association strength contribution vector; calculating the logarithmic function value with base two of the absolute value of each eigenvalue in the consumer behavior data context semantic association strength contribution vector to obtain the consumer behavior data context semantic association strength contribution logarithmic representation vector; calculating the positional dot product between the consumer behavior data context semantic coding vector and the consumer behavior data context semantic association strength contribution logarithmic representation vector, and then summing the obtained dot product vector position by position to obtain the consumer behavior data context semantic association strength contribution value; calculating the natural exponential function value with the natural constant e as the base and the consumer behavior data context semantic association strength contribution value as the exponential power to obtain the consumer behavior data context semantic feature association strength prompt value.

[0026] Furthermore, based on the sequence of the prompt values ​​of the semantic feature association strength of the consumer behavior data context, the time queue of the semantic temporal association coding feature of the consumer behavior data is subjected to dynamic aggregation analysis of feature information to obtain an aggregate representation of the significant propagation of the consumer behavior temporal sequence. Specifically, the median of the sequence of prompt values ​​of the semantic feature association strength of the consumer behavior data context is first extracted as the threshold of the semantic feature association strength of the consumer behavior data context; one reason why the median of the prompt value sequence is used as the threshold value instead of the average is that the median is insensitive to extreme values ​​and can better reflect the central trend of the distribution of the semantic temporal association coding feature of the consumer behavior data. This approach helps to maintain the stability of the propagation of the semantic temporal association coding feature of the consumer behavior data in the network and avoid overall performance degradation due to a few extreme cases. Next, the consumer behavior data context semantic feature association strength gate threshold is used as the switch threshold of the gate control unit, and the time queue of the consumer behavior data context semantic coding vector and the sequence of the consumer behavior data context semantic feature association strength prompt value are input into the gate control unit to obtain the time queue of the consumer behavior data context semantic coding vector after intensity modulation; specifically, the previously extracted consumer behavior data context semantic coding vector time queue and the corresponding consumer behavior data context semantic feature association strength prompt value sequence are input into the gate control unit, and the previously determined gate threshold is used as the switch threshold of the gate control unit to achieve intensity modulation of the consumer behavior data context semantic coding vector. This process controls the propagation of the consumer behavior data semantic temporal association coding feature information at different nodes by adjusting the weights, so that those features with higher association strength and more relevant to the subsequent target user consumer behavior prediction and consumer voucher type recommendation tasks are more likely to be retained and passed to the subsequent layers. In this way, those irrelevant or redundant information to the subsequent tasks can be ignored, while those features that are critical to specific tasks can be retained. Finally, the time queue of the intensity modulated consumer behavior data context semantic encoding vector is input into the feature dynamic propagation module based on the forward LTSM model to obtain the consumer behavior temporal significant propagation aggregation feature vector as the consumer behavior temporal significant propagation aggregation representation. In particular, LSTM is a special recurrent neural network (RNN) that is particularly good at dealing with long-term dependency problems. In this context, LSTM is used to capture the dependency between the semantic temporal association encoding features of consumer behavior data at different nodes in the feature propagation process and generate a comprehensive feature vector, which not only reflects the semantic temporal association encoding features of consumer behavior data at the current node, but also includes the temporal influence of the context, providing a more accurate basis for subsequent tasks.

[0027] Among them, the specific process of using the consumer behavior data context semantic feature association strength gate threshold as the switching threshold of the gating unit and inputting the time queue of the consumer behavior data context semantic coding vector and the sequence of the consumer behavior data context semantic feature association strength prompt value into the gating unit includes: in response to the consumer behavior data context semantic feature association strength prompt value in the sequence of the consumer behavior data context semantic feature association strength prompt value being greater than or equal to the consumer behavior data context semantic feature association strength gate threshold, dividing the consumer behavior data context semantic feature association strength prompt value by the consumer behavior data context semantic feature association strength gate threshold and then multiplying it by position with the corresponding consumer behavior data context semantic coding vector; otherwise, setting it to zero to obtain the time queue of the consumer behavior data context semantic coding vector after the intensity modulation.

[0028] In summary, in the above embodiment, the time queue of the semantic temporal association coding feature of the consumer behavior data is subjected to dynamic propagation aggregation processing of the consumer behavior temporal feature to obtain a significant propagation aggregation representation of the consumer behavior temporal sequence, including: performing dynamic propagation aggregation processing of the consumer behavior temporal feature on the time queue of the semantic temporal association coding feature of the consumer behavior data using the following dynamic aggregation formula to obtain a significant propagation aggregation representation of the consumer behavior temporal sequence; wherein the dynamic aggregation formula is: in, is the time queue of the context semantic encoding vector of the consumption behavior data, are the first, second, and third in the time queue of the context semantic encoding vector of the consumer behavior data. and The contextual semantic encoding vector of consumer behavior data, For the The first one in the context semantic encoding vector of the consumer behavior data Position feature values, is the length of the contextual semantic encoding vector of consumer behavior data, For the The first one in the context semantic encoding vector of the consumer behavior data Position feature values, represents the logarithmic function value with base 2, represents the value of the natural exponential function with the natural constant e as the base, The first value in the sequence of the association strength prompt value of the contextual semantic feature of the consumer behavior data The contextual semantic feature association strength prompt value of consumer behavior data, Indicates extracting the median, is the threshold of the correlation strength of the contextual semantic features of consumer behavior data, represents the masking operation, It is the time queue of the context semantic encoding vector of the consumption behavior data after intensity modulation. are the first, second, and third time queues of the context semantic encoding vectors of the intensity modulated consumer behavior data. and The context semantic encoding vector of consumption behavior data after intensity modulation, is a sequence encoding operation based on the forward LTSM model, Aggregate feature vectors for significant propagation of consumption behavior time series.

[0029] In particular, the consumer voucher recommendation module 350 is used to recommend consumer vouchers based on the significant propagation aggregation representation of the consumer behavior time series to obtain a recommended type label for the consumer voucher. In a specific example of the present application, the significant propagation aggregation feature vector of the consumer behavior time series is input into a classifier-based consumer voucher intelligent recommender to obtain a recommendation result, and the recommendation result is used to represent the recommended type label of the consumer voucher. Specifically, the semantic temporal dynamic aggregation feature information of the target user's consumer behavior is used for classification processing to recommend an adapted consumer voucher type for the target user. In this way, a detailed user portrait can be constructed based on the consumer's preferences and consumption habits in order to predict consumption behavior, thereby realizing personalized consumer voucher recommendations and issuance, so that the consumer vouchers can be adapted to the actual needs of consumers, and the utilization rate of consumer vouchers and consumer satisfaction can be improved.

[0030] In the technical solution of the present application, each consumer behavior data context semantic coding vector in the time queue of the consumer behavior data context semantic coding vector respectively represents the global context semantic coding features of the consumer behavior data at each timestamp in the data set of the target user's consumer behavior data with the data set as the background corpus. In this way, when the time queue of the context semantic encoding vector of the consumer behavior data is input into the feature dynamic propagation network with the current node as the feature prompt gate, it uses the current node as the feature prompt gate to perform significant temporal dimension aggregation of the time queue of the context semantic encoding vector of the consumer behavior data. However, considering that the feature dynamic propagation network with the current node as the feature prompt gate uses the current node as the prior anchor by default to perform temporal transfer aggregation significance measurement, the setting of the prior anchor will naturally introduce some interference components, which will cause the overall feature manifold representation of the consumer behavior temporal significant propagation aggregation feature vector obtained by feature dynamic propagation aggregation in the high-dimensional feature space to have fine-grained feature structural holes. The existence of fine-grained feature structural holes will not only make the features of the consumer behavior temporal significant propagation aggregation feature vector corresponding to the class probability label insufficient in semantic coverage, but will also cause the outlier class regression inference mapping to miss the target, affecting the accuracy of the recommendation results obtained by the consumer behavior temporal significant propagation aggregation feature vector input into the classifier-based consumer voucher intelligent recommender.

[0031] In the technical solution of the present application, the consumption behavior time series significant propagation aggregate feature vector is input into a consumption voucher intelligent recommender based on a classifier to obtain a recommendation result, including: Inputting the consumption behavior time series significant propagation aggregation feature vector into a pre-classifier based on a Softmax function to obtain a consumption behavior time series fine-grained class probability label vector; Calculate the product between the consumer behavior time series fine-grained class probability label vector and its transposed vector to obtain the consumer behavior time series fine-grained full label domain modulation matrix; Divide the eigenvalues ​​of each position in the fine-grained full-label domain modulation matrix of the consumer behavior time series by the square root of the scale value of the significant propagation aggregated eigenvector of the consumer behavior time series to obtain the fine-grained full-label domain scale modulation matrix of the consumer behavior time series; Inputting the consumption behavior temporal fine-grained full-label domain scale modulation matrix into a quasi-probability domain attention sparse module based on a multi-level gating function to obtain the consumption behavior temporal fine-grained full-label domain scale modulation matrix; Taking the consumption behavior time series significant propagation aggregate feature vector as the query feature vector, matrix multiplying the consumption behavior time series fine-grained full label domain scale modulation matrix and the consumption behavior time series significant propagation aggregate feature vector to obtain an optimized consumption behavior time series significant propagation aggregate feature vector; The optimized consumption behavior time series significant propagation aggregate feature vector is input into the classifier-based consumption voucher intelligent recommender to obtain the recommendation result.

[0032] The optimization process of the consumption behavior time series significant propagation aggregation feature vector is expressed as follows: in, is the significant propagation aggregation feature vector of the consumption behavior time series, is the first significant propagation aggregation feature vector of the consumption behavior time series eigenvalues, is the fine-grained class probability label vector of the consumption behavior time series, is the scale of the significant propagation aggregation feature vector of the consumption behavior time series, is the consumption behavior temporal fine-grained full-label domain scale modulation matrix, is the consumption behavior temporal fine-grained full-label domain scale modulation matrix, represents matrix multiplication, Represents the consumption behavior temporal fine-grained full-label domain scale modulation weight, is a predetermined threshold, For masking, is the optimized consumption behavior time series significant propagation aggregate feature vector.

[0033] Therefore, by adding class probability domain-level optimizable disturbances to the original feature vector to strengthen the dependency entanglement between different variables in the original feature vector based on label domain modulation, the source domain feature vector can be optimized based on class probability query scale modulation. In this way, the significant inherent modal information in the source domain features can be more effectively retained, and the interference components in the feature distribution can be effectively masked, thereby improving the adversarial robustness of the manifold expression of the feature vector. In this way, the accuracy of the recommendation results obtained by the consumption voucher intelligent recommender based on the classifier is improved, so that personalized consumption voucher recommendations and issuance can be realized based on consumer preferences and consumption habits, so that consumption vouchers can be adapted to the actual needs of consumers, and the utilization rate of consumption vouchers and consumer satisfaction can be improved.

[0034] As described above, the consumption voucher issuance control system 300 based on big data according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a consumption voucher issuance control algorithm based on big data. In a possible implementation, the consumption voucher issuance control system 300 based on big data according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the consumption voucher issuance control system 300 based on big data can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the consumption voucher issuance control system 300 based on big data can also be one of the many hardware modules of the wireless terminal.

[0035] Alternatively, in another example, the big data-based consumption voucher issuance control system 300 and the wireless terminal may also be separate devices, and the big data-based consumption voucher issuance control system 300 may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0036] Furthermore, a consumption voucher issuance control method based on big data is also provided.

[0037] Figure 4 Flow chart of a consumption voucher issuance control method based on big data according to an embodiment of the present application. Figure 4 As shown, according to the big data-based consumption voucher issuance control method of the embodiment of the present application, the steps include: S1, after obtaining the authorization of the target user, obtaining a data set of consumption behavior data of the target user; S2, performing time regularization and semantic embedding coding on the data set of consumption behavior data of the target user to obtain a time queue of semantic embedding coding features of the consumption behavior data; S3, performing consumption behavior time series association coding on the time queue of semantic embedding coding features of the consumption behavior data to obtain a time queue of semantic time series association coding features of the consumption behavior data; S4, performing dynamic propagation aggregation processing of consumption behavior time series features on the time queue of semantic time series association coding features of the consumption behavior data to obtain a consumption behavior time series significant propagation aggregation representation; S5, recommending consumption vouchers based on the consumption behavior time series significant propagation aggregation representation to obtain a recommended type label for the consumption vouchers.

[0038] In summary, the big data-based consumption voucher issuance control method according to the embodiment of the present application is explained. After obtaining the authorization of the target user, the time series information of the consumption behavior data of the target user is collected, and the data processing and semantic understanding algorithms based on artificial intelligence and natural language analysis are introduced in the back end to analyze the time series information of the consumption behavior data of these target users, so as to capture the characteristic dynamic aggregation information of the semantics of the consumption behavior data of the target user in the time series, so as to recommend the type of consumption vouchers for the target user. In this way, a detailed user portrait can be constructed based on the consumer's preferences and consumption habits in order to predict consumption behavior, thereby realizing personalized consumption voucher recommendation and issuance.

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

Claims

1. A consumption voucher issuance control system based on big data, characterized in that: include: A user consumption behavior data collection module is used to obtain a data set of consumption behavior data of a target user after obtaining authorization from the target user; A user consumption behavior data semantic coding module, used to perform time sequence regularization and semantic embedding coding on the target user's consumption behavior data set to obtain a time queue of the consumption behavior data semantic embedding coding features; A consumption behavior semantic temporal association coding module is used to perform consumption behavior temporal association coding on the time queue of the semantic embedding coding feature of the consumption behavior data to obtain a time queue of the semantic temporal association coding feature of the consumption behavior data; A consumer behavior semantic temporal information aggregation module is used to perform dynamic propagation aggregation processing of consumer behavior temporal features on the time queue of the semantic temporal association coding features of the consumer behavior data to obtain a significant propagation aggregation representation of the consumer behavior temporal sequence; wherein the consumer behavior semantic temporal information aggregation module includes: a semantic feature association strength extraction unit, which is used to extract the current consumer behavior data semantic temporal association coding features from the time queue of the consumer behavior data semantic temporal association coding features, and then calculate the feature association strength prompt value of the time series of the consumer behavior data semantic temporal association coding features relative to the current consumer behavior data semantic temporal association coding features to obtain a sequence of consumer behavior data context semantic feature association strength prompt values; a semantic feature information dynamic aggregation analysis unit, which is used to perform feature information dynamic aggregation analysis on the time queue of the consumer behavior data semantic temporal association coding features based on the sequence of consumer behavior data context semantic feature association strength prompt values ​​to obtain the significant propagation aggregation representation of the consumer behavior temporal sequence; The consumption voucher recommendation module is used to recommend consumption vouchers based on the consumption behavior time series significant propagation aggregation representation to obtain a recommended type label of the consumption voucher.

2. The consumption voucher issuance control system based on big data according to claim 1 is characterized in that: The user consumption behavior data semantic encoding module is used to: Arrange the data set of the consumption behavior data of the target user according to the time dimension to obtain a time queue of the consumption behavior data; The consumer behavior data in the time queue of the consumer behavior data are semantically embedded and encoded using a consumer behavior semantic embedding coding matrix to obtain a time queue of consumer behavior data semantic embedding coding vectors as a time queue of consumer behavior data semantic embedding coding features.

3. The consumption voucher issuance control system based on big data according to claim 2 is characterized in that: The consumer behavior semantic temporal association encoding module is used to: input the time queue of the consumer behavior data semantic embedding encoding vector into a consumer behavior temporal context encoder based on a converter structure to obtain the time queue of consumer behavior data context semantic encoding vector as the time queue of the consumer behavior data semantic temporal association encoding feature.

4. The consumption voucher issuance control system based on big data according to claim 3 is characterized in that: The semantic feature association strength extraction unit comprises: A current consumer behavior data semantic temporal association coding feature extraction subunit is used to extract the current consumer behavior data context semantic coding vector from the time queue of the consumer behavior data context semantic coding vector as the current consumer behavior data semantic temporal association coding feature; The feature association strength prompt value calculation subunit is used to use the current consumer behavior data context semantic coding vector as the feature prompt gate, calculate the feature association strength prompt value of each consumer behavior data context semantic coding vector in the time queue of the consumer behavior data context semantic coding vector relative to the current consumer behavior data context semantic coding vector to obtain a sequence of consumer behavior data context semantic feature association strength prompt values.

5. The consumption voucher issuance control system based on big data according to claim 4 is characterized in that: The feature association strength prompt value calculation subunit is used to: Calculating the positional division between the consumer behavior data context semantic encoding vector and the current consumer behavior data context semantic encoding vector to obtain a consumer behavior data context semantic association strength contribution vector; Calculating the logarithmic function value with base 2 of the absolute value of each eigenvalue in the consumer behavior data context semantic association strength contribution vector to obtain a consumer behavior data context semantic association strength contribution logarithmic representation vector; After calculating the position-by-position dot product between the consumer behavior data context semantic encoding vector and the consumer behavior data context semantic association strength contribution logarithmic representation vector, the obtained dot product vectors are summed position by position to obtain the consumer behavior data context semantic association strength contribution value; The natural exponential function value with the natural constant e as the base and the consumption behavior data context semantic association strength contribution value as the exponential power is calculated to obtain the consumption behavior data context semantic feature association strength prompt value.

6. The consumption voucher issuance control system based on big data according to claim 5 is characterized in that: The semantic feature information dynamic aggregation analysis unit includes: A median extraction subunit, used to extract the median of the sequence of the consumer behavior data context semantic feature association strength prompt value as the consumer behavior data context semantic feature association strength gate threshold; A gated modulation subunit, used to use the consumer behavior data context semantic feature association strength gate threshold as the switch threshold of the gated unit, and input the time queue of the consumer behavior data context semantic encoding vector and the sequence of the consumer behavior data context semantic feature association strength prompt value into the gated unit to obtain the time queue of the consumer behavior data context semantic encoding vector after intensity modulation; The feature dynamic propagation subunit is used to input the time queue of the intensity modulated consumer behavior data context semantic encoding vector into the feature dynamic propagation module based on the forward LTSM model to obtain the consumer behavior temporal significant propagation aggregation feature vector as the consumer behavior temporal significant propagation aggregation representation.

7. The consumption voucher issuance control system based on big data according to claim 6 is characterized in that: The gate modulation subunit is used for: In response to a consumer behavior data context semantic feature association strength prompt value in the sequence of consumer behavior data context semantic feature association strength prompt values ​​being greater than or equal to the consumer behavior data context semantic feature association strength gate threshold, the consumer behavior data context semantic feature association strength prompt value is divided by the consumer behavior data context semantic feature association strength gate threshold and then multiplied by the corresponding consumer behavior data context semantic encoding vector by position; otherwise, it is set to zero to obtain the time queue of the consumer behavior data context semantic encoding vector after intensity modulation.

8. The consumption voucher issuance control system based on big data according to claim 7 is characterized in that: The consumption voucher recommendation module is used to: input the consumption behavior time series significant propagation aggregation feature vector into the classifier-based consumption voucher intelligent recommender to obtain a recommendation result, and the recommendation result is used to represent the recommendation type label of the consumption voucher.

9. A consumption voucher issuance control method based on big data, characterized in that: include: After obtaining authorization from the target user, obtain a data set of consumption behavior data of the target user; Performing time sequence regularization and semantic embedding coding on the data set of the consumption behavior data of the target user to obtain a time queue of semantic embedding coding features of the consumption behavior data; Performing consumer behavior temporal association coding on the time queue of the semantic embedding coding feature of the consumer behavior data to obtain a time queue of the semantic temporal association coding feature of the consumer behavior data; Performing dynamic propagation aggregation processing of consumer behavior temporal features on the time queue of semantic temporal correlation coding features of the consumer behavior data to obtain a significant propagation aggregation representation of consumer behavior temporal features; Based on the time series significant propagation aggregation representation of the consumption behavior, consumption voucher recommendations are made to obtain a recommended type label of the consumption voucher.

Images