CN107590690B

CN107590690B - Data processing method and device and server

Info

Publication number: CN107590690B
Application number: CN201710790349.3A
Authority: CN
Inventors: 肖锦文; 赵嘉寅; 周琳
Original assignee: Advanced New Technologies Co Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2017-09-05
Filing date: 2017-09-05
Publication date: 2021-02-23
Anticipated expiration: 2037-09-05
Also published as: CN107590690A

Abstract

The embodiment of the specification provides a data processing method, a data processing device and a server. The method comprises the following steps: acquiring original historical transaction data, and determining feature vectors of transaction shops and transaction users in the historical transaction data; taking the feature vectors of a first preset number of trading users corresponding to a target trading shop as the trading feature vectors of the target trading shop; taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain corresponding transaction embedded characteristic data of the trading stores and the trading users.

Description

Data processing method and device and server

Technical Field

The embodiment of the specification relates to the technical field of data optimization processing, in particular to a data processing method, a data processing device and a server.

Background

With the development of the internet consumption era, more and more people can purchase commodities on some electronic commerce platforms, and in order to improve the commodity purchase rate, the association degree between a user and a merchant can be inferred based on historical transaction data in the electronic commerce platform, so that the user with high association degree is recommended to the merchant, and the merchant can carry out commodity marketing and user maintenance in a targeted manner; or recommend highly associated merchants to the user.

In the prior art, the association degree between a merchant and a user needs to be determined, and a large number of characteristics such as the transaction amount of the last 1 day, the transaction amount of the last week, the monthly transaction frequency and the like can be set and constructed manually according to original historical transaction data; then, the association degree between the merchant and the user is measured according to the comparison between the features, for example, the transaction amount of the user a in the last week of the merchant X is 280 yuan, the transaction amount of the user B in the last week of the merchant X is 80 yuan, and accordingly, according to the feature (the transaction amount in the last week), it can be determined that the association degree between the user a and the merchant X is greater than the association degree between the user B and the merchant X. However, the features used for determining the association degree between the user and the merchant in the prior art need to be set and constructed manually, so that invalid features are easy to occur, and the problem of low processing efficiency often exists for a large amount of historical transaction data. Therefore, there is a need to provide faster or more efficient solutions.

Disclosure of Invention

The embodiment of the specification aims to provide a data processing method, a data processing device and a data processing server, which can quickly and effectively obtain transaction embedded characteristic data representing transaction characteristics and improve data processing efficiency.

The embodiment of the specification is realized by the following steps:

a method of data processing, comprising:

acquiring historical transaction data, and determining feature vectors of transaction shops and transaction users in the historical transaction data;

taking the feature vectors of a first preset number of trading users corresponding to a target trading shop as the trading feature vectors of the target trading shop;

taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user;

and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain corresponding transaction embedded characteristic data of the trading stores and the trading users.

A data processing apparatus comprising:

the historical transaction data acquisition module is used for acquiring historical transaction data;

the characteristic vector determining module is used for determining the characteristic vectors of the trading shops and the trading users in the historical trading data;

the system comprises a first trading feature vector determining module, a second trading feature vector determining module and a third trading feature vector determining module, wherein the first trading feature vector determining module is used for taking feature vectors of a first preset number of trading users corresponding to a target trading store as trading feature vectors of the target trading store;

the second trading feature vector determining module is used for taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user;

and the characterization learning module is used for performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain corresponding transaction embedded characteristic data of the trading stores and the trading users.

A data processing server comprising a processor and a memory, the memory storing computer program instructions for execution by the processor, the computer program instructions comprising:

As can be seen from the above, one or more embodiments of the present specification provide for the generation of a transaction by obtaining raw historical transaction data; then, directly taking the feature vectors of a first preset number of trading users corresponding to the target trading shop as the trading feature vectors of the target trading shop; taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vectors of the target trading stores and the transaction characteristic vectors of the target trading users, and adding the transaction characteristics into the corresponding characteristic vectors of the trading stores and the trading users to further obtain transaction embedded characteristic data capable of characterizing the transaction characteristics. The embodiment of the specification directly utilizes the original transaction data to learn the transaction embedded characteristic data capable of representing the transaction characteristics, so that the learned transaction embedded characteristic data can be effectively ensured to directly represent the transaction characteristics, and the data processing efficiency can be improved.

Drawings

In order to more clearly illustrate one or more embodiments or prior art solutions of the present disclosure, the drawings used in the description of the embodiments or prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the present disclosure, and other drawings can be obtained by those skilled in the art without inventive efforts.

FIG. 1 is a schematic flow chart diagram illustrating one embodiment of a data processing method provided herein;

fig. 2 is a schematic flow chart of an embodiment of performing transaction characterization learning on a trading store and a trading user by using a transaction feature vector of the target trading store and a transaction feature vector of the target trading user to obtain transaction embedded feature data of the corresponding trading store and the trading user, which is provided by the present specification;

FIG. 3 is a flow diagram illustrating one embodiment of determining hidden relationship feature data provided herein;

FIG. 4 is a diagram illustrating an embodiment of calculating the association degree between a user and a merchant based on data obtained by the data processing method provided in the present specification;

fig. 5 is a schematic structural diagram of an embodiment of a data processing apparatus provided in this specification.

Detailed Description

The embodiment of the specification provides a data processing method, a data processing device and a server.

In order to make those skilled in the art better understand the technical solutions in the present specification, the technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is obvious that the described embodiments are only a part of the embodiments of the present specification, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without any creative effort shall fall within the protection scope of the present specification.

Currently, on some electronic commerce platforms, merchants often need to perform commodity marketing, user maintenance and the like to improve the purchase rate of commodities. The historical transaction data may be used to reflect the user's preference of the store as a record of the transaction behavior generated when the user makes a purchase at the store. Therefore, it is often the case that the correlation between the user and the shop is determined by extracting the correlation features from the historical transaction data, and the purchase rate of the product is improved by performing the correlation recommendation processing. In the existing process of manually setting and constructing a large number of characteristics according to original historical transaction data, the manually constructed characteristics often have certain subjectivity and are easy to have invalid characteristics, so that the extracted characteristics cannot effectively represent the transaction characteristics between a user and a shop (the transaction characteristics can include characteristics reflecting the transaction behaviors of the user). Based on the data, the characterization capability of the characteristic data used for calculating the association degree between the user and the shop on the transaction characteristics can be improved.

A specific embodiment of the data processing method of the present specification is described below. FIG. 1 is a schematic flow chart diagram of one embodiment of a data processing method provided herein, which provides method steps as described in the embodiments or flowcharts, but may include more or fewer steps based on routine or non-inventive labor. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. In actual implementation, the system or client product may execute sequentially or in parallel (e.g., in the context of parallel processors or multi-threaded processing) according to the embodiments or methods shown in the figures. Specifically, as shown in fig. 1, the method may include:

s102: acquiring historical transaction data, and determining the characteristic vectors of the transaction shops and the transaction users in the historical transaction data.

In the embodiment of the present specification, a large amount of historical transaction data may be acquired, and feature vectors of each trading store and each trading user corresponding to the historical transaction data may be determined.

Specifically, the historical transaction data may include a user and a store corresponding to the transaction. The feature vector of the trading store may include a random feature vector of a preset dimension, and a value range of each feature value of the random feature vector is 0-1. Specifically, for example, the preset dimension is three-dimensional, and the search random feature vector may be (0.6, 0.4, 0.5). The feature vector of the trading user may include a random feature vector of a preset dimension, and the sum of feature values of the preset dimension of the random feature vector is 1. The feature vector of the trading store has the same dimension as the feature vector of the trading user.

S104: and taking the feature vectors of the trading users of the first preset number corresponding to the target trading shop as the trading feature vectors of the target trading shop.

In this embodiment of the present specification, a feature vector of a first preset number of trading users corresponding to a target trading store may be used as a trading feature vector of the target trading store. In particular, in some embodiments, the historical transaction data may also include transaction time. Correspondingly, the taking the feature vectors of the trading users of the first preset number corresponding to the target trading store as the trading feature vectors of the target trading store may include:

selecting feature vectors of a first preset number of trading users corresponding to the target trading shop according to trading time in trading data corresponding to the target trading shop;

and taking the feature vectors of the trading users of the first preset number as the trading feature vectors of the target trading stores.

In a specific embodiment, the selecting the feature vectors of the trading users of the first preset number corresponding to the target trading store according to the trading time may include selecting the feature vectors of the trading users of the first preset number according to the trading time sequence, continuously or at regular intervals; the method can further comprise the step of selecting the feature vectors of the trading users of the first preset number in a certain time period according to the trading time. Specifically, theThe first preset number may be a number preset according to actual requirements, for example, 5, and correspondingly, the trading feature vector of the target trading store may be (U)₁，U₂，U₃，U₄，U₅) Wherein, U₁，U₂，U₃，U₄，U₅And the feature vectors are the feature vectors of 5 trading users corresponding to the target trading shops.

S106: and taking the feature vectors of the trading stores of the second preset number corresponding to the target trading user as the trading feature vectors of the target trading user.

In this embodiment of the present specification, the feature vectors of the trading stores of the second preset number corresponding to the target trading user may be used as the trading feature vectors of the target trading user. In particular, in some embodiments, the historical transaction data may also include transaction time. Correspondingly, the taking the feature vector of the trading stores of the second preset number corresponding to the target trading user as the trading feature vector of the target trading user may include:

selecting feature vectors of a second preset number of trading shops corresponding to the target trading user according to the trading time in the trading data corresponding to the target trading user;

and taking the feature vectors of the trading stores with the second preset number as the trading feature vectors of the target trading users.

In a specific embodiment, the selecting the feature vectors of the trading stores of the second preset number corresponding to the target trading user according to the trading time may include selecting the feature vectors of the trading stores of the second preset number continuously or at regular intervals according to the trading time; the method can further comprise the step of selecting the feature vectors of a second preset number of trading shops in a certain time period according to the trading time. Specifically, the second preset number may be a number set in advance according to actual requirements, and the second preset number may be different from the first preset number, and preferably, the second preset number may be the same as the first preset number. Specifically, the method comprises the following steps. Such as aThe second preset number is set to 5, and correspondingly, the transaction feature vector of the target transaction user may be (S)₁，S₂，S₃，S₄，S₅) Wherein S is₁，S₂，S₃，S₄，S₅And the feature vectors are the feature vectors of 5 trading shops corresponding to the target trading user.

S108: and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain corresponding transaction embedded characteristic data of the trading stores and the trading users.

In the embodiment of the description, the transaction characteristic vector of the target transaction shop and the transaction characteristic vector of the target transaction user can be utilized to perform transaction characterization learning of the transaction shop and the transaction user, so that the corresponding transaction embedded characteristic data of the transaction shop and the transaction user can be obtained. As shown in fig. 2, fig. 2 is a schematic flow chart of an embodiment of performing transaction characterization learning on a trading store and a trading user by using a transaction feature vector of the target trading store and a transaction feature vector of the target trading user to obtain transaction embedded feature data of the corresponding trading store and the trading user, which may specifically include:

s202: and respectively preprocessing the transaction characteristic vector of the target transaction shop and the transaction characteristic vector of the target transaction user to obtain the transaction characteristic vectors of the target transaction shop and the target transaction user.

Specifically, in the embodiment of the present specification, since the transaction feature vector includes one or more feature vectors, in order to ensure that the transaction feature vector and the corresponding feature vector are feature vectors with the same dimension, so as to facilitate subsequent calculation, the transaction feature vector may be preprocessed. Specifically, the pretreatment may include any one of:

average value calculation processing, summation calculation processing, maximum value taking processing and weighted average value calculation processing.

Taking the average value calculation process as an example, the average value calculation process may be performed on a plurality of feature vectors corresponding to the transaction feature vectors, and then the obtained average value vector is used as the corresponding transaction characterization vector. In addition, in the case where the transaction feature vector corresponds to one feature vector, the average value of one feature vector is calculated and processed to obtain the feature vector.

S204: and normalizing the transaction characterization vector of the target trading store and the feature vector of the target trading store, and calculating the difference between the normalized transaction characterization vector and the feature vector of the target trading store based on a preset loss function.

The embodiment of the specification can perform normalization processing on the transaction representation vector of the target trading store and the feature vector of the target trading store, so that the two vectors are guaranteed to be vectors with the same dimension and the same magnitude when the loss function is calculated. Specifically, the normalization process may include a mode of using a softmax flexible maximum transfer function, but the embodiments of the present specification are not limited thereto. The sum of the corresponding normalized transaction characterization vector and the characteristic value of the characteristic vector is 1, and the value range of each characteristic value is 0-1. For example a three-dimensional vector (0.1, 0.3, 0.6).

After the normalization processing, the difference degree between the transaction characterization vector and the feature vector after the target trading shop normalization can be calculated based on a preset loss function. The value of the corresponding preset loss function is taken as the degree of difference. Specifically, the preset loss function in the embodiments of the present disclosure may include at least one of:

relative entropy function, 0-1 loss function, logarithmic loss function, exponential loss function, negative cross entropy and mean square error function.

Taking the relative entropy function as an example, H (S, Z) ═ Σ (S · log (S/Z)), where H (S, Z) represents the relative entropy between the transaction characterization vector and the feature vector after normalization of the target trading store; s represents a feature vector after the target trading shop is normalized; and (4) carrying out normalized transaction characterization vectors on the Z target transaction shops. The corresponding calculated relative entropy value can be used as the difference degree between the normalized transaction characterization vector and the feature vector of the target trading store.

As can be seen from the above, when the value corresponding to the preset loss function is larger, the difference degree between the corresponding vectors is larger; conversely, when the value corresponding to the predetermined loss function is smaller, the difference between the corresponding vectors is smaller.

S206: and adjusting the feature vectors of a first preset number of trading users corresponding to the target trading shop to change the trading feature vectors of the target trading shop, and repeating the steps from the preprocessing of the trading feature vectors of the target trading shop to the calculation of the difference degree until the difference degree between the normalized trading feature vectors and the feature vectors of the target trading shop meets a first preset condition.

In order to reduce the difference between the normalized transaction representation vector and the feature vector of the target trading store, the feature vectors of a first preset number of trading users corresponding to the target trading store may be adjusted to change the trading feature vector of the target trading store, and then the steps of preprocessing the trading feature vector of the target trading store in S202 and S204 are repeated until the difference between the normalized transaction representation vector and the feature vector of the target trading store meets a first preset condition. The first preset condition may include that a difference between the current difference and the difference obtained by the last calculation is less than or equal to a first preset value, or a number of times that a difference between adjacent differences is less than or equal to the first preset value is greater than or equal to a first preset number. Specifically, the first preset value and the first preset times may be preset in combination with actual application requirements.

When the difference degree between the normalized transaction representation vector of the target trading store and the feature vector meets a first preset condition, the difference degree between the normalized transaction representation vector of the target trading store and the current feature vector is small and tends to be stable.

S208: and taking the feature vector of the trading user corresponding to the target trading shop when the first preset condition is met as the feature vector of the target trading user, carrying out normalization processing on the feature vector of the target trading user and the trading characterization vector, and calculating the difference between the feature vector and the trading characterization vector after the target trading user is normalized based on a preset loss function.

Specifically, the normalization process and the calculation based on the preset loss function may refer to the above related steps, which are not described herein again.

S210: and adjusting the feature vectors of a second preset number of trading shops corresponding to the target trading user to change the trading feature vectors of the target trading user, and repeating the steps from the preprocessing of the trading feature vectors of the target trading user to the calculation of the difference degree until the difference degree between the normalized trading feature vectors and the feature vectors of the target trading user meets a second preset condition.

Specifically, the second preset condition may include that a difference between the current difference and the difference obtained by the last calculation is less than or equal to a second preset value, or a number of times that a difference between adjacent differences is less than or equal to the second preset value is greater than or equal to the second preset number. Specifically, the second preset value and the second preset times may be preset in combination with actual application requirements. The second preset value can be the same as or different from the first preset value; the second preset number may be the same as or different from the first preset number.

When the difference degree between the transaction representation vector and the feature vector after the target transaction user is normalized meets a second preset condition, the difference degree between the transaction representation vector after the target transaction user is normalized and the current feature vector is small and tends to be stable.

S212: and taking the feature vector of the trading shop corresponding to the target trading user when the feature vector meets the second preset condition as the feature vector of the target trading shop, and repeating the steps from the preprocessing to the calculation of the difference degree until the difference degree between the normalized trading representation vector of the target trading shop and the feature vector meets the first preset condition and the difference degree between the normalized trading representation vector of the target trading user and the feature vector meets the second preset condition.

S214: and respectively taking the feature vectors of the trading users and the trading stores as the trading embedded feature data of the corresponding trading users and the trading stores when the difference between the normalized trading representation vector and the feature vector of the target trading store meets a first preset condition and the difference between the normalized trading representation vector and the feature vector of the target trading user meets a second preset condition.

In the embodiment of the description, the feature vectors of the trading users and the trading stores are continuously and alternately adjusted, and the difference between the corresponding vectors calculated based on the preset loss function is repeated, so that the trading features can be added to the feature vectors of the corresponding trading users and the trading stores in the process of reducing the calculated value of the preset loss function, and further, the trading embedded feature data capable of representing the trading features is obtained.

In addition, it should be noted that the step sequence recited in the embodiments of the present specification is only one manner of execution sequence of many steps, and does not represent the only execution sequence. For example, in the above steps S204 and S206, the feature vectors of the transaction characterization vectors of the target trading stores are normalized, the corresponding difference degree is calculated, and the feature vectors are adjusted; then, the feature vectors of the transaction characterization vectors of the target transaction users are normalized, and corresponding difference degree calculation and feature vector adjustment processing are carried out. In other embodiments, the feature vector of the transaction characterization vector of the target trading user may be normalized, the corresponding difference degree may be calculated, and the feature vector may be adjusted, and then the feature vector of the transaction characterization vector of the target trading store may be normalized, the corresponding difference degree may be calculated, and the feature vector may be adjusted. In this case, of course, the feature vector of the trading store corresponding to the target trading user when the second preset condition is met may be used as the feature vector of the target trading store.

In other embodiments, to better characterize transaction characteristics, multiple loss functions may be employed to generate multiple transaction embedded characteristic data to mine different transaction characteristics. Correspondingly, the method may further include:

calculating the difference degree between the characteristic vector of the target trading user and the trading characterization vector of the target trading user based on at least two types of preset loss functions, and calculating the difference degree between the normalized trading characterization vector of the target trading store and the characteristic vector of the target trading store;

correspondingly, the transaction embedded characteristic data of the transaction user and the transaction shop at least comprise two types of transaction embedded characteristic data.

In other embodiments, as shown in fig. 3, fig. 3 is a schematic flow chart diagram of one embodiment of determining hidden relationship characteristic data provided in the present specification. Specifically, the method may include:

s302: any two types of transaction embedded characteristic data of a transaction user and a transaction shop are respectively selected.

S304: any two types of transaction embedded characteristic data of a transaction user are respectively used as an input layer and an output layer, and a first preset hidden layer is added in the middle to construct a first neural network.

Any two types of transaction embedded feature data respectively serving as an input layer and an output layer can comprise that one transaction embedded feature data serves as the input layer, and the other transaction embedded feature data serves as the output layer; and interchanging the transaction embedded characteristic data of the input layer and the output layer.

Specifically, the first preset hidden layer may include one or more hidden layers. The activation function between the first preset hidden layer and the input layer when the first neural network is constructed may adopt a tanh hyperbolic tangent function, but the embodiments of the present specification are not limited thereto.

S306: any two types of transaction embedded characteristic data of the transaction shop are respectively used as an input layer and an output layer, and a second preset hidden layer is added in the middle to construct a second neural network.

The steps for constructing the second neural network can be referred to the above description for constructing the first neural network, and are not described herein again.

S308: and obtaining hidden relation characteristic data between any two types of transaction embedded characteristic information of the transaction user based on the learning training of the constructed first neural network.

S310: and obtaining hidden relation characteristic data between any two types of transaction embedded characteristic information of the transaction shop based on the learning training of the constructed second neural network.

According to the embodiment of the specification, the transaction characteristics can be represented from different angles due to different types of transaction embedded characteristic data, and the association degree between the user and the shop can be determined better subsequently by acquiring the hidden relation characteristic data capable of reflecting the relation between the different types of transaction characteristic data.

Since the degree of association between the user and the store can be reflected in the transaction between the user and the store, the degree of association between the user and the store can be determined by using the transaction embedded feature data representing the transaction features, and the corresponding recommendation processing can be performed based on the degree of association. Correspondingly, the method may further include:

calculating the association degree between the trading user and the trading shop by using the trading embedded characteristic data of the trading shop and the trading user;

and recommending the trading shop and the trading user corresponding to the trading embedded characteristic data with the relevance degree larger than or equal to the preset relevance value.

Specifically, in the embodiment of the present disclosure, the calculating the association degree between the trading user and the trading store by using the trading embedded feature data of the trading store and the trading user may include calculating a distance between the trading user and the trading store. In a specific embodiment, the distance between the transaction embedded feature data may be an euclidean distance between the transaction embedded feature data, and when a numerical value of the euclidean distance calculated based on the two transaction embedded feature data is smaller, it may be indicated that the better the association degree between the corresponding transaction user and the transaction store is, the higher the association degree is; conversely, when the numerical value of the euclidean distance calculated based on the two transaction embedded feature data is larger, it can be indicated that the degree of association between the corresponding transaction user and the transaction store is worse and the degree of association is lower.

Of course, the distance between the transaction embedded feature data in the embodiments of the present disclosure is not limited to the above euclidean distance, and may also include a cosine distance, a manhattan distance, and the like.

Specifically, the preset correlation value may include a recommendation condition preset according to an actual application requirement. And when the association degree between the transaction embedded characteristic data of the transaction user and the transaction shop is greater than the preset association value, recommending the corresponding transaction user and the transaction shop.

In other embodiments, the method may further comprise:

respectively taking the transaction embedded characteristic data and the corresponding hidden relation characteristic data of the transaction shop and the transaction user as the characterization data of the corresponding transaction shop and the corresponding transaction user;

calculating the association degree between the trading user and the trading shop by using the characterization data of the trading shop and the trading user;

and recommending the trading shop and the trading user corresponding to the characterization data with the relevance degree larger than or equal to the preset relevance value.

Here, calculating the association degree between the trading user and the trading store using the characterization data of the trading store and the trading user may include calculating a distance between the characterization data of the trading user and the trading store. For the distance between the characterization data to be calculated, the distance between the transaction embedded feature data is calculated as described above, and is not described herein again.

In other embodiments, the method may further comprise:

performing associated recommendation training on transaction embedded characteristic data of a transaction shop and a transaction user based on a preset machine learning algorithm to obtain a first associated recommendation model;

and performing recommendation processing between the trading user and the trading shop based on the output result of the first associated recommendation model.

The embodiment of the specification can perform association recommendation training on transaction embedded feature data of a trading store and a trading user based on a preset machine learning algorithm to obtain a first association recommendation model, determine the association probability between the trading user and the trading store by using the first association recommendation model, and perform recommendation processing by taking the probability higher than the preset probability value as an output result. Specifically, the preset probability value may be set according to a requirement for association between the user and the store in actual application.

In other embodiments, the method may further comprise:

performing association recommendation training on transaction embedded feature data and corresponding hidden relation feature data of the transaction shop and the transaction user based on a preset machine learning algorithm to obtain a second association recommendation model;

and performing recommendation processing between the trading user and the trading shop based on the output result of the second associated recommendation model.

The embodiment of the specification can perform association recommendation training on transaction embedded feature data and corresponding hidden relation feature data of a trading store and a trading user based on a preset machine learning algorithm to obtain a second association recommendation model, determine the association probability between the trading user and the trading store by using the second association recommendation model, and perform recommendation processing by taking the association probability higher than the preset probability value as an output result. Specifically, the preset probability value may be set according to a requirement for association between the user and the store in actual application.

It can be seen that one or more embodiments of a data processing method of the present specification operate by obtaining raw historical transaction data; then, directly taking the feature vectors of a first preset number of trading users corresponding to the target trading shop as the trading feature vectors of the target trading shop; taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vectors of the target trading stores and the transaction characteristic vectors of the target trading users, and adding the transaction characteristics into the corresponding characteristic vectors of the trading stores and the trading users to further obtain transaction embedded characteristic data capable of characterizing the transaction characteristics. The embodiment of the specification directly utilizes the original transaction data to learn the transaction embedded characteristic data capable of representing the transaction characteristics, so that the learned transaction embedded characteristic data can be effectively ensured to directly represent the transaction characteristics, and the data processing efficiency can be improved.

As shown in fig. 4, fig. 4 is a schematic diagram of an embodiment of calculating the association degree between a user and a merchant based on data obtained by the data processing method provided in the present specification; an embodiment of calculating the association degree between the user and the merchant based on the data obtained by the data processing method provided by the embodiment of the present specification is described below with reference to fig. 4, and as can be seen from fig. 4, in a specific embodiment, 5 users transacting with a store are selected by taking the store as a center according to historical transaction data: user 1, user 2, user 3, user 4, and user 5; and selecting 5 shops transacted with the user by taking the user as a center according to historical transaction data: store 1, store 2, store 3, store 4, and store 5; secondly, preprocessing of average value calculation is carried out on the feature vectors corresponding to the 5 users and the 5 shops; then, by continuously and alternately adjusting feature vectors of trading users and trading shops to repeat normalization processing and respectively calculating the difference degree between corresponding vectors based on a relative entropy function and a variance function, trading features can be added into the feature vectors of corresponding trading users and trading shops in the process of reducing the calculated value of a preset loss function, and further, trading embedded feature data U1, U2, S1 and S2 capable of representing the trading features are obtained; then, hidden relation characteristic data H _ U12 which can reflect the relation between U1 and U2 is determined by utilizing the transaction embedded characteristic data U1 and U2 of the user; and determining hidden relation feature data H _ S12 which can reflect the relation between S1 and S2 by using the transaction embedded feature data S1 and S2 of the shop; then, U1, U2, and H _ U12 are used as the token data of the user, and S1, S2, and H _ S12 are used as the token data of the shop; finally, a degree of association between the user and the store is determined based on a calculation between the characterization data of the user and the store. The feature data used for determining the association degree between the user and the shop not only comprises a plurality of transaction embedded feature data capable of representing transaction features, but also comprises hidden relationship feature data capable of reflecting the relationship between the transaction embedded feature data, so that features learned by different processes based on original historical transaction data are fused, and the association degree between the user and the shop can be better reflected.

In another aspect of the present specification, a data processing apparatus is further provided, and fig. 5 is a schematic structural diagram of an embodiment of the data processing apparatus provided in the specification, and as shown in fig. 5, the apparatus 500 may include:

a historical transaction data acquisition module 510, which may be configured to acquire historical transaction data;

a feature vector determination module 520, configured to determine feature vectors of trading stores and trading users in the historical trading data;

a first trading feature vector determining module 530, configured to use feature vectors of a first preset number of trading users corresponding to a target trading store as trading feature vectors of the target trading store;

the second trading feature vector determining module 540 may be configured to use feature vectors of a second preset number of trading stores corresponding to a target trading user as the trading feature vectors of the target trading user;

the characterization learning module 550 may be configured to perform transaction characterization learning on the trading stores and the trading users by using the transaction feature vector of the target trading store and the transaction feature vector of the target trading user to obtain the corresponding transaction embedded feature data of the trading stores and the trading users.

In another embodiment, the characterization learning module 550 may include:

the preprocessing unit can be used for respectively preprocessing the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain the transaction characterization vectors of the target trading store and the target trading user;

the first normalization processing unit can be used for performing normalization processing on the transaction characterization vector of the target trading store and the feature vector of the target trading store;

the first difference calculating unit may be configured to calculate a difference between the normalized transaction representation vector and the feature vector of the target trading store based on a preset loss function;

a first adjustment processing unit, configured to adjust feature vectors of a first preset number of trading users corresponding to the target trading store to change the trading feature vector of the target trading store, and repeat the steps from the preprocessing of the trading feature vector of the target trading store to the calculation of the difference until the difference between the normalized trading feature vector of the target trading store and the feature vector meets a first preset condition;

the second normalization processing unit may be configured to use, when the first preset condition is met, the feature vector of the trading user corresponding to the target trading store as the feature vector of the target trading user, and perform normalization processing on the feature vector of the target trading user and the trading characterization vector;

the second difference calculating unit may be configured to calculate a difference between the normalized feature vector of the target transaction user and the transaction characterization vector based on a preset loss function;

a second adjustment processing unit, configured to adjust feature vectors of a second preset number of trading stores corresponding to the target trading user to change the trading feature vector of the target trading user, and repeat the steps from the preprocessing of the trading feature vector of the target trading user to the calculation of the difference degree until the difference degree between the normalized trading feature vector and the feature vector of the target trading user meets a second preset condition;

the data processing unit may be configured to use the feature vector of the trading store corresponding to the target trading user when the second preset condition is met as the feature vector of the target trading store, and repeat the steps from the preprocessing to the calculation of the disparity degree until the disparity degree between the normalized trading representation vector of the target trading store and the feature vector meets the first preset condition and the disparity degree between the normalized trading representation vector of the target trading user and the feature vector meets the second preset condition;

the trading embedded feature data determining unit may be configured to use feature vectors of the trading user and the trading store when the difference between the normalized trading feature vector and the feature vector of the target trading store meets a first preset condition and the difference between the normalized trading feature vector and the feature vector of the target trading user meets a second preset condition as the trading embedded feature data of the corresponding trading user and the trading store, respectively.

In another embodiment, the apparatus 500 may further include:

the data processing module can be used for calculating the difference degree between the characteristic vector of the target trading user and the trading characterization vector of the target trading user based on at least two types of preset loss functions, and calculating the difference degree between the normalized trading characterization vector of the target trading store and the characteristic vector of the target trading store;

In another embodiment, the apparatus 500 may further include:

the transaction embedded characteristic data selection module can be used for respectively selecting any two types of transaction embedded characteristic data of a transaction user and a transaction shop;

the first neural network construction module can be used for respectively taking any two types of transaction embedded characteristic data of a transaction user as an input layer and an output layer, and adding a first preset hidden layer in the middle to construct a first neural network;

the second neural network construction module can be used for respectively taking any two types of transaction embedded characteristic data of the transaction stores as an input layer and an output layer, and adding a second preset hidden layer in the middle to construct a second neural network;

the first learning training module can be used for learning training based on the constructed first neural network to obtain hidden relation feature data between any two types of transaction embedded feature information of transaction users;

and the second learning training module can be used for learning training based on the constructed second neural network to obtain hidden relation characteristic data between any two types of transaction embedded characteristic information of the transaction shop.

In another embodiment, the pre-treatment may include any one of:

In another embodiment, the historical transaction data may further include: the transaction time.

In another embodiment, the first transaction feature vector determination module 530 may include:

the first feature vector selecting unit may be configured to select feature vectors of a first preset number of trading users corresponding to the target trading store according to trading time in the trading data corresponding to the target trading store;

the first trading feature vector determination unit may be configured to use feature vectors of the first preset number of trading users as trading feature vectors of the target trading store.

In another embodiment, the second transaction feature vector determination module 540 may include:

the second feature vector selecting unit may be configured to select feature vectors of a second preset number of trading stores corresponding to the target trading user according to trading time in the trading data corresponding to the target trading user;

the second trading feature vector determination unit may be configured to use the feature vectors of the second preset number of trading stores as the trading feature vector of the target trading user.

In another embodiment, the apparatus may further include:

the first association degree calculation module can be used for calculating the association degree between the trading user and the trading shop by utilizing the trading embedded characteristic data of the trading shop and the trading user;

the first recommendation processing module can be used for performing recommendation processing between the trading stores and the trading users corresponding to the trading embedded characteristic data with the relevance degree being greater than or equal to the preset relevance value.

In another embodiment, the apparatus 500 may further include:

the characterization data determining module can be used for respectively taking the transaction embedded characteristic data and the corresponding hidden relation characteristic data of the trading stores and the trading users as the characterization data of the corresponding trading stores and the corresponding trading users;

the second association degree calculation module can be used for calculating the association degree between the trading user and the trading store by utilizing the characterization data of the trading store and the trading user;

the second recommendation processing module can be used for performing recommendation processing between the trading stores and the trading users corresponding to the characterization data with the relevance degree larger than or equal to the preset relevance value.

In another embodiment, the apparatus 500 may further include:

the first association recommendation model determining module can be used for performing association recommendation training on transaction embedded characteristic data of a transaction shop and a transaction user based on a preset machine learning algorithm to obtain a first association recommendation model;

and the third recommendation processing module can be used for performing recommendation processing between the trading user and the trading shop based on the output result of the first associated recommendation model.

In another embodiment, the apparatus 500 may further include:

the second association recommendation model determining module can be used for performing association recommendation training on transaction embedded feature data and corresponding hidden relation feature data of the transaction shop and the transaction user based on a preset machine learning algorithm to obtain a second association recommendation model;

and the fourth recommendation processing module can be used for performing recommendation processing between the trading user and the trading shop based on the output result of the second associated recommendation model.

The data processing method or apparatus provided in the embodiments of the present specification may be implemented in a computer by a processor executing corresponding program instructions, for example, implemented at a PC end using a c + + language of a windows operating system, or implemented at an intelligent terminal using, for example, android and iOS system programming languages, and implemented based on processing logic of a quantum computer. Accordingly, another aspect of the present specification also provides a data processing server comprising a processor and a memory, the memory storing computer program instructions executed by the processor, the computer program instructions may include:

Specifically, in the embodiment of the present disclosure, the processor may include a Central Processing Unit (CPU), and may also include other single-chip microcomputers, logic gates, integrated circuits, and the like with logic processing capability, or a suitable combination thereof. The memory may include a non-volatile memory or the like.

The performing of the transaction characterization learning of the trading stores and the trading users by using the transaction feature vector of the target trading store and the transaction feature vector of the target trading users to obtain the transaction embedded feature data of the corresponding trading stores and the corresponding trading users may include:

respectively preprocessing the transaction characteristic vector of the target transaction shop and the transaction characteristic vector of the target transaction user to obtain the transaction characteristic vectors of the target transaction shop and the target transaction user;

normalizing the transaction characterization vector of the target trading store and the feature vector of the target trading store, and calculating the difference between the normalized transaction characterization vector and the feature vector of the target trading store based on a preset loss function;

adjusting the feature vectors of a first preset number of trading users corresponding to the target trading shop to change the trading feature vectors of the target trading shop, and repeating the steps from the preprocessing of the trading feature vectors of the target trading shop to the calculation of the difference degree until the difference degree between the normalized trading feature vectors and the feature vectors of the target trading shop meets a first preset condition;

taking the feature vector of the trading user corresponding to the target trading shop when the first preset condition is met as the feature vector of the target trading user, carrying out normalization processing on the feature vector of the target trading user and the trading characterization vector, and calculating the difference between the feature vector and the trading characterization vector after the target trading user is normalized based on a preset loss function;

adjusting the feature vectors of a second preset number of trading shops corresponding to the target trading user to change the trading feature vectors of the target trading user, and repeating the steps from the preprocessing of the trading feature vectors of the target trading user to the calculation of the difference degree until the difference degree between the normalized trading feature vectors and the feature vectors of the target trading user meets a second preset condition;

taking the feature vector of the trading shop corresponding to the target trading user when the feature vector meets the second preset condition as the feature vector of the target trading shop, and repeating the steps from the preprocessing to the calculation of the difference degree until the difference degree between the normalized trading representation vector of the target trading shop and the feature vector meets the first preset condition and the difference degree between the normalized trading representation vector of the target trading user and the feature vector meets the second preset condition;

and respectively taking the feature vectors of the trading users and the trading stores as the trading embedded feature data of the corresponding trading users and the trading stores when the difference between the normalized trading representation vector and the feature vector of the target trading store meets a first preset condition and the difference between the normalized trading representation vector and the feature vector of the target trading user meets a second preset condition.

In another embodiment, the computer program instructions may further comprise:

any two types of transaction embedded characteristic data of a transaction user and a transaction shop are selected respectively;

any two types of transaction embedded characteristic data of a transaction user are respectively used as an input layer and an output layer, and a first preset hidden layer is added in the middle to construct a first neural network;

any two types of transaction embedded characteristic data of the transaction shop are respectively used as an input layer and an output layer, and a second preset hidden layer is added in the middle to construct a second neural network;

obtaining hidden relation feature data between any two types of transaction embedded feature information of a transaction user based on the learning training of the constructed first neural network;

and obtaining hidden relation characteristic data between any two types of transaction embedded characteristic information of the transaction shop based on the learning training of the constructed second neural network.

In another embodiment, the preprocessing may include any one of the following:

In another embodiment, wherein the historical transaction data further comprises: the transaction time.

In another embodiment, the taking the feature vector of the trading users of the first preset number corresponding to the target trading store as the trading feature vector of the target trading store includes:

In another embodiment, the taking the feature vector of the trading stores of the second preset number corresponding to the target trading user as the trading feature vector of the target trading user includes:

In another embodiment, the computer program instructions further comprise:

In another embodiment, the computer program instructions may further comprise:

Thus, embodiments of a data processing method, apparatus, or server herein provide for obtaining raw historical transaction data; then, directly taking the feature vectors of a first preset number of trading users corresponding to the target trading shop as the trading feature vectors of the target trading shop; taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; and performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vectors of the target trading stores and the transaction characteristic vectors of the target trading users, and adding the transaction characteristics into the corresponding characteristic vectors of the trading stores and the trading users to further obtain transaction embedded characteristic data capable of characterizing the transaction characteristics. The embodiment of the specification directly utilizes the original transaction data to learn the transaction embedded characteristic data capable of representing the transaction characteristics, so that the learned transaction embedded characteristic data can be effectively ensured to directly represent the transaction characteristics, and the data processing efficiency can be improved.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain the corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical modules. For example, a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), is an integrated circuit whose Logic functions are determined by programming the Device by a user. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually making an Integrated Circuit chip, such Programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development and writing, but the original code before compiling is also written by a specific Programming Language, which is called Hardware Description Language (HDL), and HDL is not only one but many, such as abel (advanced Boolean Expression Language), ahdl (alternate Hardware Description Language), traffic, pl (core universal Programming Language), HDCal (jhdware Description Language), lang, Lola, HDL, laspam, hardward Description Language (vhr Description Language), vhal (Hardware Description Language), and vhigh-Language, which are currently used in most common. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.

The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.

The apparatuses, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or implemented by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functions of the various elements may be implemented in the same one or more software and/or hardware implementations of the present description.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage, graphene storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, apparatus or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

This description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the device and server embodiments, since they are substantially similar to the method embodiments, the description is simple, and the relevant points can be referred to the partial description of the method embodiments.

The above description is only an example of the present specification, and is not intended to limit the present specification. Various modifications and alterations to this description will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of the present specification should be included in the scope of the claims.

Claims

1. A method of data processing, comprising:

taking the feature vectors of a first preset number of trading users corresponding to a target trading shop as the trading feature vectors of the target trading shop; the target trading store is one of the trading stores;

taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; the target trading user is one of the trading users;

performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain transaction embedded characteristic data of the corresponding trading stores and the corresponding trading users;

respectively taking the transaction embedded characteristic data and the corresponding hidden relation characteristic data of the transaction shop and the transaction user as the characterization data of the corresponding transaction shop and the corresponding transaction user; wherein the hidden relationship feature data reflects a relationship between the transaction embedded feature data;

2. The method of claim 1, wherein the performing transaction characterization learning of trading stores and trading users by using the transaction feature vector of the target trading store and the transaction feature vector of the target trading user to obtain the transaction embedded feature data of the corresponding trading stores and trading users comprises:

3. The method of claim 2, wherein the method further comprises:

4. The method of claim 3, wherein the method further comprises:

5. The method of any of claims 2 to 4, wherein the pre-processing comprises any of:

average value calculation processing, summation calculation processing and maximum value taking processing.

6. The method of any of claims 1 to 4, wherein the historical transaction data further comprises: the transaction time.

7. The method of claim 6, wherein the using, as the trading feature vector of the target trading store, the feature vector of the first preset number of trading users corresponding to the target trading store comprises:

8. The method according to claim 6, wherein the taking the feature vector of the trading store with the second preset number corresponding to the target trading user as the trading feature vector of the target trading user comprises:

9. The method of any of claims 1 to 4, wherein the method further comprises:

10. The method of any of claims 1 to 4, wherein the method further comprises:

11. The method of claim 4, wherein the method further comprises:

12. A data processing apparatus comprising:

the system comprises a first trading feature vector determining module, a second trading feature vector determining module and a third trading feature vector determining module, wherein the first trading feature vector determining module is used for taking feature vectors of a first preset number of trading users corresponding to a target trading store as trading feature vectors of the target trading store; the target trading store is one of the trading stores;

the second trading feature vector determining module is used for taking the feature vectors of a second preset number of trading shops corresponding to the target trading user as the trading feature vectors of the target trading user; the target trading user is one of the trading users;

the characterization learning module is used for performing transaction characterization learning on the trading stores and the trading users by using the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading users to obtain corresponding transaction embedded characteristic data of the trading stores and the trading users;

the characterization data determining module is used for respectively taking the transaction embedded characteristic data and the corresponding hidden relation characteristic data of the trading stores and the trading users as the characterization data of the corresponding trading stores and the corresponding trading users; wherein the hidden relationship feature data reflects a relationship between the transaction embedded feature data;

the second association degree calculation module is used for calculating the association degree between the trading user and the trading shop by utilizing the characterization data of the trading shop and the trading user;

and the second recommendation processing module is used for performing recommendation processing between the trading stores and the trading users corresponding to the characterization data with the relevance degree larger than or equal to the preset relevance value.

13. The apparatus of claim 12, wherein the characterization learning module comprises:

the preprocessing unit is used for respectively preprocessing the transaction characteristic vector of the target trading store and the transaction characteristic vector of the target trading user to obtain the transaction characteristic vectors of the target trading store and the target trading user;

the first normalization processing unit is used for performing normalization processing on the transaction representation vector of the target trading store and the feature vector of the target trading store;

the first difference calculating unit is used for calculating the difference between the normalized transaction representation vector and the feature vector of the target transaction shop based on a preset loss function;

a first adjustment processing unit, configured to adjust feature vectors of a first preset number of trading users corresponding to the target trading store to change a trading feature vector of the target trading store, and repeat the steps from the preprocessing of the trading feature vector of the target trading store to the calculation of a difference until a difference between the normalized trading feature vector of the target trading store and the feature vector meets a first preset condition;

the second normalization processing unit is used for taking the feature vector of the trading user corresponding to the target trading shop as the feature vector of the target trading user when the first preset condition is met, and performing normalization processing on the feature vector of the target trading user and the trading representation vector;

the second difference degree calculation unit is used for calculating the difference degree between the normalized feature vector of the target trading user and the trading characterization vector based on a preset loss function;

the second adjustment processing unit is used for adjusting the feature vectors of a second preset number of trading shops corresponding to the target trading user to change the trading feature vector of the target trading user, and repeating the steps from the preprocessing of the trading feature vector of the target trading user to the calculation of the difference degree until the difference degree between the normalized trading feature vector and the feature vector of the target trading user meets a second preset condition;

the data processing unit is used for taking the feature vector of the trading shop corresponding to the target trading user when the feature vector meets the second preset condition as the feature vector of the target trading shop, and repeating the steps from the preprocessing to the calculation of the difference until the difference between the normalized trading representation vector of the target trading shop and the feature vector meets the first preset condition and the difference between the normalized trading representation vector of the target trading user and the feature vector meets the second preset condition;

and the transaction embedded characteristic data determining unit is used for respectively taking the characteristic vectors of the transaction user and the transaction store as the transaction embedded characteristic data of the corresponding transaction user and the transaction store when the difference degree between the transaction characteristic vector and the characteristic vector after the target transaction store is normalized meets a first preset condition and the difference degree between the transaction characteristic vector and the characteristic vector after the target transaction user is normalized meets a second preset condition.

14. The apparatus of claim 13, wherein the apparatus further comprises:

the data processing module is used for calculating the difference between the characteristic vector of the target trading user and the trading characterization vector of the target trading user based on at least two types of preset loss functions and calculating the difference between the normalized trading characterization vector of the target trading store and the characteristic vector of the target trading store;

15. The apparatus of claim 14, wherein the apparatus further comprises:

the transaction embedded characteristic data selection module is used for respectively selecting any two types of transaction embedded characteristic data of a transaction user and a transaction shop;

the first neural network construction module is used for respectively taking any two types of transaction embedded characteristic data of a transaction user as an input layer and an output layer, and adding a first preset hidden layer in the middle to construct a first neural network;

the second neural network construction module is used for respectively taking any two types of transaction embedded characteristic data of the transaction stores as an input layer and an output layer, and adding a second preset hidden layer in the middle to construct a second neural network;

the first learning training module is used for obtaining hidden relation characteristic data between any two types of transaction embedded characteristic information of a transaction user based on the learning training of the constructed first neural network;

and the second learning training module is used for obtaining hidden relation characteristic data between any two types of transaction embedded characteristic information of the transaction shop based on the learning training of the constructed second neural network.

16. The apparatus of any one of claims 13 to 15, wherein the pre-processing comprises any one of:

17. The apparatus of any of claims 13 to 15, wherein the historical transaction data further comprises: the transaction time.

18. The apparatus of claim 17, wherein the first transaction feature vector determination module comprises:

the first feature vector selecting unit is used for selecting feature vectors of a first preset number of trading users corresponding to the target trading shop according to the trading time in the trading data corresponding to the target trading shop;

and the first trading feature vector determining unit is used for taking the feature vectors of the trading users of the first preset number as the trading feature vectors of the target trading shops.

19. The apparatus of claim 17, wherein the second transaction feature vector determination module comprises:

the second feature vector selecting unit is used for selecting feature vectors of a second preset number of trading shops corresponding to the target trading user according to the trading time in the trading data corresponding to the target trading user;

and the second trading feature vector determining unit is used for taking the feature vectors of the trading shops with the second preset number as the trading feature vectors of the target trading users.

20. The apparatus of any one of claims 12 to 15, wherein the apparatus further comprises:

the first association degree calculation module is used for calculating the association degree between the trading user and the trading shop by utilizing the trading embedded characteristic data of the trading shop and the trading user;

and the first recommendation processing module is used for performing recommendation processing between the trading stores and the trading users corresponding to the trading embedded characteristic data with the relevance degree being more than or equal to the preset relevance value.

21. The apparatus of any one of claims 12 to 15, wherein the apparatus further comprises:

the first association recommendation model determining module is used for performing association recommendation training on transaction embedded characteristic data of a transaction shop and a transaction user based on a preset machine learning algorithm to obtain a first association recommendation model;

and the third recommendation processing module is used for performing recommendation processing between the trading user and the trading shop based on the output result of the first associated recommendation model.

22. The apparatus of claim 15, wherein the apparatus further comprises:

the second association recommendation model determining module is used for performing association recommendation training on the transaction embedded characteristic data and the corresponding hidden relation characteristic data of the transaction shop and the transaction user based on a preset machine learning algorithm to obtain a second association recommendation model;

and the fourth recommendation processing module is used for performing recommendation processing between the trading user and the trading shop based on the output result of the second associated recommendation model.

23. A data processing server comprising a processor and a memory, the memory storing computer program instructions for execution by the processor, the computer program instructions comprising:

24. The server according to claim 23, wherein the performing transaction characterization learning of trading stores and trading users by using the transaction feature vector of the target trading store and the transaction feature vector of the target trading user to obtain the transaction embedded feature data of the corresponding trading stores and trading users comprises:

25. The server of claim 24, wherein the computer program instructions further comprise:

26. The server of claim 25, wherein the computer program instructions further comprise:

27. The server of any one of claims 24 to 26, wherein the pre-processing comprises any one of:

28. The server of any one of claims 23 to 26, wherein the historical transaction data further comprises: the transaction time.

29. The server according to claim 28, wherein the taking the feature vector of the first preset number of trading users corresponding to the target trading store as the trading feature vector of the target trading store comprises:

30. The server according to claim 28, wherein the taking the feature vector of the trading store of the second preset number corresponding to the target trading user as the trading feature vector of the target trading user comprises:

31. The server of any of claims 23 to 26, wherein the computer program instructions further comprise:

32. The server of any of claims 23 to 26, wherein the computer program instructions further comprise:

33. The server of claim 26, wherein the computer program instructions further comprise: