CN110020938B

CN110020938B - Transaction information processing method, device, equipment and storage medium

Info

Publication number: CN110020938B
Application number: CN201910062959.0A
Authority: CN
Inventors: 高利翠; 肖凯
Original assignee: Advanced New Technologies Co Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2019-01-23
Filing date: 2019-01-23
Publication date: 2024-01-16
Anticipated expiration: 2039-01-23
Also published as: CN110020938A

Abstract

The embodiment of the application provides a transaction information processing method, device, equipment and storage medium. The method comprises the following steps: generating a behavioral sequence of the target subject based on the transaction behavioral data of the target subject; extracting transaction behavior characteristics of a target subject from a behavior sequence of the target subject, wherein the transaction behavior characteristics comprise subject variable information of transaction behavior data, and the subject variable information comprises information representing attribution information of the target subject; generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on dimension information of the transaction behavior feature; and counting each dimension information of the embedded vector based on the subject variable information so as to predict the transaction risk of the target subject based on the counting result. According to the technical scheme, the accuracy of risk assessment of transaction behaviors with unobvious abnormal characteristics can be improved, and the data processing efficiency can be improved.

Description

Transaction information processing method, device, equipment and storage medium

Technical Field

The present invention relates to the field of big data technologies, and in particular, to a transaction information processing method, a transaction information processing device, a transaction information processing apparatus, and a storage medium.

Background

With the rapid development of internet technology, more and more people choose to conduct various business transactions such as shopping, transferring accounts, remittance and the like through the internet, and how to guarantee the security of the business transactions becomes a focus of attention.

In one technical scheme, risk of current transaction behavior of a bank account is evaluated according to historical transaction behavior data of the bank account, and whether the current transaction behavior of the bank account is abnormal transaction is determined based on an evaluation result. However, in such a solution, it is difficult to accurately identify transaction risk for transaction behaviors with insignificant abnormal characteristics, such as transaction behaviors of a new bank account.

Disclosure of Invention

An object of an embodiment of the present application is to provide a transaction information processing method, a transaction information processing apparatus, a transaction information processing device, and a storage medium, to solve a problem of difficulty in accurately identifying transaction risk of a transaction behavior with insignificant abnormal characteristics.

In order to solve the technical problems, the embodiment of the application is realized as follows:

according to a first aspect of an embodiment of the present application, there is provided a transaction information processing method, including: generating a behavior sequence of a target subject based on transaction behavior data of the target subject; extracting transaction behavior characteristics of the target subject from a behavior sequence of the target subject, wherein the transaction behavior characteristics comprise subject variable information of the transaction behavior data, and the subject variable information comprises information representing attribution information of the target subject; generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature; and counting each dimension information of the embedded vector based on the subject variable information so as to predict the transaction risk of the target subject based on a counting result.

In some embodiments of the present application, based on the foregoing solution, generating, by word embedding, an embedded vector corresponding to the transaction behavior feature based on dimension information of the transaction behavior feature includes: clustering all feature items of the transaction behavior feature, and determining dimension information of the transaction behavior feature based on a clustering result; and mapping the transaction behavior characteristics into corresponding embedded vectors through a word embedding model based on the dimension information of the transaction behavior characteristics.

In some embodiments of the present application, based on the foregoing solution, the word embedding model is a long-short-term memory LSTM model, and mapping the transaction behavior feature into a corresponding embedding vector through the word embedding model based on dimension information of the transaction behavior feature includes: inputting dimension information of the transaction behavior characteristics into the LSTM model; extracting corresponding dimension characteristics from the transaction behavior characteristics through a hidden layer of the LSTM model based on the dimension information of the transaction behavior characteristics; outputting each dimension characteristic of the extracted transaction behavior characteristic to a preset full-connection layer, and generating a corresponding embedded vector through the preset full-connection layer.

In some embodiments of the present application, based on the foregoing solution, the method further includes: acquiring historical transaction behavior data of a plurality of target subjects; extracting transaction behavior features of the plurality of target subjects from the historical transaction behavior data; training the LSTM model based on the transaction behavior features of the plurality of target subjects and the dimension information of the transaction behavior features.

In some embodiments of the present application, based on the foregoing scheme, counting the respective dimension information of the embedded vector based on the subject variable information includes: counting each dimension information of the embedded vector based on the complete information of the main variable information; and/or counting each dimension information of the embedded vector based on the attribution information of the target subject contained in the subject variable information.

In some embodiments of the present application, based on the foregoing solution, the principal variable information is bank account information, and counting each dimension information of the embedded vector based on the attribution information of the target principal contained in the principal variable information includes: and counting the dimension information of the embedded vector based on the bank identification code contained in the bank account information.

In some embodiments of the present application, based on the foregoing scheme, performing statistics on each dimension information of the embedded vector includes: and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

In some embodiments of the present application, predicting the transaction risk of the target subject based on the statistics based on the foregoing scheme includes: classifying the target subject into a new account or an old account based on the subject variable information; if the target subject is a new account, predicting transaction risk of the target subject based on a statistical result corresponding to the attribution information of the target subject contained in the subject variable information; and if the target subject is a old account, predicting the transaction risk of the target subject based on a statistical result corresponding to the complete information of the subject variable information.

In some embodiments of the present application, predicting the transaction risk of the target subject based on the statistics based on the foregoing scheme includes: acquiring statistical results of each dimension of the embedded vector corresponding to the main variable; and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

According to a second aspect of the embodiments of the present application, there is provided a transaction information processing apparatus including: a sequence generating unit, configured to generate a behavior sequence of a target subject based on transaction behavior data of the target subject; a feature extraction unit, configured to extract a transaction behavior feature of the target subject from a behavior sequence of the target subject, where the transaction behavior feature includes subject variable information of the transaction behavior data, and the subject variable information includes information indicating attribution information of the target subject; the embedded vector generation unit is used for generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature; and the statistics unit is used for counting each dimension information of the embedded vector based on the subject variable information so as to predict the transaction risk of the target subject based on a counting result.

In some embodiments of the present application, based on the foregoing scheme, the embedded vector generating unit includes: the dimension information determining unit is used for carrying out clustering processing on each characteristic item of the transaction behavior characteristics and determining dimension information of the transaction behavior characteristics based on a clustering result; and the mapping unit is used for mapping the transaction behavior characteristics into corresponding embedded vectors through a word embedding model based on the dimension information of the transaction behavior characteristics.

In some embodiments of the present application, based on the foregoing solution, the word embedding model is a long-short-term memory LSTM model, and the mapping unit includes: an input unit for inputting the transaction behavior feature and dimension information of the transaction behavior feature to the LSTM model; the feature extraction unit is used for extracting corresponding dimension features from the transaction behavior features through a hidden layer of the LSTM model based on the dimension information of the transaction behavior features; and the feature output unit is used for outputting each dimension feature of the extracted transaction behavior feature to a preset full-connection layer, and generating a corresponding embedded vector through the preset full-connection layer.

In some embodiments of the present application, based on the foregoing aspect, the transaction information processing apparatus further includes: a history data acquisition unit configured to acquire history transaction behavior data of a plurality of target subjects; a behavior feature extraction unit for extracting transaction behavior features of the plurality of target subjects from the historical transaction behavior data; and the training unit is used for training the LSTM model based on the transaction behavior characteristics of the target subjects and the dimension information of the transaction behavior characteristics.

In some embodiments of the present application, based on the foregoing scheme, the statistical unit includes: the first statistics unit is used for counting each dimension information of the embedded vector based on the complete information of the main variable information; and/or a second statistics unit, configured to perform statistics on each dimension information of the embedded vector based on the attribution information of the target subject included in the subject variable information.

In some embodiments of the present application, based on the foregoing solution, the principal variable information is bank account information, and the second statistical unit is configured to: and counting the dimension information of the embedded vector based on the bank identification code contained in the bank account information.

In some embodiments of the present application, based on the foregoing scheme, the statistical unit is configured to: and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

In some embodiments of the present application, based on the foregoing scheme, the statistical unit includes: the dividing unit is used for dividing the target subject into a new account or an old account based on the subject variable information; the first prediction unit is used for predicting the transaction risk of the target subject based on a statistical result corresponding to the attribution information of the target subject contained in the subject variable information if the target subject is a new account; and the second prediction unit is used for predicting the transaction risk of the target subject based on the statistical result corresponding to the complete information of the subject variable information if the target subject is an old account.

In some embodiments of the present application, based on the foregoing scheme, the statistical unit is configured to: acquiring statistical results of each dimension of the embedded vector corresponding to the main variable; and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

According to a third aspect of the embodiments of the present application, there is provided a transaction information processing apparatus including: a processor; and a memory configured to store computer-executable instructions that, when executed, cause the processor to implement the steps of the transaction information processing method of any of the first aspects above.

According to a fourth aspect of embodiments of the present application, there is provided a storage medium storing computer executable instructions which, when executed, implement the steps of the transaction information processing method of any one of the above-described first aspects.

According to the technical scheme, on one hand, the transaction behavior characteristics of the target main body containing the attribution information are determined based on the transaction behavior sequence of the target main body, so that the transaction behavior characteristics of the target main body are conveniently counted based on the attribution information of the target main body; on the other hand, the dimension information based on the transaction behavior features generates the embedded vector corresponding to the transaction behavior features in a word embedding mode, so that the transaction behavior features of the target main body can be fully extracted in each dimension, the calculated amount can be reduced, and the data processing efficiency is improved; in still another aspect, statistics is performed on each dimension information of the embedded vector based on the subject variable information, and transaction risk of the new target subject can be predicted based on the statistical result, so that accuracy of risk assessment of transaction behaviors with unobvious abnormal features can be improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present application, and that other drawings may be obtained according to these drawings without inventive effort to a person skilled in the art.

FIG. 1 illustrates a schematic block diagram of an application scenario of a transaction information processing method provided in accordance with some embodiments of the present application;

FIG. 2 illustrates a flow diagram of a transaction information processing method provided in accordance with some embodiments of the present application;

FIG. 3 illustrates a flow diagram for generating an embedded vector provided in accordance with some embodiments of the present application;

FIG. 4 is a flow chart illustrating a transaction information processing method provided in accordance with further embodiments of the present application;

FIG. 5 shows a flow diagram of a transaction information processing method provided in accordance with further embodiments of the present application;

FIG. 6 illustrates a schematic block diagram of a transaction information processing device provided in accordance with some embodiments of the present application;

FIG. 7 illustrates a schematic block diagram of an embedded vector generation unit provided in accordance with some embodiments of the present application;

FIG. 8 illustrates a schematic block diagram of a mapping unit provided in accordance with some embodiments of the present application; and

fig. 9 illustrates a schematic block diagram of a transaction information processing device provided in accordance with some embodiments of the present application.

Detailed Description

In order to better understand the technical solutions in the present application, the following description will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.

Fig. 1 illustrates a schematic block diagram of an application scenario of a transaction information processing method provided according to some embodiments of the present application. Referring to fig. 1, the application scenario may include: at least one client 110 and a server 120. The client 110 communicates with the server 120 via a network 130. The payment application of the bank or the third party payment mechanism is installed on the client 110, and the user can perform transaction actions of offline or online payment, such as two-dimension code payment, small-amount secret payment, transfer or recharging, and the like through the payment application on the client 110, and record transaction action data related to the corresponding transaction actions. The server 120 confirms the transaction performed by the client 110, completes the corresponding transaction, and records the transaction data related to the transaction.

It should be noted that, the client 110 may be a mobile phone, a tablet computer, a desktop computer, a portable notebook computer, a POS (Point Of sale) terminal, or the like. The server side 120 may be a physical server including an independent host, or a virtual server carried by a host cluster, or a cloud server. The network 130 may be a wired network or a wireless network, for example, the network 130 may be a public switched telephone network (Public Switched Telephone Network, PSTN) or the internet.

It should be noted that, the steps in the transaction information processing method in the exemplary embodiment of the present application may be partially executed by the client 110, partially executed by the server 120, or may be entirely executed by the server 120, which is not particularly limited in the present invention.

A transaction information processing method according to an exemplary embodiment of the present application is described below with reference to fig. 2 in conjunction with the application scenario of fig. 1. It should be noted that the above application scenario is only shown for the convenience of understanding the spirit and principle of the present invention, and the embodiments of the present invention are not limited in any way. Rather, embodiments of the invention may be applied to any scenario where applicable.

Fig. 2 is a flow chart illustrating a transaction information processing method according to some embodiments of the present application, which may be applied to the server side 120 in fig. 1. Referring to fig. 2, the transaction information processing method includes steps S210 to S240, and the transaction information processing method in the exemplary embodiment of fig. 2 is described in detail below.

Referring to fig. 2, in step S210, a behavior sequence of a target subject is generated based on transaction behavior data of the target subject.

In an exemplary embodiment, the target entity is an entity capable of reflecting the attribution information of the user performing the transaction, where the attribution information may be geographical location information of the user performing the transaction, and the target entity may be bank account information corresponding to the transaction, such as a bank card number, an IP (Internet Protocol, network protocol) address, a phone number, an identification card number, or other suitable entities, such as an IMEI (International Mobile Equipment Identity, an international mobile equipment identity) of a device corresponding to the transaction, a MAC (Media Access Control, a media access control) address of the device, and the like, which is also within the scope of protection of the present application. In the following embodiments, a bank card number will be described as an example.

In an example embodiment, transaction behavior data of a target subject may be acquired over a predetermined period of time, which may be a period of 15 days, 1 month, 2 months, or 3 months, or the like. The transaction behavior data of the target subject may include transfer information, payment information, and collection information of the target subject, and may also include other data related to the transaction behavior, such as login data, recharging data, or shopping data, which is not particularly limited in this application.

Further, the behavior sequence of the target subject is generated based on the transaction behavior data of the target subject in a predetermined time period, the behavior sequence of the target subject is the transaction behavior data of the target subject in a plurality of time periods sequenced according to the occurrence time of the transaction, for example, if the behavior sequence of the target subject is generated based on the transaction behavior data of the target subject in a predetermined time period, for example, 1 month, the transaction behavior data in the predetermined time period can be divided based on a predetermined interval, for example, 1 day, so as to obtain the transaction behavior data in a plurality of time periods, and the transaction behavior data in each time period is sequenced according to the occurrence time, so as to generate the behavior sequence in the predetermined time period corresponding to the target subject.

In step S220, a transaction behavior feature of the target subject is extracted from the behavior sequence of the target subject, wherein the transaction behavior feature includes subject variable information of the target subject corresponding to the transaction behavior data, and the subject variable information includes information indicating attribution information of the target subject.

In an example embodiment, the transaction performance characteristics of the target subject may include subject variable information, transaction amount, payee information or payer information, transaction location, transaction time, and transaction type characteristics of the target subject, as well as other suitable transaction performance characteristics such as characteristic information such as payment means, coupon usage information, etc. The subject variable information of the target subject includes information indicating the home information of the target subject, for example, when the target subject is a bank card, the subject variable information of the target subject includes a bank card number, a bank card BIN (Bank Identification Number, bank identification code) included in the bank card number can indicate the home information of the bank card, the bank card BIN is typically indicated by a 6-bit number, and appears in the first 6 bits of the bank card number, for example, a card BIN of a commercial bank in the city of partnership is 603601, a card BIN of a commercial bank in the city of no tin is 603265, a card BIN of a commercial bank in the countryside of the city of permanence is 603694, and a card BIN of a commercial bank in the large company is 603708. When the target subject is an identity card, the subject variable information of the target subject contains an identity card number, and the first 6 bits of the identity card number can represent the attribution information of the identity card, for example, 110000 in Beijing city, 120000 in Tianjin city, 130000 in Hebei province, 140000 in Shanxi province, and 150000 in inner Mongolia.

Further, in an example embodiment, word segmentation processing is performed on transaction behavior data in a behavior sequence of a target subject, words related to transaction risk identification, such as subject variable, transaction amount, transaction type, transaction time and the like, are extracted from the transaction behavior data in the word segmented behavior sequence, words related to transaction risk identification are encoded, such as one-hot one-time encoding, and feature vectors of transaction behavior features corresponding to the behavior sequence of the target subject are generated. For example, when the word related to the transaction risk identification includes the subject variable of the target subject, the transaction amount, and the transaction type, the one-time thermal code of the subject variable of the target subject is [1, 0], the one-time thermal code of the transaction amount is [0,1, 0], the one-time thermal code of the transaction type is [0,1, 0], and the feature vector of the corresponding transaction behavior feature is generated based on the code of the word related to the transaction risk identification in the transaction behavior data.

Although the transaction behavior data is encoded by the single-hot encoding in the present exemplary embodiment, other suitable encoding methods may be used to encode the transaction behavior data, for example, a bag-of-word model, a TF-IDF (Term Frequency-inverse document Frequency) model, etc., which is not particularly limited in this application.

Next, in step S230, an embedding vector corresponding to the transaction behavior feature is generated by means of word embedding based on the dimension information of the transaction behavior feature of the target subject.

In an example embodiment, the dimension information of the transaction behavior feature of the target subject is used to identify the same or similar feature information in the transaction behavior feature, and the dimension information of the transaction behavior feature of the target subject may include: one or more of transaction behavior attribution dimension information, transaction behavior type dimension information, transaction amount dimension information, and transaction time dimension information. For example, the transaction behavior attribution dimension information may include attribution dimension information such as bank card number information, IP address information, mobile phone number information, identity card number information, and the like; the transaction behavior type dimension information may include: type dimension information such as consumer transactions, insurance transactions, financial transactions, borrowing and repayment transactions, wage transactions and the like; the transaction amount dimension information may include a plurality of amount intervals such as [0, 5000], [5000, 10000], [10000, 20000], [20000, 50000] and the like amount dimension information; the transaction time dimension information may include a plurality of time intervals, for example, 2018.1.1 to 2018.5.1, and the plurality of time intervals may be [2018.1.1, 2018.2.1], [2018.2.1, 2018.3.1], [2018.3.1, 2018.4.1], [2018.4.1, 2018.5.1], or the like dimension information.

Further, based on the dimension information of the transaction behavior feature, an embedded vector corresponding to the transaction behavior feature is generated in a word embedding mode. Word embedding is a distributed representation of words by which a high-dimensional word space can be mapped into a low-dimensional vector space. In an example embodiment, after the dimensional information of the transaction behavior feature is determined, the multi-dimensional transaction behavior feature may be mapped to a low-dimensional embedded vector corresponding to the dimensional information through a word embedded model, for example, an LSTM (Long Short-Term Memory) model, so that the calculation amount can be reduced and the data processing efficiency can be improved. For example, the dimension information that determines the transaction behavior feature includes four-dimensional information, namely transaction behavior attribution dimension information, transaction behavior type dimension information, transaction amount dimension information, and transaction time dimension information, and the transaction behavior feature may be mapped to the four-dimensional embedded vectors through a word embedding model.

It should be noted that, in this exemplary embodiment, the Word embedding manner may be an LSTM model, or may be a Word2Vec model or a Glove model, or may be another suitable neural network model, for example, a Skip-Gram model or a CBOW (Continuous Bag of Words model), which is also within the scope of the present application.

In step S240, statistics are performed on each dimension information of the embedded vector based on the subject variable information of the target subject, so as to predict the transaction risk of the target subject based on the statistics result.

In an example embodiment, when the target subject is a bank card number, the subject variable information of the target subject includes bank card number information, and the bank card number information includes home information of the target subject, that is, card BIN information of the bank card, and may be counted on each dimension information of the embedded vector based on complete information of the bank card number, that is, the entire card number, or may be counted on each dimension information of the embedded vector based on the card BIN of the bank card. When the target subject is identity card information, the subject variable of the target subject contains identity card number information, the identity card number information contains attribution information of the target subject, namely address information represented by the first 6 bits of the identity card number, and statistics can be carried out on each dimension information of the embedded vector based on complete information of the identity card number, namely the whole identity card number, or on each dimension information of the embedded vector based on the first 6 bits of the identity card number.

For example, in an exemplary embodiment, when the target subject is a bank card number, statistics may be performed based on a maximum value, a minimum value, or an average value of each dimension information of the target subject, for example, a maximum value, a minimum value, or an average value of transaction amounts corresponding to each card number in a predetermined period of time may be counted, and a maximum value, a minimum value, or an average value of transaction amounts corresponding to each card BIN in a predetermined period of time may also be counted. In addition, the combination statistics operation may be performed on each dimension information, that is, the transaction behavior type dimension, the transaction amount dimension, the transaction behavior attribution dimension, and the like may be performed, for example, the total amount and the average value of the transaction amounts of the consumption type or the insurance type transaction types corresponding to each card number in a predetermined period of time may be counted, or the total amount and the average value of the transaction amounts of the consumption type or the insurance type transaction types corresponding to each card BIN may be counted.

In this exemplary embodiment, the main body variable information of the target main body, for example, the bank card and the identity card, is the first 6 bits of information, but the exemplary embodiment of the present application is not limited thereto, for example, the main body variable information of the bank card may also be the first 5 bits of the bank card number, and the main body variable information of the identity card may also be the first 9 bits of the identity card number, that is, the number of bits including the year of birth, as long as the main body variable information of the target main body can distinguish the attribution of the target main body, which is not particularly limited in the present application.

Further, in an example embodiment, the transaction risk of the target subject is predicted based on the statistics result of each dimension information of the embedded vector, for example, after the current transaction behavior data of the target subject is obtained, whether the maximum value of the transaction amount in each transaction behavior type dimension in the transaction behavior data exceeds the maximum value of the historical statistics result in the transaction behavior type dimension of the target subject is determined, if the maximum value is exceeded, it is determined that the transaction behavior of the target subject has risk, and risk reminding information is sent to the user. According to the technical scheme in the implementation of the example, even if the transaction data of a new main body such as a new card is very small, the transaction risk of the new main body can be determined as long as the corresponding dimension of the new main body such as the corresponding card BIN appears, so that the accuracy of risk assessment of the transaction behavior of the new main body is improved.

In addition, in an example embodiment, the risk type, that is, the high risk or the low risk, of the target subject may be further marked based on the transaction behavior data of the target subject in the preset period, the embedded vector corresponding to the target subject is divided into a positive sample and a negative sample based on the risk type of the marked target subject, the risk assessment model is trained based on the positive sample and the negative sample, the transaction risk of the current transaction behavior of the target subject is predicted based on the training result of the risk assessment model, and the risk assessment model may be a neural network model, a decision tree model or a support vector machine model, which is not particularly limited in this application.

According to the transaction information processing method in the exemplary embodiment of fig. 2, on one hand, the transaction behavior characteristics including the attribution information of the target subject are determined based on the transaction behavior sequence of the target subject, so that statistics on the transaction behavior characteristics of the target subject are facilitated based on the attribution information of the target subject; on the other hand, the dimension information based on the transaction behavior features generates the embedded vector corresponding to the transaction behavior features in a word embedding mode, so that the transaction behavior features of the target main body can be fully extracted in each dimension, the calculated amount can be reduced, and the data processing efficiency is improved; in still another aspect, statistics is performed on each dimension information of the embedded vector based on the subject variable information, and transaction risk of the new target subject can be predicted based on the statistical result, so that accuracy of risk assessment of transaction behaviors with unobvious abnormal features can be improved.

Fig. 3 illustrates a flow diagram for generating an embedded vector provided in accordance with some embodiments of the present application.

Referring to fig. 3, in step S310, each feature item of the transaction behavior feature of the target subject is clustered, and dimension information of the transaction behavior feature is determined based on the clustering result.

In an example embodiment, in the case that dimension information of a transaction behavior feature of a target subject is not determined, word segmentation processing may be performed on transaction behavior data in a behavior sequence of the target subject, words related to transaction risk recognition are extracted from the transaction behavior data in the behavior sequence after word segmentation, corresponding word vectors are generated, distances between the word vectors are calculated, class clusters of the words related to the transaction risk recognition in the transaction behavior data are determined based on the calculated distances, and the class clusters are used as dimension information of the transaction behavior feature. For example, the cluster of classes after clustering may include a principal variable class, a transaction amount class, a transaction type class, a transaction time class, etc., which are used as dimension information of transaction behavior characteristics.

It should be noted that, the distance between the respective word vectors may be a hamming distance, a euclidean distance, a cosine distance, but the distance in the exemplary embodiment of the present application is not limited thereto, and may be, for example, a mahalanobis distance, a manhattan distance, or the like.

In step S320, the transaction behavior feature is mapped to a corresponding embedded vector through a word embedding model based on the dimension information of the transaction behavior feature.

In an example embodiment, the word embedding model is an LSTM model, which is a deep learning model that may include an input layer, a hidden layer, and an output layer. The input layer is used for receiving the extracted transaction behavior characteristics and dimension information of the transaction behavior characteristics, the hidden layer is used for extracting corresponding dimension characteristics from the transaction behavior characteristics based on the dimension information of the transaction behavior characteristics, the output layer is used for outputting each dimension characteristic of the extracted transaction behavior characteristics to a preset full-connection layer, and an embedded vector corresponding to the transaction behavior characteristics is generated through the preset full-connection layer.

Furthermore, in an example embodiment, word embedding models are trained based on historical transaction behavior data of a target subject, and transaction behavior features are mapped to corresponding embedding vectors based on the trained word embedding models. For example, historical transaction behavior data of a plurality of target subjects within a preset time period can be obtained, corresponding transaction behavior features are extracted from the historical transaction behavior data of the plurality of target subjects, and word embedding models are trained based on the transaction behavior features of the target subjects and corresponding dimension information.

Fig. 4 is a flow chart illustrating a transaction information processing method according to further embodiments of the present application.

Referring to fig. 4, in step S410, a behavior sequence of a target subject is generated based on transaction behavior data of the target subject.

In an example embodiment, the target subject may be a bank card, the transaction behavior data of the target subject may be collection behavior data, and the behavior sequence of the target subject is collection sequence, i.e., collection event 1 to collection event t.

In step S420, the transaction behavioral characteristics of the target subject are extracted from the behavioral sequence of the target subject.

In an example embodiment, the transaction behavior feature of the target subject may be a feature of subject variable information, transaction amount, payee information or payer information, transaction location, transaction time, transaction type, etc. of the target subject, and may also include feature information that may also include other suitable transaction behavior features such as payment mode, coupon usage information, etc. In fig. 4, the characteristics of the target subject are represented by variables 1_1 to n_1, 1_2 to n_2, … …, 1_t to n_t, each representing a characteristic.

In step S430, an embedding vector corresponding to the transaction behavior feature is generated by word embedding based on the dimension information of the transaction behavior feature of the target subject.

In an example embodiment, the dimension information of the transaction behavioral characteristics of the target subject may include: one or more of transaction behavior attribution dimension information, transaction behavior type dimension information, transaction amount dimension information, and transaction time dimension information. For example, the transaction behavior attribution dimension information may include attribution dimension information such as bank card number information, IP address information, cell phone number information, identification card number information, and the like.

Further, word embedding is a distributed representation of words by which a high-dimensional word space can be mapped into a low-dimensional vector space. In an example embodiment, after the dimensional information of the transaction behavior feature is determined, the multi-dimensional transaction behavior feature may be mapped to a low-dimensional embedded vector corresponding to the dimensional information through a word embedding model, such as an LSTM (Long Short-Term Memory) model. For example, the dimension information that determines the transaction behavior feature includes four-dimensional information, namely transaction behavior attribution dimension information, transaction behavior type dimension information, transaction amount dimension information, and transaction time dimension information, and the transaction behavior feature may be mapped to the four-dimensional embedded vectors through a word embedding model.

The LSTM model is a deep learning model and may include an input layer, a hidden layer, and an output layer. The input layer is used for receiving the extracted transaction behavior characteristics and dimension information of the transaction behavior characteristics, the hidden layer is used for extracting corresponding dimension characteristics from the transaction behavior characteristics based on the dimension information of the transaction behavior characteristics, the output layer is used for outputting each dimension characteristic of the extracted transaction behavior characteristics to a preset full-connection layer, and an embedded vector corresponding to the transaction behavior characteristics is generated through the preset full-connection layer.

In step S440, statistics are performed on each dimension information of the embedded vector based on the subject variable information of the target subject.

Fig. 5 is a flow chart illustrating a transaction information processing method according to further embodiments of the present application.

Referring to fig. 5, in step S510, transaction behavior data of a target subject in a preset period of time is acquired, for example, the transaction behavior data of the target subject in 3 months is deposited and washed, noise data in the transaction behavior data is removed, and the transaction behavior data of the target subject is stored for storing the acquired transaction behavior data, for example, in the form of an offline data table.

In step S520, dimensional behavior characteristics corresponding to the target subject are determined based on the transaction behavior data, and the dimensional behavior characteristics DBF (Dimension Behavior Feature, dimensional behavior characteristics) are behavior characteristics obtained by integrating the detailed data or the behavior sequence of the transaction behavior in a certain dimension after processing and refining. For example, dimension information of the transaction behavior data may be obtained, and the transaction behavior data of the target subject may be mapped into corresponding dimension behavior features based on the dimension information.

In an example embodiment, step S520 may include steps S524 to S528, which are described in detail below.

In step S524, a behavior sequence of the target subject is generated based on the transaction behavior data of the target subject. The behavior sequence of the target subject is transaction behavior data of the target subject in a plurality of time intervals sequenced according to occurrence time of transactions, for example, if the behavior sequence of the target subject is generated based on the transaction behavior data of the target subject in a predetermined time period, for example, 3 months, the transaction behavior data in the predetermined time period can be divided based on a predetermined interval, for example, 1 day, so as to obtain the transaction behavior data in a plurality of time intervals, and the transaction behavior data in each time interval is sequenced according to occurrence time, so as to generate the behavior sequence in the predetermined time period corresponding to the target subject.

In step S526, an embedding vector corresponding to the behavior sequence of the target subject is generated by means of word embedding. Word embedding is a distributed representation of words by which a high-dimensional word space can be mapped into a low-dimensional vector space. In an example embodiment, after the dimensional information of the transaction behavioral characteristics is determined, the multi-dimensional transaction behavioral characteristics may be mapped to a low-dimensional embedded vector corresponding to the dimensional information by a word embedding model, such as an LSTM model. For example, the dimension information that determines the transaction behavior feature includes four-dimensional information, namely transaction behavior attribution dimension information, transaction behavior type dimension information, transaction amount dimension information, and transaction time dimension information, and the transaction behavior feature may be mapped to the four-dimensional embedded vectors through a word embedding model.

In step S528, dimensional behavior features corresponding to the target subject are determined based on the embedded vectors corresponding to the behavior sequences of the target subject. In an example embodiment, respective dimensional information of the embedded vector is counted based on subject variable information of the target subject, and dimensional behavior characteristics corresponding to the target subject are determined based on the counted result. When the target main body is a bank card number, the main body variable information of the target main body comprises bank card number information, the bank card number information comprises the attribution information of the target main body, namely, the card BIN information of the bank card, the dimension information of the embedded vector can be counted based on the whole information of the bank card number, namely, the whole card number, and the dimension information of the embedded vector can be counted based on the card BIN of the bank card. When the target subject is identity card information, the subject variable of the target subject contains identity card number information, the identity card number information contains attribution information of the target subject, namely address information represented by the first 6 bits of the identity card number, and statistics can be carried out on each dimension information of the embedded vector based on complete information of the identity card number, namely the whole identity card number, or on each dimension information of the embedded vector based on the first 6 bits of the identity card number.

Next, returning to step S530, in step S530, a risk assessment model is built based on the dimensional behavioral characteristics of the target subject. For example, the risk assessment model may be trained in combination with dimensional behavior characteristics of the target subject and conventional risk variables, which may be a neural network model, a decision tree model, or a support vector machine model, and the conventional risk variables may include standard deviation of transaction amount, average value of transaction amount, and abnormal behavior variables.

Further, the target subjects may be further clustered based on dimension behavior characteristics of the target subjects, for example, when the target subjects are bank cards, the target subjects may be clustered by the card BIN information of the target subjects, and if the card BIN information of the target subjects is the same, the target subjects are classified into a group. By grouping the target subjects, new subjects, such as new cards, can be divided into appropriate groups, so that the accuracy of risk assessment of transaction behavior of the new subjects can be improved.

Further, in step S540, a regular risk variable of the target subject is generated based on the transaction behavior data of the target subject, for example, the transaction behavior data of the target subject is counted, and a regular risk variable such as an average value, a standard deviation, a transaction probability, etc. of the transaction amount of the target subject on each transaction type is determined.

Next, in step 550, a model policy corresponding to the risk assessment model is determined. For different risk types, different wind control models such as anti-fraud models, anti-theft models, anti-cheating models and the like need to be built. Different models are adopted for different business types under each risk type, and an anti-fraud model is taken as an example, and can be further subdivided into an offline fraudster model, an online account transfer model, an online card transfer model and the like. For a service type, the model policy may adopt a parallel mode of multiple models, that is, a parallel mode of adding a certain rule to each model score threshold, for example: model strategy 1: model A score is greater than 0.6 and transaction amount is greater than 1000 yuan; model policy 2 is model B score greater than 0.7 and transaction amount greater than 10000 yuan. By adopting a plurality of models to evaluate the transaction risk of the target subject, the accuracy of risk evaluation can be further improved.

In step S560, the risk of the transaction of the target subject is evaluated based on the model policy in step S550 and the decision result is output. For example, after the output scores of different models are standardized, a comprehensive score is integrally output, the transaction risk of the target subject is evaluated based on the comprehensive score, the grade of the transaction risk is determined, and a final decision such as a risk prompt or a safety prompt is output.

Furthermore, risk decision logs of the target subjects, model output scores, risk grade labels and other risk data can be deposited, so that model strategies of a risk assessment model of the target subjects can be further adjusted based on the deposited risk data of the target subjects, and accuracy of risk assessment is improved.

In an example embodiment of the present application, a transaction information processing apparatus is also provided. Referring to fig. 6, the transaction information processing apparatus 600 includes: a sequence generation unit 610, a feature extraction unit 620, an embedded vector generation unit 630, and a statistics unit 640. Wherein the sequence generating unit 610 is configured to generate a behavior sequence of a target subject based on transaction behavior data of the target subject; the feature extraction unit 620 is configured to extract a transaction behavior feature of the target subject from a behavior sequence of the target subject, where the transaction behavior feature includes subject variable information of the transaction behavior data, and the subject variable information includes information indicating attribution information of the target subject; the embedded vector generating unit 630 is configured to generate an embedded vector corresponding to the transaction behavior feature by means of word embedding based on the dimension information of the transaction behavior feature; the statistics unit 640 is configured to perform statistics on each dimension information of the embedded vector based on the subject variable information, so as to predict a transaction risk of the target subject based on a statistical result.

In some embodiments of the present application, referring to fig. 7, based on the foregoing scheme, the embedded vector generating unit 630 includes: a dimension information determining unit 710, configured to perform clustering processing on each feature item of the transaction behavior feature, and determine dimension information of the transaction behavior feature based on a clustering result; and a mapping unit 720, configured to map the transaction behavior feature into a corresponding embedded vector through a word embedding model based on the dimension information of the transaction behavior feature.

In some embodiments of the present application, based on the foregoing solution, referring to fig. 8, the word embedding model is a long-short-term memory LSTM model, and the mapping unit 720 includes: an input unit 722 for inputting the transaction behavior feature and dimension information of the transaction behavior feature to the LSTM model; a feature extraction unit 724, configured to extract corresponding dimension features from the transaction behavior features through the hidden layer of the LSTM model based on the dimension information of the transaction behavior features; and the feature output unit 726 is configured to output each dimension feature of the extracted transaction behavior feature to a preset full-connection layer, and generate a corresponding embedded vector through the preset full-connection layer.

In some embodiments of the present application, based on the foregoing scheme, the transaction information processing apparatus 600 further includes: a history data acquisition unit configured to acquire history transaction behavior data of a plurality of target subjects; a behavior feature extraction unit for extracting transaction behavior features of the plurality of target subjects from the historical transaction behavior data; and the training unit is used for training the LSTM model based on the transaction behavior characteristics of the target subjects and the dimension information of the transaction behavior characteristics.

In some embodiments of the present application, based on the foregoing scheme, the statistics unit 640 includes: the first statistics unit is used for counting each dimension information of the embedded vector based on the complete information of the main variable information; and/or a second statistics unit, configured to perform statistics on each dimension information of the embedded vector based on the attribution information of the target subject included in the subject variable information.

In some embodiments of the present application, based on the foregoing scheme, the statistics unit 640 is configured to: and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

In some embodiments of the present application, based on the foregoing scheme, the statistics unit 640 includes: the dividing unit is used for dividing the target subject into a new account or an old account based on the subject variable information; the first prediction unit is used for predicting the transaction risk of the target subject based on a statistical result corresponding to the attribution information of the target subject contained in the subject variable information if the target subject is a new account; and the second prediction unit is used for predicting the transaction risk of the target subject based on the statistical result corresponding to the complete information of the subject variable information if the target subject is an old account.

In some embodiments of the present application, based on the foregoing scheme, the statistics unit 640 is configured to: acquiring statistical results of each dimension of the embedded vector corresponding to the main variable; and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

According to the transaction information processing apparatus in the example embodiment of fig. 6, on the one hand, the transaction behavior characteristics including the attribution information of the target subject are determined based on the transaction behavior sequence of the target subject, so that statistics on the transaction behavior characteristics of the target subject are facilitated based on the attribution information of the target subject; on the other hand, the dimension information based on the transaction behavior features generates the embedded vector corresponding to the transaction behavior features in a word embedding mode, so that the transaction behavior features of the target main body can be fully extracted in each dimension, the calculated amount can be reduced, and the data processing efficiency is improved; in still another aspect, statistics is performed on each dimension information of the embedded vector based on the subject variable information, and transaction risk of the new target subject can be predicted based on the statistical result, so that accuracy of risk assessment of transaction behaviors with unobvious abnormal features can be improved.

The transaction information processing device provided by the embodiment of the application can realize each process in the embodiment of the method and achieve the same functions and effects, and is not repeated here.

Further, the embodiment of the application also provides a transaction information processing device, as shown in fig. 9.

The transaction information processing device may be configured or configured to vary considerably, and may include one or more processors 901 and a memory 902, where the memory 902 may store one or more storage applications or data. Wherein the memory 902 may be transient storage or persistent storage. The application programs stored in the memory 902 may include one or more modules (not shown in the figures), each of which may include a series of computer executable instructions for use in a transaction information processing device. Still further, the processor 901 may be arranged to communicate with the memory 902 and execute a series of computer executable instructions in the memory 902 on the transaction information processing device. The transaction information processing device may also include one or more power supplies 903, one or more wired or wireless network interfaces 904, one or more input/output interfaces 905, one or more keyboards 906, and the like.

In a particular embodiment, a transaction information processing device includes a memory, and one or more programs, wherein the one or more programs are stored in the memory, and the one or more programs may include one or more modules, and each module may include a series of computer-executable instructions for the transaction information processing device, and execution of the one or more programs by one or more processors comprises computer-executable instructions for: generating a behavior sequence of a target subject based on transaction behavior data of the target subject; extracting transaction behavior characteristics of the target subject from a behavior sequence of the target subject, wherein the transaction behavior characteristics comprise subject variable information of the transaction behavior data, and the subject variable information comprises information representing attribution information of the target subject; generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature; and counting each dimension information of the embedded vector based on the subject variable information so as to predict the transaction risk of the target subject based on a counting result. .

Optionally, the computer executable instructions, when executed, generate an embedded vector corresponding to the transaction behavior feature by way of word embedding based on the dimensional information of the transaction behavior feature, including: clustering all feature items of the transaction behavior feature, and determining dimension information of the transaction behavior feature based on a clustering result; and mapping the transaction behavior characteristics into corresponding embedded vectors through a word embedding model based on the dimension information of the transaction behavior characteristics.

Optionally, the word embedding model is a long-short term memory LSTM model, and mapping the transaction behavior feature to a corresponding embedding vector through the word embedding model based on dimension information of the transaction behavior feature includes: inputting dimension information of the transaction behavior characteristics into the LSTM model; extracting corresponding dimension characteristics from the transaction behavior characteristics through a hidden layer of the LSTM model based on the dimension information of the transaction behavior characteristics; outputting each dimension characteristic of the extracted transaction behavior characteristic to a preset full-connection layer, and generating a corresponding embedded vector through the preset full-connection layer.

Optionally, the computer executable instructions, when executed, further comprise: acquiring historical transaction behavior data of a plurality of target subjects; extracting transaction behavior features of the plurality of target subjects from the historical transaction behavior data; training the LSTM model based on the transaction behavior features of the plurality of target subjects and the dimension information of the transaction behavior features.

Optionally, the computer executable instructions, when executed, perform statistics on respective dimension information of the embedded vector based on the subject variable information, including: counting each dimension information of the embedded vector based on the complete information of the main variable information; and/or counting each dimension information of the embedded vector based on the attribution information of the target subject contained in the subject variable information.

Optionally, when the computer executable instructions are executed, the principal variable information is bank account information, and counting each dimension information of the embedded vector based on the attribution information of the target principal contained in the principal variable information includes: and counting the dimension information of the embedded vector based on the bank identification code contained in the bank account information.

Optionally, the computer executable instructions, when executed, perform statistics on respective dimension information of the embedded vector, including: and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

Optionally, the computer executable instructions, when executed, predict a risk of transaction for the target subject based on the statistics, comprising: classifying the target subject into a new account or an old account based on the subject variable information; if the target subject is a new account, predicting transaction risk of the target subject based on a statistical result corresponding to the attribution information of the target subject contained in the subject variable information; and if the target subject is a old account, predicting the transaction risk of the target subject based on a statistical result corresponding to the complete information of the subject variable information.

Optionally, the computer executable instructions, when executed, predict a risk of transaction for the target subject based on the statistics, comprising: acquiring statistical results of each dimension of the embedded vector corresponding to the main variable; and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

The transaction information processing device provided by the embodiment of the present application can implement each process in the foregoing method embodiment, and achieve the same functions and effects, which are not repeated here.

In addition, the embodiment of the present application further provides a storage medium, which is configured to store computer executable instructions, and in a specific embodiment, the storage medium may be a usb disk, an optical disc, a hard disk, etc., where the computer executable instructions stored in the storage medium when executed by a processor can implement the following flow: generating a behavior sequence of a target subject based on transaction behavior data of the target subject; extracting transaction behavior characteristics of the target subject from a behavior sequence of the target subject, wherein the transaction behavior characteristics comprise subject variable information of the transaction behavior data, and the subject variable information comprises information representing attribution information of the target subject; generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature; and counting each dimension information of the embedded vector based on the subject variable information so as to predict the transaction risk of the target subject based on a counting result.

Optionally, the computer executable instructions stored on the storage medium, when executed by the processor, generate an embedded vector corresponding to the transaction behavior feature by word embedding based on the dimension information of the transaction behavior feature, including: clustering all feature items of the transaction behavior feature, and determining dimension information of the transaction behavior feature based on a clustering result; and mapping the transaction behavior characteristics into corresponding embedded vectors through a word embedding model based on the dimension information of the transaction behavior characteristics.

Optionally, the computer executable instructions stored on the storage medium, when executed by the processor, the word embedding model is a long-short term memory LSTM model, and mapping the transaction behavior feature into a corresponding embedding vector through the word embedding model based on dimension information of the transaction behavior feature, including: inputting dimension information of the transaction behavior characteristics into the LSTM model; extracting corresponding dimension characteristics from the transaction behavior characteristics through a hidden layer of the LSTM model based on the dimension information of the transaction behavior characteristics; outputting each dimension characteristic of the extracted transaction behavior characteristic to a preset full-connection layer, and generating a corresponding embedded vector through the preset full-connection layer.

Optionally, the storage medium stores computer executable instructions that when executed by the processor further comprise: acquiring historical transaction behavior data of a plurality of target subjects; extracting transaction behavior features of the plurality of target subjects from the historical transaction behavior data; training the LSTM model based on the transaction behavior features of the plurality of target subjects and the dimension information of the transaction behavior features.

Optionally, the computer executable instructions stored on the storage medium, when executed by the processor, perform statistics on the respective dimension information of the embedded vector based on the subject variable information, including: counting each dimension information of the embedded vector based on the complete information of the main variable information; and/or counting each dimension information of the embedded vector based on the attribution information of the target subject contained in the subject variable information.

Optionally, the computer executable instructions stored in the storage medium, when executed by the processor, make statistics on each dimension information of the embedded vector based on the attribution information of the target subject contained in the subject variable information, where the subject variable information is bank account information, including: and counting the dimension information of the embedded vector based on the bank identification code contained in the bank account information.

Optionally, the computer executable instructions stored on the storage medium, when executed by the processor, perform statistics on the respective dimension information of the embedded vector, including: and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

Optionally, the storage medium stores computer executable instructions that, when executed by the processor, predict transaction risk of the target subject based on statistics, comprising: classifying the target subject into a new account or an old account based on the subject variable information; if the target subject is a new account, predicting transaction risk of the target subject based on a statistical result corresponding to the attribution information of the target subject contained in the subject variable information; and if the target subject is a old account, predicting the transaction risk of the target subject based on a statistical result corresponding to the complete information of the subject variable information.

Optionally, the storage medium stores computer executable instructions that, when executed by the processor, predict transaction risk of the target subject based on statistics, comprising: acquiring statistical results of each dimension of the embedded vector corresponding to the main variable; and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

The computer readable storage medium provided in the embodiments of the present application can implement each process in the foregoing method embodiments and achieve the same functions and effects, and are not repeated here.

In the 90 s of the 20 th century, improvements to one technology could clearly be distinguished as improvements in hardware (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software (improvements to the process flow). However, with the development of technology, many improvements of the current method flows can be regarded as direct improvements of hardware circuit structures. Designers almost always obtain corresponding hardware circuit structures by programming improved method flows into hardware circuits. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device, PLD) (e.g., field programmable gate array (Field Programmable Gate Array, FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A designer programs to "integrate" a digital system onto a PLD without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, such programming is mostly implemented by using "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before the compiling is also written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), but not just one of the hdds, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of 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, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable logic controllers, and embedded microcontrollers, 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 of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in a pure computer readable program code, it is well possible to implement the same functionality by logically programming the method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Such a controller may thus be regarded as a kind of hardware component, and means for performing various functions included therein may also be regarded as structures within the hardware component. Or even means for achieving the various functions may be regarded as either software modules implementing the methods or structures within hardware components.

The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, 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 functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present application.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application 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 application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations 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 one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

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

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 storage media for a computer 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 magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

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 one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises the element.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application 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 application 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 application 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.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.

The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and changes may be made to the present application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. which are within the spirit and principles of the present application are intended to be included within the scope of the claims of the present application.

Claims

1. A transaction information processing method, comprising:

generating a behavior sequence of each subject based on transaction behavior data of the subject;

extracting transaction behavior characteristics of each subject from the behavior sequence of each subject, wherein the transaction behavior characteristics comprise subject variable information of the transaction behavior data, and the subject variable information comprises information representing attribution information of the subject;

Generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature;

counting each dimension information of the embedded vector based on the subject variable information of a target subject so as to predict the transaction risk of the target subject based on a counting result;

the statistics on each dimension information of the embedded vector based on the subject variable information of the target subject is performed to predict the transaction risk of the target subject based on the statistical result, and the method comprises the following steps:

classifying a target subject into a new account or an old account based on subject variable information of the target subject;

if the target main body is an old account, counting each dimension information of the embedded vector based on the complete information of the main body variable information of the target main body to obtain a first statistic result, and predicting the transaction risk of the target main body according to the first statistic result; the main variable information comprises a bank card number or an identity card number, and the complete information of the main variable information is the complete number information of the bank card number or the identity card number;

if the target subject is a new account, counting each dimension information of the embedded vector based on the attribution information of the target subject contained in the subject variable information of the target subject to obtain a second statistical result, and predicting the transaction risk of the target subject according to the second statistical result; the attribution information is part number information which can distinguish attribution of the main body in the bank card number or the identity card number.

2. The transaction information processing method according to claim 1, wherein generating an embedding vector corresponding to the transaction behavior feature by means of word embedding based on dimension information of the transaction behavior feature includes:

clustering all feature items of the transaction behavior feature, and determining dimension information of the transaction behavior feature based on a clustering result;

and mapping the transaction behavior characteristics into corresponding embedded vectors through a word embedding model based on the dimension information of the transaction behavior characteristics.

3. The transaction information processing method according to claim 2, wherein the word embedding model is a long-short-term memory LSTM model, and mapping the transaction behavior feature into a corresponding embedding vector through the word embedding model based on dimension information of the transaction behavior feature includes:

inputting dimension information of the transaction behavior characteristics into the LSTM model;

extracting corresponding dimension characteristics from the transaction behavior characteristics through a hidden layer of the LSTM model based on the dimension information of the transaction behavior characteristics;

outputting each dimension characteristic of the extracted transaction behavior characteristic to a preset full-connection layer, and generating a corresponding embedded vector through the preset full-connection layer.

4. The transaction information processing method according to claim 3, further comprising:

acquiring historical transaction behavior data of a plurality of subjects;

extracting transaction behavior features of the plurality of subjects from the historical transaction behavior data;

training the LSTM model based on the transaction behavior features of the plurality of target subjects and the dimension information of the transaction behavior features.

5. The transaction information processing method according to claim 4, wherein the subject variable information is bank account information, and counting each dimension information of the embedded vector based on the home information of the target subject included in the subject variable information includes:

and counting the dimension information of the embedded vector based on the bank identification code contained in the bank account information.

6. The transaction information processing method according to claim 4, wherein counting each dimension information of the embedded vector includes:

and counting the maximum value, the minimum value or the average value of each dimension information of the embedded vector.

7. The transaction information processing method according to any one of claims 1 to 6, characterized in that predicting transaction risk of the target subject based on a statistical result includes:

Acquiring statistical results of each dimension of the embedded vector corresponding to the main variable;

and predicting the transaction risk of the target subject through a decision tree model based on the statistical result.

8. A transaction information processing apparatus, comprising:

a sequence generating unit for generating a behavior sequence of each subject based on transaction behavior data of each subject;

a feature extraction unit, configured to extract a transaction behavior feature of each subject from a behavior sequence of each subject, where the transaction behavior feature includes subject variable information of the transaction behavior data, and the subject variable information includes information indicating attribution information of the subject;

the embedded vector generation unit is used for generating an embedded vector corresponding to the transaction behavior feature in a word embedding mode based on the dimension information of the transaction behavior feature;

the statistics unit is used for counting each dimension information of the embedded vector based on the subject variable information of the target subject so as to predict the transaction risk of the target subject based on a counting result;

9. A transaction information processing apparatus, characterized by comprising: a processor; and a memory configured to store computer-executable instructions that, when executed, cause the processor to implement the steps of the transaction information processing method of any of the preceding claims 1 to 7.

10. A storage medium storing computer executable instructions which when executed implement the steps of the transaction information processing method of any one of the preceding claims 1 to 7.