Label generation method and device based on artificial intelligence, computer equipment and medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a label generation method and device based on artificial intelligence, computer equipment and a storage medium, and the method comprises the steps: constructing a view based on account transaction flow data; identifying the account transaction flow data to obtain a first fund collection transaction; constructing a collection label set based on the first fund collection transaction; screening a specified tag set matched with the specified keyword from the collection tag set, and screening a second fund collection transaction based on the specified tag set; screening a suspected fund collection transaction from the account transaction flow data based on a transaction client view, and combining the second fund collection transaction with the suspected fund collection transaction to obtain a third fund collection transaction; performing transaction analysis on the third fund collection transaction based on a transaction judgment model to obtain a fourth fund collection transaction; and merging and labeling the first fund collection transaction and the fourth fund collection transaction to obtain a target transaction label. According to the invention, the identification accuracy of the fund collection transaction label is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence development technology and the field of financial technology, and in particular to artificial intelligence-based label generation methods, devices, computer equipment and storage media. Background Art

[0002] In today's business environment, bank transaction flow data has become an indispensable and important record of corporate financial activities. These data not only record the flow of corporate funds in detail, but also contain rich information, which has extremely high application value for corporate operational decisions, financial health assessments, and bank financial service optimization. Fund collection transactions, as a key type of bank transaction flow, specifically refer to the behavior of enterprises automatically or manually collecting funds in different bank accounts to a designated account according to preset rules or conditions. For banks, such transactions are not only an important indicator of customer fund flow, but also an important basis for banks to build customer profiles, customize financial service products, and optimize risk control strategies.

[0003] At present, there are many technical solutions for mining and labeling fund collection transactions on the market, among which the most common is the rule-based mining method. This type of method usually relies on manually set rule sets to identify fund collection transactions by matching and screening the characteristics of transaction data. However, with the continuous expansion of banking business and the increasing diversification of customer needs, the form and characteristics of fund collection transactions are also constantly changing. Traditional rule models are often difficult to adapt to such rapid changes, resulting in recognition failures or missed recognition when facing new types of fund collection transactions. In addition, the update and optimization of rule models usually require manual intervention, which not only increases operating costs, but may also lead to untimely model updates, further reducing the accuracy of fund collection transaction label recognition. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose an artificial intelligence-based label generation method, device, computer equipment and storage medium to solve the technical problem of low recognition accuracy in the existing rule-based model-based fund collection transaction label recognition method.

[0005] In order to solve the above technical problems, the embodiment of the present application provides a label generation method based on artificial intelligence, which adopts the following technical solution:

[0006] Collecting account transaction flow data in each subsystem, and constructing corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views;

[0007] Based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction;

[0008] Constructing a collection tag set corresponding to the first fund collection transaction;

[0009] Screening out a designated tag set matching a designated keyword from the collection tag set, and screening out a second fund collection transaction corresponding to the designated tag set;

[0010] Screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merging the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction;

[0011] Performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction;

[0012] Merging the first fund collection transaction and the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction;

[0013] The fifth fund collection transaction is labeled to obtain a corresponding target transaction label.

[0014] Furthermore, the step of constructing a corresponding view based on the account transaction flow data specifically includes:

[0015] Cleaning the account transaction flow data to obtain the corresponding designated account transaction flow data;

[0016] Call the preset transaction dimension;

[0017] Based on a preset data weaving strategy, the transaction flow data of the designated account is organized and processed according to the transaction dimension to obtain a view corresponding to the transaction dimension.

[0018] Furthermore, the step of constructing a collection tag set corresponding to the first fund collection transaction specifically includes:

[0019] Obtaining transaction content in the first fund collection transaction;

[0020] Segmenting the transaction content to obtain corresponding segmentation results;

[0021] Performing part-of-speech tagging on the word segmentation result to obtain a corresponding part-of-speech tagging result;

[0022] Based on the part-of-speech tagging result, a preset business tagging rule is called to tag the transaction content to obtain the collection tag set.

[0023] Furthermore, the step of filtering out a specified tag set matching a specified keyword from the collected tag set specifically includes:

[0024] Call the preset similarity algorithm;

[0025] Calculate the similarity between each tag in the aggregated tag set and the specified keyword based on the similarity algorithm;

[0026] Filter out a target similarity greater than a preset similarity threshold from all the similarities;

[0027] Filtering out a target tag corresponding to the target similarity from the collected tag set;

[0028] The specified tag set is generated based on the target tag.

[0029] Furthermore, the step of screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view specifically includes:

[0030] Based on the transaction customer view, query the counterparty information corresponding to the specified transaction; wherein the specified transaction is any one of all transactions included in the account transaction flow data;

[0031] Call the preset recognition rules;

[0032] Analyze the counterparty information based on the identification rules to determine whether the counterparty information meets the conditions for fund pooling transactions;

[0033] If so, the designated transaction is marked as a suspected fund collection transaction; otherwise, the designated transaction is marked as a non-fund collection transaction.

[0034] Furthermore, the view also includes a transaction account view and a transaction institution view; and before the step of performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction, it also includes:

[0035] Constructing corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view;

[0036] Dividing the transaction sample data into a training set and a test set;

[0037] Call the preset deep learning model;

[0038] Based on a preset training strategy, the deep learning model is trained using the training set to obtain a trained specified model;

[0039] Validating the specified model based on the test set;

[0040] If the specified model passes the verification, the specified model is used as the transaction determination model.

[0041] Furthermore, the step of constructing corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view specifically includes:

[0042] Collecting data on the transaction account view, the transaction institution view, and the transaction customer view to obtain corresponding initial data;

[0043] Performing data cleaning on the initial data to obtain corresponding processed data;

[0044] Performing feature extraction on the processed data to obtain corresponding feature data;

[0045] Preprocessing the feature data to obtain processed feature data;

[0046] The processed feature data is used as the transaction sample data.

[0047] In order to solve the above technical problems, the embodiment of the present application further provides a label generation device based on artificial intelligence, which adopts the following technical solution:

[0048] A first construction module is used to collect account transaction flow data in each subsystem and construct corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views;

[0049] an identification module, configured to identify a fund collection transaction on the account transaction flow data based on a preset fund collection identification model to obtain a corresponding first fund collection transaction;

[0050] A second building module, used to build a collection label set corresponding to the first fund collection transaction;

[0051] A first screening module, used to screen out a designated tag set matching a designated keyword from the collection tag set, and screen out a second fund collection transaction corresponding to the designated tag set;

[0052] a second screening module, configured to screen out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merge the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction;

[0053] An analysis module, configured to perform transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction;

[0054] a merging module, configured to merge the first fund collection transaction with the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction;

[0055] A generation module is used to perform labeling processing on the fifth fund collection transaction to obtain a corresponding target transaction label.

[0056] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0057] Collecting account transaction flow data in each subsystem, and constructing corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views;

[0058] Based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction;

[0059] Constructing a collection tag set corresponding to the first fund collection transaction;

[0060] Screening out a designated tag set matching a designated keyword from the collection tag set, and screening out a second fund collection transaction corresponding to the designated tag set;

[0061] Screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merging the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction;

[0062] Performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction;

[0063] Merging the first fund collection transaction and the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction;

[0064] The fifth fund collection transaction is labeled to obtain a corresponding target transaction label.

[0065] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0066] Collecting account transaction flow data in each subsystem, and constructing corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views;

[0067] Based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction;

[0068] Constructing a collection tag set corresponding to the first fund collection transaction;

[0069] Screening out a designated tag set matching a designated keyword from the collection tag set, and screening out a second fund collection transaction corresponding to the designated tag set;

[0070] Screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merging the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction;

[0071] Performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction;

[0072] Merging the first fund collection transaction and the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction;

[0073] The fifth fund collection transaction is labeled to obtain a corresponding target transaction label.

[0074] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0075] The present application first collects account transaction flow data in each subsystem, and constructs a corresponding view based on the account transaction flow data; wherein the view at least includes a transaction customer view; then, based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction; then, a collection tag set corresponding to the first fund collection transaction is constructed; a specified tag set matching a specified keyword is screened out from the collection tag set, and a second fund collection transaction corresponding to the specified tag set is screened out; subsequently, suspected fund collection transactions are screened out from the account transaction flow data based on the transaction customer view, and the second fund collection transaction is merged with the suspected fund collection transaction to obtain a corresponding third fund collection transaction; further, based on a preset transaction determination model, a transaction analysis is performed on the third fund collection transaction to obtain a corresponding fourth fund collection transaction; and the first fund collection transaction is merged with the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction; finally, the fifth fund collection transaction is labeled to obtain a corresponding target transaction tag. By combining the use of a funds collection identification model with a transaction determination model, the present application can achieve multi-dimensional and multi-level funds collection transaction identification for account transaction flow data, effectively reduce the occurrence of identification failures or missed identifications, and greatly improve the identification accuracy of funds collection transaction labels. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0077] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0078] Figure 2 A flowchart of an embodiment of a label generation method based on artificial intelligence according to the present application;

[0079] Figure 3 is a structural schematic diagram of an embodiment of an artificial intelligence-based label generation device according to the present application;

[0080] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0081] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0082] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0083] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0084] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0085] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0086] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0087] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0088] It should be noted that the artificial intelligence-based label generation method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the artificial intelligence-based label generation device is generally set in the server / terminal device.

[0089] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0090] Continue to refer Figure 2 , showing a flowchart of an embodiment of an artificial intelligence-based label generation method according to the present application. According to different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted. The artificial intelligence-based label generation method provided in the embodiment of the present application can be applied to any scenario where label generation is required, and the artificial intelligence-based label generation method can be applied to products in these scenarios, for example, label generation in the financial insurance field. The artificial intelligence-based label generation method comprises the following steps:

[0091] Step S201, collecting account transaction flow data in each subsystem, and constructing corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views.

[0092] In this embodiment, the electronic device (e.g. Figure 1The server / terminal device shown in the figure) can obtain account transaction flow data through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future. The execution subject of this application is a label generation system, or simply referred to as a system. The above-mentioned subsystems may be financial institutions, such as various core subsystems within a bank (such as core business systems, payment systems, credit systems, etc.). Among them, the account transaction flow data in each subsystem can be collected regularly or in real time by utilizing an API interface or a database connection. The account transaction flow data may include at least key fields such as transaction date, transaction time, transaction type, transaction amount, counterparty information, account information, etc. In addition, the specific implementation process of constructing the corresponding view based on the account transaction flow data will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here.

[0093] Step S202: identifying a funds collection transaction on the account transaction flow data based on a preset funds collection identification model to obtain a corresponding first funds collection transaction.

[0094] In this embodiment, the above-mentioned fund collection identification model refers to a model that trains and analyzes fund transaction data through a classification algorithm of machine learning (ML) technology to identify specific patterns or behaviors such as fund collection. Specifically, a suitable machine learning classification algorithm (such as logistic regression, decision tree, random forest, etc.) can be selected. Then, the historical transaction data is used as a training set to train the fund collection identification model. During the model training process, it is necessary to continuously adjust the parameters to improve the accuracy and generalization ability of the model. Subsequently, by inputting the newly collected account transaction flow data into the trained fund collection identification model, preliminary fund collection transaction identification is performed, and the identification result output by the fund collection identification is the preliminary fund collection transaction, that is, the first fund collection transaction mentioned above.

[0095] Among them, the above view may also include a transaction flow view, and a series of expert rules may be sorted out according to the characteristics of banking business, the characteristics of fund collection transactions and the transaction flow view included in the above view. Expert rules may include conditions such as transaction amount threshold, counterparty type, transaction frequency, etc. The expert rules can also be used to optimize the fund collection identification model to improve the accuracy and efficiency of the model's fund collection identification. Specifically, expert rules can provide a useful supplement for training fund collection identification models. For example, in the early stage of model training, expert rules can be used to generate some initial training data or features, so as to help the model learn the characteristics of behaviors such as fund collection more quickly. In addition, expert rules can also be used to explain and verify the prediction results of the model. When the model predicts that a fund transaction has a collection risk, the expert rules can further analyze the specific circumstances and background of the transaction to verify whether the prediction results of the model are accurate.

[0096] Step S203: construct a collection tag set corresponding to the first fund collection transaction.

[0097] In this embodiment, the specific implementation process of constructing the collection label set corresponding to the first fund collection transaction will be further described in detail in subsequent specific embodiments of the present application, and will not be elaborated on here.

[0098] Step S204: Filter out a designated tag set that matches a designated keyword from the collection tag set, and filter out a second fund collection transaction corresponding to the designated tag set.

[0099] In this embodiment, the specific implementation process of selecting the specified tag set that matches the specified keyword from the collection tag set will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here. In addition, by using the screening conditions or features defined in the specified tag set, the transactions in the account transaction flow data are screened, such as operations involving querying, filtering and matching the transaction data, and then obtaining transaction records that meet the definition of the specified tag set from the screening process. These transaction records are marked as similar fund collection transactions, that is, the second fund collection transaction mentioned above.

[0100] Step S205: screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merging the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction.

[0101] In this embodiment, the specific implementation process of screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here. In addition, the union of the second fund collection transaction and the suspected fund collection transaction can be obtained and named as the third fund collection transaction.

[0102] Step S206: performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction.

[0103] In this embodiment, the third fund collection transaction can be input into the transaction determination model, so that the third fund collection transaction can be analyzed by the transaction determination model, and the corresponding fourth fund collection transaction can be output. The construction process of the above transaction determination model will be further described in detail in the subsequent specific embodiments, and will not be elaborated on here.

[0104] Step S207: merging the first fund collection transaction and the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction.

[0105] In this embodiment, the union of the first fund collection transaction and the fourth fund collection transaction may be obtained and named as the fifth fund collection transaction.

[0106] Step S208, labeling the fifth fund collection transaction to obtain a corresponding target transaction label.

[0107] In this embodiment, the generation of the final transaction label can be completed by marking the generated fifth fund collection transaction with a fund collection transaction label.

[0108] The present application first collects account transaction flow data in each subsystem, and constructs a corresponding view based on the account transaction flow data; wherein the view at least includes a transaction customer view; then, based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction; then, a collection tag set corresponding to the first fund collection transaction is constructed; a specified tag set matching a specified keyword is screened out from the collection tag set, and a second fund collection transaction corresponding to the specified tag set is screened out; subsequently, suspected fund collection transactions are screened out from the account transaction flow data based on the transaction customer view, and the second fund collection transaction is merged with the suspected fund collection transaction to obtain a corresponding third fund collection transaction; further, based on a preset transaction determination model, a transaction analysis is performed on the third fund collection transaction to obtain a corresponding fourth fund collection transaction; and the first fund collection transaction is merged with the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction; finally, the fifth fund collection transaction is labeled to obtain a corresponding target transaction tag. By combining the use of a funds collection identification model with a transaction determination model, the present application can achieve multi-dimensional and multi-level funds collection transaction identification for account transaction flow data, effectively reduce the occurrence of identification failures or missed identifications, and greatly improve the identification accuracy of funds collection transaction labels.

[0109] In some optional implementations, the step of constructing a corresponding view based on the account transaction flow data in step S201 includes the following steps:

[0110] The account transaction flow data is cleaned to obtain the corresponding designated account transaction flow data.

[0111] In this embodiment, the cleaning process includes removing duplicate, invalid or abnormal data in the account transaction flow data, thereby obtaining the transaction flow data of the designated account.

[0112] Call the preset transaction dimension.

[0113] In this embodiment, the above-mentioned transaction dimensions may at least include the transaction flow dimension, and may also include dimensions such as transaction accounts, transaction institutions, and transaction customers.

[0114] Based on a preset data weaving strategy, the transaction flow data of the designated account is organized and processed according to the transaction dimension to obtain a view corresponding to the transaction dimension.

[0115] In this embodiment, the above-mentioned data weaving strategy refers to a strategy for data organization and processing using data weaving calculations. By using data weaving technology, the cleaned transaction flow data of the designated account can be organized according to dimensions such as transaction flow, transaction account, transaction institution, and transaction customer, and views can be established respectively. Among them, the view may include a transaction flow view, a transaction account view, a transaction institution view, and a transaction customer view. In addition, data weaving technology involves operations such as data mapping, data conversion, and data aggregation to ensure the consistency and accuracy of the data.

[0116] The present application cleans the account transaction flow data to obtain the corresponding designated account transaction flow data; then calls the preset transaction dimension; subsequently, based on the preset data weaving strategy, the designated account transaction flow data is organized and processed according to the transaction dimension to obtain a view corresponding to the transaction dimension. The present application cleans the account transaction flow data to obtain the designated account transaction flow data, calls the transaction dimension, and then, based on the use of the data weaving strategy, organizes and processes the designated account transaction flow data according to the transaction dimension, thereby realizing the rapid and accurate construction of a view corresponding to the transaction dimension, improving the efficiency and intelligence of view construction.

[0117] In some optional implementations of this embodiment, step S203 includes the following steps:

[0118] Obtain transaction content in the first fund collection transaction.

[0119] In this embodiment, the transaction content may include a summary name, transaction remarks or transaction remarks in the first fund collection transaction.

[0120] The transaction content is segmented to obtain corresponding segmentation results.

[0121] In this embodiment, the transaction content may be segmented by using a natural language processing tool to obtain corresponding segmentation results.

[0122] Perform part-of-speech tagging on the word segmentation result to obtain a corresponding part-of-speech tagging result.

[0123] In this embodiment, the word segmentation results can be tagged with parts of speech by utilizing the part-of-speech tagging function of a natural language processing tool to identify parts of speech such as nouns, verbs, and adjectives, and obtain corresponding part-of-speech tagging results.

[0124] Based on the part-of-speech tagging result, a preset business tagging rule is called to tag the transaction content to obtain the collection tag set.

[0125] In this embodiment, after obtaining the part-of-speech tagging results, the transaction content can be further tagged in combination with the preset business tagging rules to obtain the corresponding collection tag set. Among them, the business tagging rules can be constructed according to the actual tagging needs, for example, they may include: if the transaction description contains verbs such as "transfer" and "payment" and involves an amount, it may be a payment transaction. If the transaction description contains nouns such as "rent" and "salary", it may be a specific type of transaction. If the transaction description contains prepositional phrases such as "for" and "purpose", it may be possible to further refine the transaction type.

[0126] The present application obtains a segmentation result by segmenting the transaction content in the first fund collection transaction, then performs part-of-speech tagging on the segmentation result to obtain a part-of-speech tagging result, and then labels the transaction content based on the part-of-speech tagging result and the use of business tagging rules, thereby achieving rapid and accurate construction of a collection label set corresponding to the first fund collection transaction, improving the construction efficiency of the collection label set and ensuring the data accuracy of the obtained collection label set.

[0127] In some optional implementations, the step of filtering out a specified tag set matching a specified keyword from the collected tag set in step S204 includes the following steps:

[0128] Call the preset similarity algorithm.

[0129] In this embodiment, the similarity algorithm may specifically adopt a cosine similarity algorithm.

[0130] The similarity between each tag in the aggregated tag set and the designated keyword is calculated based on the similarity algorithm.

[0131] In this embodiment, the above-mentioned designated keyword is specifically “fund collection.” The cosine similarity algorithm evaluates the similarity between two vectors by calculating the cosine value of the angle between them, thereby being able to calculate the similarity between each tag in the collection tag set and the designated keyword.

[0132] A target similarity greater than a preset similarity threshold is selected from all the similarities.

[0133] In this embodiment, there is no specific limitation on the value of the similarity threshold, which can be set according to actual business needs, for example, it can be set to 0.88.

[0134] Filter out a target tag corresponding to the target similarity from the collected tag set.

[0135] In this embodiment, the target tag refers to a tag whose similarity with the designated keyword is greater than the similarity threshold.

[0136] The specified tag set is generated based on the target tag.

[0137] In this embodiment, all target labels are combined to obtain a corresponding designated label set.

[0138] The present application calls a preset similarity algorithm; then calculates the similarity between each tag in the collected tag set and the specified keyword based on the similarity algorithm; then filters out the target similarity greater than the preset similarity threshold from all the similarities; subsequently filters out the target tag corresponding to the target similarity from the collected tag set, and generates the specified tag set based on the target tag. The present application calculates the similarity between each tag in the collected tag set and the specified keyword based on the use of a similarity algorithm; then filters out the target tag corresponding to the target similarity greater than the preset similarity threshold from the collected tag set, and then can quickly and accurately generate the specified tag set based on the target tag, thereby improving the generation efficiency of the specified tag set and ensuring the data accuracy of the obtained specified tag set.

[0139] In some optional implementations, the screening out of suspected fund collection transactions from the account transaction flow data based on the transaction customer view in step S205 includes the following steps:

[0140] Based on the transaction customer view, the counterparty information corresponding to the designated transaction is queried; wherein the designated transaction is any one of all transactions included in the account transaction flow data.

[0141] In this embodiment, the generated transaction customer view includes all necessary counterparty information, such as company name, company type, equity relationship, etc.

[0142] Call the preset recognition rules.

[0143] In this embodiment, according to business needs, identification rules for realizing the identification function of suspected fund collection transactions are pre-built. Specifically, first clarify which types of counterparties are regarded as objects of suspected fund collection transactions. In this embodiment, these objects include holding subsidiaries, actual controllers of enterprises, shareholders of enterprises, corporate legal persons, etc. Then, for each type of counterparty, specific identification rules are formulated. For example, for holding subsidiaries, it can be determined by querying the equity relationship; for actual controllers of enterprises, it may be necessary to query the enterprise registration information or related announcements.

[0144] The counterparty information is analyzed based on the identification rules to determine whether the counterparty information meets the conditions for fund pooling transactions.

[0145] In this embodiment, for each transaction (designated transaction) included in the account transaction flow data, after querying the counterparty information of the formulated transaction through the transaction customer view, the counterparty information can be judged according to the defined identification rules. If the counterparty meets the conditions of a suspected fund collection transaction (such as being a holding subsidiary, actual controller of the enterprise, etc.), the designated transaction is marked as a suspected fund collection transaction; otherwise, the designated transaction is marked as a non-fund collection transaction.

[0146] If so, the designated transaction is marked as a suspected fund collection transaction; otherwise, the designated transaction is marked as a non-fund collection transaction.

[0147] In this embodiment, all identified possible suspected fund collection transactions can be aggregated to form a transaction set, and the transaction set is defined as the suspected fund collection transaction.

[0148] This application queries the counterparty information corresponding to the designated transaction based on the transaction client view; then calls the preset identification rules; subsequently analyzes the counterparty information based on the identification rules to determine whether the counterparty information meets the conditions of the fund collection transaction; if so, marks the designated transaction as a suspected fund collection transaction, otherwise marks the designated transaction as a non-fund collection transaction. This application queries the counterparty information corresponding to the designated transaction based on the transaction client view, and then analyzes the counterparty information based on the use of identification rules, so that the identification and processing of suspected fund collection transactions can be realized quickly and accurately, which improves the identification and processing of suspected fund collection transactions and ensures the accuracy of the data obtained for suspected fund collection transactions.

[0149] In some optional implementations of this embodiment, the view further includes a transaction account view and a transaction institution view; before step S206, the electronic device may further perform the following steps:

[0150] Corresponding transaction sample data is constructed based on the transaction account view, the transaction institution view and the transaction customer view.

[0151] In this embodiment, the specific implementation process of constructing the corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view will be described in further detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0152] The transaction sample data is divided into a training set and a test set.

[0153] In this embodiment, the transaction sample data can be divided into a training set and a test set according to a preset division ratio, wherein the selection of the division ratio is not specifically limited, for example, it can be set to 7:3.

[0154] Call the preset deep learning model.

[0155] In this embodiment, the above-mentioned deep learning model can specifically adopt an LSTM deep learning model. The model construction of the deep learning model includes: selecting an LSTM network structure, including the settings of an input layer, a hidden layer, and an output layer. Then, the hyperparameters of the model are determined according to the requirements, such as a learning rate, a batch size, and the number of iterations.

[0156] Based on a preset training strategy, the deep learning model is trained using the training set to obtain a trained specified model.

[0157] In this embodiment, the training set is input into the deep learning model. The model is then trained using a corresponding loss function (such as cross entropy loss) and an optimization algorithm (such as Adam optimizer). During the training process, the performance of the model is evaluated by monitoring the change of the loss function and the accuracy of the validation set, thereby obtaining a trained specified model.

[0158] The specified model is validated based on the test set.

[0159] In this embodiment, the trained specified model can be verified by using the test set to ensure the generalization ability of the model and obtain corresponding verification results. If the performance indicators of the generated specified model meet the preset performance requirements, the specified model is determined to have passed the verification, otherwise the specified model is determined to have failed the verification.

[0160] If the specified model passes the verification, the specified model is used as the transaction determination model.

[0161] In this embodiment, if the specified model fails the verification, the structure and hyperparameters of the model are adjusted according to the verification results to improve the performance of the model, until a model whose performance indicators meet the preset performance requirements is finally obtained and used as the above-mentioned transaction determination model.

[0162] This application constructs corresponding transaction sample data based on the transaction account view, the transaction institution view, and the transaction customer view; then divides the transaction sample data into a training set and a test set; then calls a preset deep learning model; subsequently, based on a preset training strategy, uses the training set to train the deep learning model to obtain a trained designated model; further verifies the designated model based on the test set; if the designated model passes the verification, the designated model is used as the transaction determination model. This application trains and verifies the deep learning model based on the use of training sets and test sets, so that a transaction determination model that meets the requirements can be constructed quickly and accurately, thereby improving the construction efficiency of the transaction determination model and ensuring the model effect of the obtained transaction determination model.

[0163] In some optional implementations of this embodiment, the constructing corresponding transaction sample data based on the transaction account view, the transaction institution view, and the transaction customer view includes the following steps:

[0164] Data is collected on the transaction account view, the transaction institution view, and the transaction customer view to obtain corresponding initial data.

[0165] In this embodiment, the account information of the account transaction flow data, such as account balance, transaction history, etc., can be obtained from the transaction account view, the institution information related to the account transaction flow data, such as institution type, geographical location, business scope, etc., can be obtained from the transaction institution view, and the customer information of the account transaction flow data, such as customer type, credit rating, historical transaction behavior, etc., can be obtained from the transaction customer view. Then, the obtained account information, institution information and customer information are integrated to obtain the above initial data.

[0166] The initial data is cleaned to obtain corresponding processed data.

[0167] In this embodiment, the data cleaning may include removing duplicate data, invalid data and abnormal data, then filling or deleting missing data, and standardizing the data to ensure comparability between different features.

[0168] Feature extraction is performed on the processed data to obtain corresponding feature data.

[0169] In this embodiment, according to business needs and model requirements, feature data that have a significant impact on the determination of fund collection transactions are extracted from the above-obtained processed data. These feature data may include transaction amount, transaction frequency, counterparty type, transaction time, etc.

[0170] The feature data is preprocessed to obtain processed feature data.

[0171] In this embodiment, the above-mentioned preprocessing may include smoothing processing, normalization processing, etc. of the time series data, so as to convert the feature data into a format suitable for LSTM model input, such as time series data.

[0172] The processed feature data is used as the transaction sample data.

[0173] The present application obtains the corresponding initial data by collecting data from the transaction account view, the transaction institution view and the transaction customer view, and then performs data cleaning, feature extraction and preprocessing on the initial data, so as to quickly and accurately construct the required transaction sample data, improve the construction efficiency of the transaction sample data, and ensure the data accuracy of the obtained transaction sample data.

[0174] In some optional implementations, the user information obtained is subject to the user's consent and complies with relevant laws and policies.

[0175] In addition, the non-Company software tools or components that appear in the embodiments of the present application are merely examples and do not represent actual use.

[0176] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0177] It should be emphasized that in order to further ensure the privacy and security of the above-mentioned target transaction tag, the above-mentioned target transaction tag can also be stored in a node of a blockchain.

[0178] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0179] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0180] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0181] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0182] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0183] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a label generation device based on artificial intelligence, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0184] like Figure 3 As shown, the artificial intelligence-based label generation device 300 described in this embodiment includes: a first construction module 301, an identification module 302, a second construction module 303, a first screening module 304, a second screening module 305, an analysis module 306, a merging module 307 and a generation module 308. Among them:

[0185] The first construction module 301 is used to collect account transaction flow data in each subsystem and construct corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views;

[0186] An identification module 302 is used to identify a fund collection transaction on the account transaction flow data based on a preset fund collection identification model to obtain a corresponding first fund collection transaction;

[0187] A second construction module 303 is used to construct a collection label set corresponding to the first fund collection transaction;

[0188] A first screening module 304 is used to screen out a designated tag set matching a designated keyword from the collection tag set, and screen out a second fund collection transaction corresponding to the designated tag set;

[0189] A second screening module 305 is used to screen out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merge the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction;

[0190] An analysis module 306, configured to perform transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction;

[0191] A merging module 307, configured to merge the first fund collection transaction with the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction;

[0192] The generating module 308 is used to perform labeling processing on the fifth fund collection transaction to obtain a corresponding target transaction label.

[0193] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0194] In some optional implementations of this embodiment, the first building module 301 includes:

[0195] A first processing submodule is used to clean the account transaction flow data to obtain the corresponding designated account transaction flow data;

[0196] The first calling submodule is used to call the preset transaction dimension;

[0197] The second processing submodule is used to organize and process the transaction flow data of the designated account according to the transaction dimension based on a preset data weaving strategy to obtain a view corresponding to the transaction dimension.

[0198] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0199] In some optional implementations of this embodiment, the second building module 303 includes:

[0200] An acquisition submodule, used to acquire the transaction content in the first fund collection transaction;

[0201] A word segmentation submodule, used to segment the transaction content to obtain corresponding word segmentation results;

[0202] The tagging submodule is used to perform part-of-speech tagging on the word segmentation result to obtain the corresponding part-of-speech tagging result;

[0203] The third processing submodule is used to call a preset business tagging rule to tag the transaction content based on the part-of-speech tagging result to obtain the collection tag set.

[0204] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0205] In some optional implementations of this embodiment, the first screening module 304 includes:

[0206] The second calling submodule is used to call a preset similarity algorithm;

[0207] A calculation submodule, configured to calculate the similarity between each tag in the aggregated tag set and the specified keyword based on the similarity algorithm;

[0208] A first screening submodule, used to screen out target similarities greater than a preset similarity threshold from all the similarities;

[0209] A second screening submodule is used to screen out a target tag corresponding to the target similarity from the collection tag set;

[0210] A generating submodule is used to generate the specified tag set based on the target tag.

[0211] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0212] In some optional implementations of this embodiment, the second screening module 305 includes:

[0213] A query submodule, configured to query the counterparty information corresponding to a specified transaction based on the transaction client view; wherein the specified transaction is any one of all transactions included in the account transaction flow data;

[0214] The third calling submodule is used to call the preset recognition rules;

[0215] A judgment submodule, used to analyze the transaction counterparty information based on the identification rule to determine whether the transaction counterparty information meets the conditions for fund pooling transactions;

[0216] The marking submodule is used to mark the designated transaction as a suspected fund collection transaction if yes, and mark the designated transaction as a non-fund collection transaction otherwise.

[0217] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0218] In some optional implementations of this embodiment, the view further includes a transaction account view and a transaction institution view; and the label generation device based on artificial intelligence further includes:

[0219] A third construction module is used to construct corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view;

[0220] A partitioning module, used for partitioning the transaction sample data into a training set and a test set;

[0221] The calling module is used to call the preset deep learning model;

[0222] A training module, used to train the deep learning model using the training set based on a preset training strategy to obtain a trained specified model;

[0223] A verification module, used for verifying the specified model based on the test set;

[0224] A determination module is used to use the specified model as the transaction determination model if the specified model passes the verification.

[0225] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0226] In some optional implementations of this embodiment, the third building block includes:

[0227] A collection submodule, used to collect data from the trading account view, the trading institution view and the trading customer view to obtain corresponding initial data;

[0228] A cleaning submodule, used for cleaning the initial data to obtain corresponding processed data;

[0229] An extraction submodule, used for performing feature extraction on the processed data to obtain corresponding feature data;

[0230] A preprocessing submodule, used to preprocess the feature data to obtain processed feature data;

[0231] A determination submodule is used to use the processed feature data as the transaction sample data.

[0232] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based label generation method in the aforementioned implementation mode, and will not be repeated here.

[0233] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0234] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0235] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0236] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the label generation method based on artificial intelligence, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0237] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions stored in the memory 41 or process data, such as computer-readable instructions for running the artificial intelligence-based label generation method.

[0238] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0239] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the artificial intelligence-based label generation method as described above.

[0240] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0241] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A label generation method based on artificial intelligence, characterized in that: The steps include: Collecting account transaction flow data in each subsystem, and constructing corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views; Based on a preset fund collection identification model, the account transaction flow data is subjected to fund collection transaction identification to obtain a corresponding first fund collection transaction; Constructing a collection tag set corresponding to the first fund collection transaction; Screening out a designated tag set matching a designated keyword from the collection tag set, and screening out a second fund collection transaction corresponding to the designated tag set; Screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merging the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction; Performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction; Merging the first fund collection transaction and the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction; The fifth fund collection transaction is labeled to obtain a corresponding target transaction label.

2. The label generation method based on artificial intelligence according to claim 1, characterized in that: The step of constructing a corresponding view based on the account transaction flow data specifically includes: Cleaning the account transaction flow data to obtain the corresponding designated account transaction flow data; Call the preset transaction dimension; Based on a preset data weaving strategy, the transaction flow data of the designated account is organized and processed according to the transaction dimension to obtain a view corresponding to the transaction dimension.

3. The label generation method based on artificial intelligence according to claim 1, characterized in that: The step of constructing a collection tag set corresponding to the first fund collection transaction specifically includes: Obtaining transaction content in the first fund collection transaction; Segmenting the transaction content to obtain corresponding segmentation results; Performing part-of-speech tagging on the word segmentation result to obtain a corresponding part-of-speech tagging result; Based on the part-of-speech tagging result, a preset business tagging rule is called to tag the transaction content to obtain the collection tag set.

4. The label generation method based on artificial intelligence according to claim 1, characterized in that: The step of filtering out a specified tag set matching a specified keyword from the collected tag set specifically includes: Call the preset similarity algorithm; Calculate the similarity between each tag in the aggregated tag set and the specified keyword based on the similarity algorithm; Filter out a target similarity greater than a preset similarity threshold from all the similarities; Filtering out a target tag corresponding to the target similarity from the collected tag set; The specified tag set is generated based on the target tag.

5. The label generation method based on artificial intelligence according to claim 1, characterized in that: The step of screening out suspected fund collection transactions from the account transaction flow data based on the transaction customer view specifically includes: Based on the transaction customer view, query the counterparty information corresponding to the specified transaction; wherein the specified transaction is any one of all transactions included in the account transaction flow data; Call the preset recognition rules; Analyze the counterparty information based on the identification rules to determine whether the counterparty information meets the conditions for fund pooling transactions; If so, the designated transaction is marked as a suspected fund collection transaction; otherwise, the designated transaction is marked as a non-fund collection transaction.

6. The label generation method based on artificial intelligence according to claim 1, characterized in that: The view also includes a transaction account view and a transaction institution view; before the step of performing transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction, the step also includes: Constructing corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view; Dividing the transaction sample data into a training set and a test set; Call the preset deep learning model; Based on a preset training strategy, the deep learning model is trained using the training set to obtain a trained specified model; Validating the specified model based on the test set; If the specified model passes the verification, the specified model is used as the transaction determination model.

7. The label generation method based on artificial intelligence according to claim 6, characterized in that: The step of constructing corresponding transaction sample data based on the transaction account view, the transaction institution view and the transaction customer view specifically includes: Collecting data on the transaction account view, the transaction institution view, and the transaction customer view to obtain corresponding initial data; Performing data cleaning on the initial data to obtain corresponding processed data; Performing feature extraction on the processed data to obtain corresponding feature data; Preprocessing the feature data to obtain processed feature data; The processed feature data is used as the transaction sample data.

8. A label generation device based on artificial intelligence, characterized in that: include: A first construction module is used to collect account transaction flow data in each subsystem and construct corresponding views based on the account transaction flow data; wherein the views at least include transaction customer views; an identification module, configured to identify a fund collection transaction on the account transaction flow data based on a preset fund collection identification model to obtain a corresponding first fund collection transaction; A second building module, used to build a collection label set corresponding to the first fund collection transaction; A first screening module, used to screen out a designated tag set matching a designated keyword from the collection tag set, and screen out a second fund collection transaction corresponding to the designated tag set; a second screening module, configured to screen out suspected fund collection transactions from the account transaction flow data based on the transaction customer view, and merge the second fund collection transaction with the suspected fund collection transaction to obtain a corresponding third fund collection transaction; An analysis module, configured to perform transaction analysis on the third fund collection transaction based on a preset transaction determination model to obtain a corresponding fourth fund collection transaction; a merging module, configured to merge the first fund collection transaction with the fourth fund collection transaction to obtain a corresponding fifth fund collection transaction; A generation module is used to perform labeling processing on the fifth fund collection transaction to obtain a corresponding target transaction label.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the artificial intelligence-based label generation method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the artificial intelligence-based label generation method as described in any one of claims 1 to 7 are implemented.

Images