A system and method for generating key suspicious transaction reports

Abstract

The present invention discloses a system and method for generating a key suspicious transaction report, which relates to the technical field of financial data security. The system includes a data processing unit, a report generating unit, a report exporting unit and a display; the data processing unit is used to acquire, store, parse, analyze and clean transaction data; the report generating unit is used to generate a key suspicious transaction report with analysis dimensions based on the cleaned data; the report exporting unit is used to generate and output a final report based on the dimensions selected by the user; the display is used for the user to input the selected dimensions. The present invention also proposes a method for realizing the system. The present invention can effectively improve the situation in the prior art that the quality of key suspicious transaction reports is not high and the data format is inconsistent with the standard data format, resulting in low report writing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of financial data security, and in particular to a system and method for generating a key suspicious transaction report. Background Art

[0002] Financial institutions have invested considerable manpower and material resources in key suspicious transaction reporting. After data analysis and information screening, the suspicious characteristics of suspicious transactions in terms of customer identity, financial transactions, etc. have become more obvious.

[0003] At present, most of the key suspicious transaction reports of financial institutions are in manual analysis mode. It is difficult for business personnel to quickly and effectively conduct accurate analysis through a large number of transaction flows. In addition, there is a lack of advanced technology and tools to help analysts conduct data statistics and feature analysis, which makes it difficult to ensure the effective analysis of suspicious transactions, resulting in low analysis quality and low efficiency of suspicious transaction reports. After the suspicious transaction analysis is completed, most of the key suspicious transaction reports are basically written manually by analysts. The whole process consumes a lot of time and energy. In addition, the lack of verification mechanism in manual writing makes it difficult to ensure the accuracy of relevant data in the report. Analysts with different experiences also have different analysis angles and writing methods for the report, resulting in uneven quality of key suspicious transaction reports submitted by the same institution. In addition, there is also the problem of inconsistency between the format of data within the institution and the reported data. When submitting key suspicious reports, related data also need to be submitted. Analysts need to integrate the required data from many data sources and prepare corresponding data tables in accordance with the requirements of the People's Bank of China, resulting in reduced efficiency in submitting key suspicious transaction reports. Summary of the invention

[0004] The object of the present invention is to provide a system and method for generating a key suspicious transaction report to solve the problems raised in the prior art.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a system for generating key suspicious transaction reports, the system comprising a data processing unit, a report generating unit, a report exporting unit and a display;

[0006] The data processing unit is used to acquire, store, parse, analyze and clean the transaction data;

[0007] The report generating unit is used to generate a key suspicious transaction report with analysis dimensions based on the cleaned data;

[0008] The report export unit is used to generate and output a final report according to the dimensions selected by the user;

[0009] The display is used for the user to input the dimension selected;

[0010] The data processing unit is connected to the report generating unit; the report generating unit is connected to the report exporting unit; the report exporting unit is connected to the display.

[0011] The data processing unit includes a data acquisition module, a data analysis module, a transaction database, a related data acquisition module, and a data cleaning module;

[0012] The data acquisition module is used to acquire key suspicious transaction data content and transmit it to the transaction database for storage; the key suspicious transaction data content is obtained by screening by staff; the transaction data is obtained by legal and compliant screening by staff;

[0013] The data analysis module is used to obtain basic information such as identity information, account information, and transaction information of transaction personnel based on the analysis of key suspicious transaction data;

[0014] The transaction database is used to store transaction data; the transaction database is a built-in database or an external database, containing a full amount of transaction information; the number of deployed databases is at least 1; the transaction database information is obtained by staff in a legal and compliant manner;

[0015] The relevant data acquisition module is used to acquire transaction data information related to the basic information from the transaction database according to the basic transaction information;

[0016] The data cleaning module is used to clean the data format of key suspicious transaction data and other related transaction data information into a unified format; the unified format is a pre-set standard data format;

[0017] The data acquisition module is communicatively connected with the data analysis module and the transaction database; the related data acquisition module is communicatively connected with the data analysis module, the transaction database and the data cleaning module; the data cleaning module is communicatively connected with the report generation unit.

[0018] The report generation unit includes a relevant data analysis module, a data feature analysis module and a report material module;

[0019] The report material module is used to store information according to preset dimensions;

[0020] The relevant data analysis module is used to obtain the dimensional information of the report from the report material module, and extract the dimensional information content based on the cleaned data after dimensional information analysis; the dimensional information may include, for example, basic information of the trader, such as name, occupation, identity information, account information, legal investigation and freezing data, number of bank card accidents, transaction information, such as transaction type, transaction object, transaction channel, transaction amount, number of transactions, transaction time, transaction location, transaction statistics during the reporting period, transaction object information statistics, source of funds statistics, etc.; the dimensional information data is obtained by legal and compliant screening by the staff;

[0021] The data feature analysis module is used to perform data feature analysis on the cleaned data based on the unextracted dimension information and relying on the big data model;

[0022] The related data analysis module is communicatively connected with the data feature analysis module and the report material module; the data feature analysis module is communicatively connected with the report material module; the report material module is communicatively connected with the report export unit.

[0023] The report export unit includes a dimension selection module, a report integration module and a report output module;

[0024] The dimension selection module is used to obtain a first instruction of an optional report dimension selected by a user from an external input device, and generate a second instruction indicating the full dimension of this report based on the first instruction;

[0025] The report integration module is used to obtain and integrate corresponding data information according to the received instructions;

[0026] The report output module is used to output the integrated key suspicious report;

[0027] The report integration module is communicatively connected to the dimension selection module and the report output module; the dimension selection module is communicatively connected to the display unit.

[0028] The display unit is used for a user to input a first instruction for a selectable dimension report.

[0029] A method for generating a key suspicious transaction report, the method comprising the following steps:

[0030] Step 1: Receive key suspicious transaction data, obtain relevant transaction data based on the key suspicious transaction data, and clean the relevant transaction data according to a preset standard data format;

[0031] Step 2: Based on the cleaned relevant transaction data, perform data analysis on the relevant transaction data based on the report dimension;

[0032] Step 3: Obtain an instruction of an optional reporting dimension selected by the user from an external input device, generate a key suspicious transaction report including the selected reporting dimension based on the instruction, and output the report.

[0033] In step 1, the key suspicious transaction data screened by the staff is received from outside the system, and the key suspicious transaction data is sent to the transaction database for storage;

[0034] The transaction database contains all transaction information and basic information types; the information storage format in the transaction database is expressed as: [B1:V B1 ,B2:V B2 ,...,B n :V Bn ]; among them, B1~B n Represents the basic information of the 1st to nth types of transaction databases, where n is a positive integer, indicating the number of types; V B1 ~V Bn Indicates B1~B n The corresponding value of; the transaction database information is obtained by the staff in a legal and compliant manner;

[0035] For each key suspicious transaction data, the regular expression combined with the context analysis algorithm is used to parse out the basic transaction information. The format is: [b1:V b1 ,b2:V b2 ,...,b m :V bm ]; where {b1,b2,...,b m}⊆{B1,B2,...,B n}, b1~b m Indicates the basic information of the first to m types of transactions, where m is a positive integer, indicating the number of types; V b1 ~V bm Indicates b1~b m The corresponding value of

[0036] Based on the extracted basic information, i As an index, set constraints and obtain transaction data information related to the basic information from the transaction database: where i is a positive integer, indicating the index of the basic information type extracted, i∈{1,2,...,m}, b i Indicates the basic information of the i-th transaction, b i ∈{b1,b2,...,b m};

[0037] When V bi When the value type is numeric, set the lower limit C and upper limit D of the value, C≤V bi≤D; filter out transaction data information whose corresponding information values ​​meet the range [C, D] from the transaction database;

[0038] When V bi When the value type is category type, a word vector is assigned to each category, and the word vectors and V corresponding to the information type are selected from the transaction database. bi Transaction data information with word vector cosine similarity of 1;

[0039] When V bi When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text, set the threshold A, and filter the text vectors and V of the corresponding information type from the transaction database. bi Transaction data information whose text vector cosine similarity satisfies [A,1];

[0040] Set constraints: For basic information b j , when V bj When the value type is numeric, set the lower limit C1 and upper limit D1; the constraint condition is: C1≤V bj ≤D1;

[0041] Where j is a positive integer, b j Indicates the basic information of the jth transaction, b j ∈{b1,b2,...,b m}, j≠i, b j ≠b i ;

[0042] When V bj When the value type is category type, a word vector is assigned to each category; the constraints are: V bj The cosine similarity between the word vector and the preset category word vector is 1;

[0043] When V bj When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text and set the threshold A1; the constraints are: V bj The cosine similarity between the text vector and the preset text vector satisfies the range [A1,1];

[0044] Selecting transaction data information that meets one or more preset constraint conditions from the filtered transaction data information as relevant transaction data information;

[0045] Perform format conversion operations on the acquired relevant transaction data information to unify all transaction data into a pre-set standard data format.

[0046] In step 2, the dimension information of the report is obtained based on the report material library, which is expressed as [β1,β2,...,βt ]; where t is a positive integer, indicating the number of dimension types, β1~β t Represents the 1st to tth dimension information;

[0047] The first dimension information is filtered out from the dimension information according to the relevant transaction data information: T1={β1,β2,...,β t}∩{b1,b2,...,b m}; The first dimension may be, for example, basic information of the trader, transaction information, etc.;

[0048] The remaining information is classified as the second dimension information: T2={b1,b2,...,b m}-{β1,β2,...,β t}∩{b1,b2,...,b m}; The second dimension information may be, for example, transaction statistics, transaction object information statistics, funding source statistics, etc. during the reporting period;

[0049] Wherein, T1 represents the first dimension information set, and T2 represents the second dimension information set;

[0050] Using the big data model to analyze the relevant transaction data based on the first dimension information and the second dimension information respectively;

[0051] Analyze relevant transaction data based on the first dimension information: Generate frequent item sets using the FP-Growth algorithm, generate association rules based on the frequent item sets, set confidence and lift thresholds to analyze the potential associations between different transactions;

[0052] Analyze relevant transaction data based on the second dimension information: For the dimension information of the second dimension information value of the relevant transaction data that is of text type, use the pre-trained language model to generate text vectors; use the K-means clustering algorithm to cluster the relevant transaction data information; use the LDA unsupervised learning algorithm to extract potential topics, and combine manual analysis to determine whether sensitive or suspicious content is involved;

[0053] The data information obtained from the analysis is stored accordingly in the report material library according to the dimension information of the report.

[0054] In step 3, a first instruction of optional report dimensions selected by a user is obtained from an external input device, where the first instruction indicates all optional dimensions included in the generated key suspicious transaction report;

[0055] In the external input device, the report dimension selection interface is displayed to the user in a visual form, and the user selects the dimension by checking the dimension button;

[0056] A second instruction indicating the full dimensions of this report is generated based on the mandatory report dimensions and the optional report dimensions indicated in the first instruction; the mandatory report dimensions are selected by the staff from {β1, β2, ..., β t}Select;

[0057] Based on the second instruction, corresponding dimensions and data information contents are obtained from the report material library, and the obtained data information contents are integrated to obtain a key suspicious transaction report; and the generated key suspicious transaction report is output.

[0058] Compared with the prior art, the beneficial effects of the present invention are: users of the present invention can freely select report dimensions and generate personalized reports as needed, further enhancing the flexibility of the system; different financial institutions or regulatory agencies can adjust the report content according to their own focus to better meet the needs of different business scenarios; the present invention allows the setting of multiple constraints to filter transaction data, and this flexibility allows the system to adapt to different business needs and regulatory requirements; as the business develops or regulatory policies change, the system can adjust the screening rules as needed to ensure its long-term effectiveness and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A structural schematic diagram of a system for generating a key suspicious transaction report according to the present invention;

[0060] Figure 2 A visual schematic diagram of a system and method for generating a key suspicious transaction report according to the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, a system for generating a key suspicious transaction report, the system comprising a data processing unit, a report generating unit, a report exporting unit and a display;

[0063] The data processing unit is used to acquire, store, parse, analyze and clean the transaction data;

[0064] The report generating unit is used to generate a key suspicious transaction report with analysis dimensions based on the cleaned data;

[0065] The report export unit is used to generate and output a final report according to the dimensions selected by the user;

[0066] The display is used for the user to input the dimension selected;

[0067] The data processing unit is connected to the report generating unit; the report generating unit is connected to the report exporting unit; the report exporting unit is connected to the display.

[0068] The data processing unit includes a data acquisition module, a data analysis module, a transaction database, a related data acquisition module, and a data cleaning module;

[0069] The data acquisition module is used to acquire key suspicious transaction data content and transmit it to the transaction database for storage; the key suspicious transaction data content is obtained by screening by staff; the transaction data is obtained by legal and compliant screening by staff;

[0070] The data analysis module is used to obtain basic information such as identity information, account information, and transaction information of transaction personnel based on the analysis of key suspicious transaction data;

[0071] The transaction database is used to store transaction data; the transaction database is a built-in database or an external database, containing a full amount of transaction information; the number of deployed databases is at least 1; the transaction database information is obtained by staff in a legal and compliant manner;

[0072] The relevant data acquisition module is used to acquire transaction data information related to the basic information from the transaction database according to the basic transaction information;

[0073] The data cleaning module is used to clean the data format of key suspicious transaction data and other related transaction data information into a unified format; the unified format is a pre-set standard data format;

[0074] The data acquisition module is communicatively connected with the data analysis module and the transaction database; the related data acquisition module is communicatively connected with the data analysis module, the transaction database and the data cleaning module; the data cleaning module is communicatively connected with the report generation unit.

[0075] The report generation unit includes a relevant data analysis module, a data feature analysis module and a report material module;

[0076] The report material module is used to store information according to preset dimensions;

[0077] The relevant data analysis module is used to obtain the dimensional information of the report from the report material module, and extract the dimensional information content based on the cleaned data after dimensional information analysis; the dimensional information may include, for example, basic information of the trader, such as name, occupation, identity information, account information, legal investigation and freezing data, number of bank card accidents, transaction information, such as transaction type, transaction object, transaction channel, transaction amount, number of transactions, transaction time, transaction location, transaction statistics during the reporting period, transaction object information statistics, source of funds statistics, etc.; the dimensional information data is obtained by legal and compliant screening by the staff;

[0078] The data feature analysis module is used to perform data feature analysis on the cleaned data based on the unextracted dimension information and relying on the big data model;

[0079] The related data analysis module is communicatively connected with the data feature analysis module and the report material module; the data feature analysis module is communicatively connected with the report material module; the report material module is communicatively connected with the report export unit.

[0080] The report export unit includes a dimension selection module, a report integration module and a report output module;

[0081] The dimension selection module is used to obtain a first instruction of an optional report dimension selected by a user from an external input device, and generate a second instruction indicating the full dimension of this report based on the first instruction;

[0082] The report integration module is used to obtain and integrate corresponding data information according to the received instructions;

[0083] The report output module is used to output the integrated key suspicious report;

[0084] The report integration module is communicatively connected to the dimension selection module and the report output module; the dimension selection module is communicatively connected to the display unit.

[0085] The display unit is used for a user to input a first instruction for a selectable dimension report.

[0086] A method for generating a key suspicious transaction report, the method comprising the following steps:

[0087] Step 1: Receive key suspicious transaction data, obtain relevant transaction data based on the key suspicious transaction data, and clean the relevant transaction data according to a preset standard data format;

[0088] Step 2: Based on the cleaned relevant transaction data, perform data analysis on the relevant transaction data based on the report dimension;

[0089] Step 3: Obtain an instruction of an optional reporting dimension selected by the user from an external input device, generate a key suspicious transaction report including the selected reporting dimension based on the instruction, and output the report.

[0090] In step 1, the key suspicious transaction data screened by the staff is received from outside the system, and the key suspicious transaction data is sent to the transaction database for storage; the transaction database information is obtained by the staff in a legal and compliant manner;

[0091] The transaction database contains all transaction information and basic information types; the information storage format in the transaction database is expressed as: [B1:V B1 ,B2:V B2 ,...,B n :V Bn ]; among them, B1~B n Represents the basic information of the 1st to nth types of transaction databases, where n is a positive integer, indicating the number of types; V B1 ~V Bn Indicates B1~B n The corresponding value of

[0092] For each key suspicious transaction data, the regular expression combined with the context analysis algorithm is used to parse out the basic transaction information. The format is: [b1:V b1 ,b2:V b2 ,...,b m :V bm ]; where {b1,b2,...,b m}⊆{B1,B2,...,B n}, b1~b m Indicates the basic information of the first to m types of transactions, where m is a positive integer, indicating the number of types; V b1 ~V bm Indicates b1~b m The corresponding value of

[0093] Based on the extracted basic information, i As an index, set constraints and obtain transaction data information related to the basic information from the transaction database: where i is a positive integer, indicating the index of the basic information type extracted, i∈{1,2,...,m}, b i Indicates the basic information of the i-th transaction, b i ∈{b1,b2,...,b m};

[0094] When V bi When the value type is numeric, set the lower limit C and upper limit D of the value, C≤V bi≤D; filter out transaction data information whose corresponding information values ​​meet the range [C, D] from the transaction database;

[0095] When V bi When the value type is category type, a word vector is assigned to each category, and the word vectors and V corresponding to the information type are selected from the transaction database. bi Transaction data information with word vector cosine similarity of 1;

[0096] When V bi When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text, set the threshold A, and filter the text vectors and V of the corresponding information type from the transaction database. bi Transaction data information whose text vector cosine similarity satisfies [A,1];

[0097] Set constraints: For basic information b j , when V bj When the value type is numeric, set the lower limit C1 and upper limit D1; the constraint condition is: C1≤V bj ≤D1;

[0098] Where j is a positive integer, b j Indicates the basic information of the jth transaction, b j ∈{b1,b2,...,b m}, j≠i, b j ≠b i ;

[0099] When V bj When the value type is category type, a word vector is assigned to each category; the constraints are: V bj The cosine similarity between the word vector and the preset category word vector is 1;

[0100] When V bj When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text and set the threshold A1; the constraints are: V bj The cosine similarity between the text vector and the preset text vector satisfies the range [A1,1];

[0101] Selecting transaction data information that meets one or more preset constraint conditions from the filtered transaction data information as relevant transaction data information;

[0102] Perform format conversion operations on the acquired relevant transaction data information to unify all transaction data into a pre-set standard data format.

[0103] In step 2, the dimension information of the report is obtained based on the report material library, which is expressed as [β1,β2,...,βt ]; where t is a positive integer, indicating the number of dimension types, β1~β t Represents the 1st to tth dimension information;

[0104] The first dimension information is filtered out from the dimension information according to the relevant transaction data information: T1={β1,β2,...,β t}∩{b1,b2,...,b m}; The first dimension may be, for example, basic information of the trader, transaction information, etc.;

[0105] The remaining information is classified as the second dimension information: T2={b1,b2,...,b m}-{β1,β2,...,β t}∩{b1,b2,...,b m}; The second dimension information may be, for example, transaction statistics, transaction object information statistics, funding source statistics, etc. during the reporting period;

[0106] Wherein, T1 represents the first dimension information set, and T2 represents the second dimension information set;

[0107] Using the big data model to analyze the relevant transaction data based on the first dimension information and the second dimension information respectively;

[0108] Analyze relevant transaction data based on the first dimension information: Generate frequent item sets using the FP-Growth algorithm, generate association rules based on the frequent item sets, set confidence and lift thresholds to analyze the potential associations between different transactions;

[0109] Analyze relevant transaction data based on the second dimension information: For the dimension information of the second dimension information value of the relevant transaction data that is of text type, use the pre-trained language model to generate text vectors; use the K-means clustering algorithm to cluster the relevant transaction data information; use the LDA unsupervised learning algorithm to extract potential topics, and combine manual analysis to determine whether sensitive or suspicious content is involved;

[0110] The data information obtained from the analysis is stored accordingly in the report material library according to the dimension information of the report.

[0111] In step 3, a first instruction of optional report dimensions selected by a user is obtained from an external input device, where the first instruction indicates all optional dimensions included in the generated key suspicious transaction report;

[0112] In the external input device, the report dimension selection interface is displayed to the user in a visual form, and the user selects the dimension by checking the dimension button;

[0113] A second instruction indicating the full dimensions of this report is generated based on the mandatory report dimensions and the optional report dimensions indicated in the first instruction; the mandatory report dimensions are selected by the staff from {β1, β2, ..., β t}Select;

[0114] Based on the second instruction, corresponding dimensions and data information contents are obtained from the report material library, and the obtained data information contents are integrated to obtain a key suspicious transaction report; and the generated key suspicious transaction report is output.

[0115] In this embodiment:

[0116] In step 1, the key suspicious transaction data screened by the staff is received from outside the system, and the key suspicious transaction data is sent to the transaction database for storage;

[0117] The transaction database contains all transaction information and basic information types; the information storage format in the transaction database is expressed as: [transaction timestamp: V (transaction timestamp), transaction amount: V (transaction amount), transaction counterparty account: V (transaction counterparty account), transaction type: V (transaction type), transaction location: V (transaction location)]; transaction database information is obtained by staff in a legal and compliant manner;

[0118] For each key suspicious transaction data, use regular expressions combined with context analysis algorithms to parse out its basic transaction information in the following format: [transaction amount: V (transaction amount), transaction location: V (transaction location), transaction type: V (transaction type), transaction counterparty account: V (transaction counterparty account)];

[0119] Based on the extracted basic information, the basic information is used as an index and constraints are set to obtain transaction data information related to the basic information from the transaction database:

[0120] V (transaction amount 1) is a numeric value. Set the lower limit C and upper limit D, C≤V(transaction amount 1)≤D. Filter out transaction data information whose corresponding information values ​​meet the range [C, D] from the transaction database.

[0121] The value type of V (transaction type 1) is category type. A word vector is assigned to each category. The transaction data information whose word vector of the corresponding information type has a cosine similarity of 1 with the word vector of V (transaction type 1) is filtered out from the transaction database.

[0122] The value type of V (transaction location 1) is text type. The pre-trained Doc2Vec model is used to generate a fixed-length vector for the text. The threshold A is set to filter out transaction data information from the transaction database whose text vectors of the corresponding information type value and the cosine similarity of the text vector of V (transaction location 1) satisfy [A, 1].

[0123] Set constraints: For basic information, when the information value type is numeric, set the lower limit C1 and upper limit D1 of the value; the constraint condition is: C1≤information value≤D1;

[0124] When the information value type is category type, a word vector is assigned to each category; the constraint condition is that the cosine similarity between the information value word vector and the preset category word vector is 1;

[0125] When the information value type is text type, the pre-trained Doc2Vec model is used to generate a fixed-length vector for the text, and the threshold A1 is set; the constraint condition is: the cosine similarity between the information value text vector and the preset text vector satisfies the range [A1,1];

[0126] Selecting transaction data information that meets one or more preset constraint conditions from the filtered transaction data information as relevant transaction data information;

[0127] Perform format conversion operations on the acquired relevant transaction data information to unify all transaction data into a pre-set standard data format.

[0128] In step 2, the dimension information of the report is obtained based on the report material library, which is expressed as [transaction amount, transaction type, counterparty account, account association information];

[0129] The first dimension information is filtered out from the dimension information according to the relevant transaction data information: T1={transaction amount, transaction type, transaction counterparty account, account association information}∩{transaction amount, transaction location, transaction type, transaction counterparty account}={transaction amount, transaction type, transaction counterparty account};

[0130] The remaining information is classified as the second dimension information: T2 = {transaction amount, transaction location, transaction type, transaction counterparty account}-{transaction amount, transaction type, transaction counterparty account} = {transaction location};

[0131] Using the big data model to analyze the relevant transaction data based on the first dimension information and the second dimension information respectively;

[0132] Analyze relevant transaction data based on the first dimension information: Generate frequent item sets using the FP-Growth algorithm, generate association rules based on the frequent item sets, set confidence and lift thresholds to analyze the potential associations between different transactions;

[0133] Analyze relevant transaction data based on the second dimension information: For the dimension information of the second dimension information value of the relevant transaction data that is of text type, use the pre-trained language model to generate text vectors; use the K-means clustering algorithm to cluster the relevant transaction data information; use the LDA unsupervised learning algorithm to extract potential topics, and combine manual analysis to determine whether sensitive or suspicious content is involved;

[0134] The data information obtained from the analysis is stored accordingly in the report material library according to the dimension information of the report.

[0135] In step 3, a first instruction of optional report dimensions selected by a user is obtained from an external input device, where the first instruction indicates all optional dimensions included in the generated key suspicious transaction report;

[0136] In the external input device, the report dimension selection interface is displayed to the user in a visual form, and the user selects the dimension by checking the dimension button;

[0137] Generate a second instruction indicating the full dimensions of this report based on the mandatory report dimensions and the optional report dimensions indicated in the first instruction; the mandatory report dimensions are selected by the staff from the dimension information obtained based on the report material library;

[0138] Based on the second instruction, corresponding dimensions and data information contents are obtained from the report material library, and the obtained data information contents are integrated to obtain a key suspicious transaction report; and the generated key suspicious transaction report is output.

[0139] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for generating a key suspicious transaction report, characterized by: The method comprises the following steps: Step 1: Receive key suspicious transaction data, obtain relevant transaction data based on the key suspicious transaction data, and clean the relevant transaction data according to a preset standard data format; Step 2: Based on the cleaned relevant transaction data, perform data analysis on the relevant transaction data based on the report dimension; Step 3: Obtaining an instruction of an optional reporting dimension selected by a user from an external input device, generating a key suspicious transaction report including the selected reporting dimension based on the instruction, and outputting the report; In step 1, the key suspicious transaction data screened by the staff is received from outside the system, and the key suspicious transaction data is sent to the transaction database for storage; The transaction database contains all transaction information and basic information types; the information storage format in the transaction database is expressed as: [B1:V B1 ,B2:V B2 ,...,B n :V Bn ]; among them, B1~B n Represents the basic information of the 1st to nth types of transaction databases, where n is a positive integer, indicating the number of types; V B1 ~V Bn Indicates B1~B n The corresponding value of For each key suspicious transaction data, the regular expression combined with the context analysis algorithm is used to parse out the basic transaction information. The format is: [b1:V b1 ,b2:V b2 ,...,b m :V bm ]; where {b1,b2,...,b m }⊆{B1,B2,...,B n }, b1~b m Indicates the basic information of the first to m types of transactions, where m is a positive integer, indicating the number of types; V b1 ~V bm Indicates b1~b m The corresponding value of Based on the extracted basic information, i As an index, set constraints and obtain transaction data information related to the basic information from the transaction database: where i is a positive integer, indicating the index of the basic information type extracted, i∈{1,2,...,m}, b i Indicates the basic information of the i-th transaction, b i ∈{b1,b2,...,b m }; When V bi When the value type is numeric, set the lower limit C and upper limit D of the value, C≤V bi ≤D; filter out transaction data information whose corresponding information values ​​meet the range [C, D] from the transaction database; When V bi When the value type is category type, a word vector is assigned to each category, and the word vectors and V corresponding to the information type are selected from the transaction database. bi Transaction data information with word vector cosine similarity of 1; When V bi When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text, set the threshold A, and filter the text vectors and V of the corresponding information type from the transaction database. bi Transaction data information whose text vector cosine similarity satisfies [A,1]; Set constraints: For basic information b j , when V bj When the value type is numeric, set the lower limit C1 and upper limit D1; the constraint condition is: C1≤V bj ≤D1; Where j is a positive integer, b j Indicates the basic information of the jth transaction, b j ∈{b1,b2,...,b m }, j≠i, b j ≠b i ; When V bj When the value type is category type, a word vector is assigned to each category; the constraints are: V bj The cosine similarity between the word vector and the preset category word vector is 1; When V bj When the value type is text, use the pre-trained Doc2Vec model to generate a fixed-length vector for the text and set the threshold A1; the constraints are: V bj The cosine similarity between the text vector and the preset text vector satisfies the range [A1,1]; Selecting transaction data information that meets one or more preset constraint conditions from the filtered transaction data information as relevant transaction data information; Performing format conversion operations on the acquired relevant transaction data information to unify all transaction data into a pre-set standard data format; In step 2, the dimension information of the report is obtained based on the report material library, which is expressed as [β1,β2,...,β t ]; where t is a positive integer, indicating the number of dimension types, β1~β t Represents the 1st to tth dimension information; The first dimension information is filtered out from the dimension information according to the relevant transaction data information: T1={β1,β2,...,β t }∩{b1,b2,...,b m }; The remaining information is classified as the second dimension information: T2={b1,b2,...,b m }-{β1,β2,...,β t }∩{b1,b2,...,b m }; Wherein, T1 represents the first dimension information set, and T2 represents the second dimension information set; Using the big data model to analyze the relevant transaction data based on the first dimension information and the second dimension information respectively; Analyze relevant transaction data based on the first dimension information: Generate frequent item sets using the FP-Growth algorithm, generate association rules based on the frequent item sets, set confidence and lift thresholds to analyze the potential associations between different transactions; Analyze relevant transaction data based on the second dimension information: For the dimension information of the second dimension information value of the relevant transaction data that is of text type, use the pre-trained language model to generate text vectors; use the K-means clustering algorithm to cluster the relevant transaction data information; use the LDA unsupervised learning algorithm to extract potential topics, and combine manual analysis to determine whether sensitive or suspicious content is involved; The data information obtained from the analysis is stored accordingly in the report material library according to the dimension information of the report.

2. A method for generating a key suspicious transaction report according to claim 1, characterized in that: In step 3, a first instruction of optional report dimensions selected by a user is obtained from an external input device, where the first instruction indicates all optional dimensions included in the generated key suspicious transaction report; In the external input device, the report dimension selection interface is displayed to the user in a visual form, and the user selects the dimension by checking the dimension button; A second instruction indicating the full dimensions of this report is generated based on the mandatory report dimensions and the optional report dimensions indicated in the first instruction; the mandatory report dimensions are selected by the staff from {β1, β2, ..., β t }Select; Based on the second instruction, corresponding dimensions and data information contents are obtained from the report material library, and the obtained data information contents are integrated to obtain a key suspicious transaction report; and the generated key suspicious transaction report is output.

3. A system for generating a key suspicious transaction report, using a method for generating a key suspicious transaction report according to any one of claims 1 to 2, characterized in that: The system includes a data processing unit, a report generating unit, a report exporting unit and a display; The data processing unit is used to acquire, store, parse, analyze and clean the transaction data; The report generating unit is used to generate a key suspicious transaction report with analysis dimensions based on the cleaned data; The report export unit is used to generate and output a final report according to the dimensions selected by the user; The display is used for the user to input the dimension selected; The data processing unit is connected to the report generating unit; the report generating unit is connected to the report exporting unit; the report exporting unit is connected to the display.

4. A system for generating a key suspicious transaction report according to claim 3, characterized in that: The data processing unit includes a data acquisition module, a data analysis module, a transaction database, a related data acquisition module, and a data cleaning module; The data acquisition module is used to acquire key suspicious transaction data content and transmit it to the transaction database for storage; the key suspicious transaction data content is obtained by screening by staff; The data analysis module is used to obtain basic transaction information based on the analysis of key suspicious transaction data; The transaction database is used to store transaction data; the transaction database is a built-in database or an external database, containing a full amount of transaction information; the number of deployed databases is at least 1; The relevant data acquisition module is used to acquire transaction data information related to the basic information from the transaction database according to the basic transaction information; The data cleaning module is used to clean the data format of key suspicious transaction data and other related transaction data information into a unified format; the unified format is a pre-set standard data format; The data acquisition module is communicatively connected with the data analysis module and the transaction database; the related data acquisition module is communicatively connected with the data analysis module, the transaction database and the data cleaning module; the data cleaning module is communicatively connected with the report generation unit.

5. A system for generating a key suspicious transaction report according to claim 4, characterized in that: The report generation unit includes a relevant data analysis module, a data feature analysis module and a report material module; The report material module is used to store information according to preset dimensions; The relevant data analysis module is used to obtain the dimension information of the report from the report material module, and extract the dimension information content based on the cleaned data after the dimension information analysis; The data feature analysis module is used to perform data feature analysis on the cleaned data based on the unextracted dimension information and relying on the big data model; The related data analysis module is communicatively connected with the data feature analysis module and the report material module; the data feature analysis module is communicatively connected with the report material module; the report material module is communicatively connected with the report export unit.

6. A system for generating a key suspicious transaction report according to claim 5, characterized in that: The report export unit includes a dimension selection module, a report integration module and a report output module; The dimension selection module is used to obtain a first instruction of an optional report dimension selected by a user from an external input device, and generate a second instruction indicating the full dimension of this report based on the first instruction; The report integration module is used to obtain and integrate corresponding data information according to the received instructions; The report output module is used to output the integrated key suspicious report; The report integration module is communicatively connected to the dimension selection module and the report output module; the dimension selection module is communicatively connected to the display unit.

7. A system for generating a key suspicious transaction report according to claim 6, characterized in that: The display unit is used for a user to input a first instruction for a selectable dimension report.

Images