CN114049190A - Financial fraud risk assessment and solution method based on transaction behavior feature extraction - Google Patents
Financial fraud risk assessment and solution method based on transaction behavior feature extraction Download PDFInfo
- Publication number
- CN114049190A CN114049190A CN202111354032.8A CN202111354032A CN114049190A CN 114049190 A CN114049190 A CN 114049190A CN 202111354032 A CN202111354032 A CN 202111354032A CN 114049190 A CN114049190 A CN 114049190A
- Authority
- CN
- China
- Prior art keywords
- transaction
- fraud
- risk
- model
- risk assessment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The financial fraud risk assessment and solution method based on transaction behavior feature extraction adopts a model modeling unit and a fraud risk assessment unit as application software for assessing and solving financial fraud; the assessment and the solution of financial fraud are realized by an artificial intelligence fraud risk assessment model trained or inferred by utilizing a static financial transaction fraud risk finger model extraction technology; the method comprises two steps of modeling a static risk fingerprint extraction technology and a fraud risk assessment model, using the static risk fingerprint extraction technology and making fraud risk assessment. The invention can grasp the essential characteristics of transaction-level fraud risk, provides comprehensive transaction risk assessment and more accurate anti-fraud risk assessment, minimizes the false detection rate while ensuring the anti-fraud rate, ensures the user experience of high-quality financial products, and can accurately capture fraud transactions. The method overcomes the problems that the existing anti-fraud technology can not effectively cope with the modernized transaction fraud, and has the defects of one-sided decision making and evaluation, low accuracy, poor adaptability and the like.
Description
Technical Field
The invention relates to the technical field of anti-fraud applied by financial departments, in particular to a financial fraud risk assessment and solution method based on transaction behavior feature extraction.
Background
With the progress of science and technology, the current financial fraud has the characteristics of specialization, automation, intellectualization, scene and the like. For example, in the existing automatic fraud activities, fraud organizations use digital technologies, such as automatic technology billing of network robots and the like, and fraud efficiency is improved by multiple times. The fraud organization can further improve the fraud capacity by using artificial intelligence, for example, natural language processing and machine learning are utilized to obtain important personal information and financial information of a user, the deceptive activity is disguised into normal financial activity in a scene, for example, monthly payment or specific human activity such as credit card application is simulated, and the deceptive property and the disguise property of the financial fraud are greatly improved.
In the prior art, the anti-fraud technology highly depends on experience, personal and enterprise credit or equipment characteristics, and establishes a blacklist or a model algorithm on the personal and account level to decide/evaluate the transaction risk. Although the related technologies, such as sesame credit, have been applied to interactive platform such as BATJ, these technologies do not essentially analyze different risk characteristics of each transaction under the same person or different individuals/accounts, and cannot cope with the modernized fraud of the transaction. Moreover, the transaction risk decision based on these technologies has the problems of one-sided, low accuracy and poor adaptability, which can inhibit normal transaction and cause negative user experience and user loss of products.
Disclosure of Invention
In order to overcome the defects that the existing anti-fraud technology is highly dependent on experience, personal and enterprise credit or equipment characteristics, can not effectively cope with modernized transaction fraud, has the defects of decision and evaluation one side, low accuracy and poor adaptability, can inhibit normal transactions, can cause negative user experience and user loss of products and the like, the invention provides a financial fraud risk evaluation and solution method based on static risk index extraction, which can grasp the intrinsic characteristics of transaction-level fraud risk and transaction behaviors under the combined action of related application units, provide comprehensive transaction risk evaluation and more accurate anti-fraud risk evaluation, ensure the anti-fraud rate and minimize the false detection rate, thereby ensuring the user experience of high-quality financial products and accurately capture the automated intelligent fraud transactions.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the financial fraud risk assessment and solution method based on transaction behavior feature extraction is characterized in that a model modeling unit and a fraud risk assessment unit are adopted as application software for assessing and solving financial fraud; the financial fraud risk index and the transaction behavior index are used for extracting an artificial intelligence fraud risk evaluation model trained or inferred by technology, so that the financial fraud is evaluated and solved; modeling a static risk finger model extraction technology and a fraud risk assessment model, using a transaction risk finger model extraction technology and performing fraud risk assessment; in the first step, the static risk fingerprint extraction technology and the fraud risk assessment model modeling have the following processes: collecting and extracting data in an original sample database through a model modeling unit, calculating static statistical characteristic values of the data in each set, directly or indirectly calculating artificial intelligence input characteristic values by using a static risk finger model, training a fraud risk assessment model, and finishing training of a risk assessment model based on the static financial transaction fraud risk finger model; the second step, using static risk fingerprinting techniques and making fraud risk assessment, has the following flow: the fraud risk evaluation unit learns new transaction occurrence, acquires transaction scene information and transaction type information, establishes a static risk index, calculates a risk evaluation model input characteristic value, performs reasoning, obtains a fraud risk score of the transaction, and compares the score with a threshold value to give fraud judgment.
Further, in step 1, in the data collected and extracted from the original sample database, the collected and extracted data includes sample data (whether the transaction is a fraudulent transaction, a distance between the user and the merchant, etc.) in a certain scene (e.g., offline transaction of credit card and debit card, online transaction of credit card and debit card, DDA transaction, etc.), a certain set of transaction parameters (e.g., merchant type, transaction location, transaction time, transaction amount, device information, etc.) within a certain historical time (e.g., one year).
Further, in the step 1, in calculating the static statistical characteristic values of the data in each set, the main purpose is to correct, normalize and calibrate the static statistical characteristics in each set, so as to obtain a risk finger model and a risk finger model in each set, wherein the risk finger model in each set needs a continuous variable conforming to mathematical probability distribution; the characteristic values mainly refer to mean values, mean square deviations, occurrence ratios and the like.
Further, in the step 1, the fraud risk assessment model is trained based on the input feature values of the database samples and the fraud transaction records of the samples.
Further, in the step 2, in the establishing of the risk finger model, the static risk finger model table corresponding to the new transaction occurrence, the transaction scene information and the transaction type information data is searched for to obtain the risk finger model.
Further, in the step 2, the calculated input characteristic values of the risk assessment model are obtained according to static risk model data.
The invention has the beneficial effects that: the invention adopts a model modeling unit and a fraud risk evaluation unit as application software; the financial fraud is evaluated and solved by an artificial intelligence fraud risk evaluation model trained or inferred by using a financial transaction fraud risk finger model extraction technology; under the combined action of the related application units, the invention can better grasp the essential characteristics of the transaction-level fraud risk, provide comprehensive transaction risk assessment and more accurate anti-fraud risk assessment, and minimize the false detection rate while ensuring the anti-fraud rate, thereby ensuring the user experience of high-quality financial products and accurately capturing the automatic intelligent fraud transaction. The method overcomes the defects that the existing anti-fraud technology highly depends on experience, personal and enterprise credit or equipment characteristics, can not effectively cope with modern transaction fraud, has the problems of one-sided decision and evaluation, low accuracy and poor adaptability, can inhibit normal transactions, and can cause negative user experience and user loss of products. Based on the above, the invention has good application prospect.
Detailed Description
The financial fraud risk assessment and solution method based on transaction behavior feature extraction adopts a model modeling unit and a fraud risk assessment unit as application software for assessing and solving financial fraud; the financial fraud risk evaluation model is trained or inferred by utilizing financial transaction fraud risk finger mode and transaction behavior finger mode extraction technology, so that the financial fraud is evaluated and solved; the method comprises two steps of modeling a risk fingerprint extraction technology and a fraud risk assessment model, using a transaction risk fingerprint extraction technology and performing fraud risk assessment.
In the first step, the risk finger model extraction technology and the fraud risk assessment model modeling have the following processes: (1) the data are collected and extracted in the original sample database through the model modeling unit, particularly in the data collected and extracted in the original sample database, the collected and extracted data comprise sample data (whether the transaction is a fraud transaction, the distance between the user and the merchant and the like) of the user in a certain scene (credit card and debit card off-line transaction, credit card and debit card on-line transaction, DDA transaction and the like), and a certain transaction parameter set (such as merchant type, transaction place, transaction time, transaction amount, equipment information and the like) in a certain period of historical time (such as one year), and the obtained data are more comprehensive, so that the comprehensive data support is provided for subsequent modeling evaluation and the like. (2) calculating statistical characteristic values of data in each transaction parameter set, wherein the main purpose is to correct the statistical characteristics of the data in each transaction parameter set of a user, normalize and calibrate the data so as to obtain a risk finger model and a risk finger model in each transaction parameter set, wherein the risk finger model in each transaction parameter set needs to accord with a continuous variable of certain mathematical probability distribution; the characteristic values mainly refer to mean values, mean square deviations, occurrence ratios and the like. (3) And (4) training a fraud risk assessment model by directly or indirectly calculating artificial intelligence input characteristic values by using a risk finger model, wherein the assessment model is obtained by AI intelligent learning by mainly using the input characteristic values of the database samples and fraud transaction records of the samples as a basis. (5) After the risk assessment model based on the financial transaction fraud risk index is trained, a foundation is laid for the subsequent fraud risk assessment unit to call the model to assess the risk of the user.
The second step, using risk fingerprint extraction techniques and making fraud risk assessment, has the following flow: (1) and the fraud risk evaluation unit is used for solving the new transaction occurrence of the user and importing the data. (2) And acquiring the transaction scene information and the transaction type information performed by the user and classifying each data to prepare for subsequent evaluation. (3) And establishing a risk finger model, wherein the realization process is obtained by searching a risk finger model table (namely the risk finger model obtained after training) corresponding to the new transaction occurrence, the transaction scene information and the transaction type information data of the user. (4) And calculating an input characteristic value of the risk assessment model, wherein the input characteristic value is calculated according to the risk model data. (5) And (4) the fraud risk evaluation unit carries out reasoning comparison on the financial transaction data of the user obtained based on the steps (1), (2), (3) and (4) to obtain a fraud risk score of the user transaction. (6) Comparing the score obtained in the step (5) with a threshold value of a risk fingerprint memory to give fraud judgment, and judging that the user behavior forms fraud risk suspicion when the fraud risk index is higher than the threshold value, otherwise, judging that the fraud risk suspicion does not exist; after the fraud risk suspicion of the user behavior is obtained, the fraud risk assessment unit can be linked with other application units, for example, the user behavior is warned, financial transactions of the user are refused to be accessed, and the like.
The invention adopts a model modeling unit and a fraud risk evaluation unit as application software; the financial fraud risk is evaluated and solved through a financial transaction fraud risk fingerprint extraction technology and an artificial intelligent fraud risk evaluation model trained or inferred by the financial transaction fraud risk fingerprint extraction technology; under the combined action of the related application units, the invention can better grasp the essential characteristics of the transaction-level fraud risk, provide comprehensive transaction risk assessment and more accurate anti-fraud risk assessment, ensure the anti-fraud rate and simultaneously minimize the false detection rate, thereby ensuring the user experience of high-quality financial products and accurately capturing the automatic intelligent fraud transaction. The method overcomes the defects that the existing anti-fraud technology highly depends on experience, personal and enterprise credit or equipment characteristics, can not effectively cope with modern transaction fraud, has the problems of one-sided decision making and evaluation, low accuracy and poor adaptability, can inhibit normal transactions, and can cause negative user experience and user loss of products.
While there have been shown and described what are at present considered the fundamental principles and essential features of the invention and its advantages, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, but is capable of other embodiments without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Furthermore, it should be understood that although the present description refers to embodiments, the embodiments do not include only one independent technical solution, and such description is only for clarity, and those skilled in the art should take the description as a whole, and the technical solutions in the embodiments may be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims (6)
1. The financial fraud risk assessment and solution method based on transaction behavior feature extraction is characterized in that a model modeling unit and a fraud risk assessment unit are adopted as application software for assessing and solving financial fraud; the financial fraud risk evaluation model is trained or inferred by utilizing financial transaction fraud risk finger mode and transaction behavior finger mode extraction technology, so that the financial fraud is evaluated and solved; modeling a static risk finger model extraction technology and a fraud risk assessment model, using a transaction risk finger model extraction technology and performing fraud risk assessment; in the first step, the static risk fingerprint extraction technology and the fraud risk assessment model modeling have the following processes: collecting and extracting data in an original sample database through a model modeling unit, calculating static statistical characteristic values of the data in each set, directly or indirectly calculating artificial intelligence input characteristic values by using a static risk finger model, training a fraud risk evaluation model, and finishing training of a risk evaluation model based on the static financial transaction fraud risk finger model; the second step, using static risk fingerprinting techniques and making fraud risk assessment, has the following flow: the fraud risk evaluation unit learns new transaction occurrence, acquires transaction scene information and transaction type information, establishes a static risk index, calculates a risk evaluation model input characteristic value, performs reasoning, obtains a fraud risk score of the transaction, and compares the score with a threshold value to give fraud judgment.
2. The method of claim 1, wherein the step 1 of collecting and extracting data from the original sample database comprises collecting sample data (whether the transaction is a fraudulent transaction, distance between the user and the merchant, etc.) in a certain scene (off-line transaction of credit card and debit card, on-line transaction of credit card and debit card, DDA transaction, etc.), a certain set of transaction parameters (e.g. merchant type, transaction location, transaction time, transaction amount, device information, etc.) during a certain historical time (e.g. one year).
3. The financial fraud risk assessment and solution method based on transaction behavior feature extraction as claimed in claim 1, wherein, in step 1, the static statistical feature values of the data in each set are calculated, and the main purpose is to correct, normalize and calibrate the static statistical features in each set, so as to obtain a risk finger model and a risk finger model in each set, wherein the risk finger model in each set needs continuous variables conforming to mathematical probability distribution; the characteristic values mainly refer to mean values, mean square deviations, occurrence ratios and the like.
4. The financial fraud risk assessment and solution method based on transaction behavior feature extraction of claim 1, wherein in the step 1, the fraud risk assessment model is trained by mainly using the input feature values of the database samples and the fraud transaction records of the samples as the basis.
5. The financial fraud risk assessment and solution method based on transaction behavior feature extraction of claim 1, wherein in the step 2, the risk finger model is established by searching a static risk finger model table corresponding to the new transaction occurrence, the transaction scenario information and the transaction type information data.
6. The financial fraud risk assessment and solution method according to claim 1, wherein the step 2 of calculating the risk assessment model input feature value is obtained according to static risk index data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111354032.8A CN114049190A (en) | 2021-11-11 | 2021-11-11 | Financial fraud risk assessment and solution method based on transaction behavior feature extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111354032.8A CN114049190A (en) | 2021-11-11 | 2021-11-11 | Financial fraud risk assessment and solution method based on transaction behavior feature extraction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114049190A true CN114049190A (en) | 2022-02-15 |
Family
ID=80209577
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111354032.8A Pending CN114049190A (en) | 2021-11-11 | 2021-11-11 | Financial fraud risk assessment and solution method based on transaction behavior feature extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114049190A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171720A (en) * | 2023-08-17 | 2023-12-05 | 哈尔滨工业大学 | Data attribution right identification system and method based on behavior fingerprint |
| CN118070141A (en) * | 2024-04-25 | 2024-05-24 | 成都乐超人科技有限公司 | Artificial intelligence-based anti-fraud transaction identification method and system |
-
2021
- 2021-11-11 CN CN202111354032.8A patent/CN114049190A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171720A (en) * | 2023-08-17 | 2023-12-05 | 哈尔滨工业大学 | Data attribution right identification system and method based on behavior fingerprint |
| CN117171720B (en) * | 2023-08-17 | 2024-03-22 | 哈尔滨工业大学 | Data attribution right identification system and method based on behavior fingerprint |
| CN118070141A (en) * | 2024-04-25 | 2024-05-24 | 成都乐超人科技有限公司 | Artificial intelligence-based anti-fraud transaction identification method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106803168B (en) | Abnormal transfer detection method and device | |
| TWI739798B (en) | Method and device for establishing data recognition model | |
| CN109035003A (en) | Anti- fraud model modelling approach and anti-fraud monitoring method based on machine learning | |
| Dorronsoro et al. | Neural fraud detection in credit card operations | |
| CN108053318B (en) | Method and device for identifying abnormal transactions | |
| CN114049190A (en) | Financial fraud risk assessment and solution method based on transaction behavior feature extraction | |
| CN105354210A (en) | Mobile game payment account behavior data processing method and apparatus | |
| CN110097451B (en) | Bank business monitoring method and device | |
| CN109919624A (en) | A kind of net loan fraud clique's identification and method for early warning based on space-time centrality | |
| CN106485528A (en) | The method and apparatus of detection data | |
| CN110210966A (en) | The processing method of User reliability social network data | |
| CN112767136A (en) | Credit anti-fraud identification method, credit anti-fraud identification device, credit anti-fraud identification equipment and credit anti-fraud identification medium based on big data | |
| CN111476296A (en) | Sample generation method, classification model training method, identification method and corresponding devices | |
| CN110335144B (en) | Personal electronic bank account security detection method and device | |
| CN109102396A (en) | A kind of user credit ranking method, computer equipment and readable medium | |
| CN110533519A (en) | Feature branch mailbox algorithm based on decision tree | |
| CN110728570B (en) | Anti-fraud fund analysis method | |
| CN111160695A (en) | Method, system, device and storage medium for identifying risk account of computer operation | |
| CN107871213B (en) | Transaction behavior evaluation method, device, server and storage medium | |
| CN117993919A (en) | Bank anti-electricity fraud data model construction method based on multi-feature fusion | |
| CN112967053A (en) | Method and device for detecting fraudulent transactions | |
| CN107679862A (en) | A kind of characteristic value of fraudulent trading model determines method and device | |
| CN111738824A (en) | Method, device and system for screening financial data processing modes | |
| CN116883173A (en) | High-frequency quantitative transaction method and system based on deep learning | |
| CN111951105A (en) | Intelligent credit wind control system based on multidimensional big data analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230720 Address after: Room 201-2, Floor 2, Building 7, No. 1, Mengdu Street, Jianye District, Nanjing, Jiangsu Province, 210000 Applicant after: Nanjing Mingcheng Zhike Information Technology Co.,Ltd. Address before: Building 1, 5500 Yuanjiang Road, Minhang District, Shanghai 201100 Applicant before: Shanghai Kaiming Zhidun Intelligent Technology Co.,Ltd. |