WO2020107756A1 - Credit anti-fraud method, system, device and computer-readable storage medium - Google Patents

Abstract

Disclosed in the present invention is a credit anti-fraud method, comprising: when a credit identity verification call is monitored, obtaining a risk level of a current identity verification customer, and obtaining customer voice data acquired during the credit identity verification call; determining a target anti-fraud strategy according to the risk level, wherein the target anti-fraud strategy is one or more among a voiceprint anti-fraud strategy, a voice anti-fraud strategy, a light voice anti-fraud strategy, a background sound anti-fraud strategy, and a voice emotion anti-fraud strategy; and performing an anti-fraud identification operation according to the customer voice data and the target anti-fraud strategy so as to obtain an anti-fraud identification result of the current identity verification customer. Also disclosed in the present invention are a credit anti-fraud system and device, and a computer-readable storage medium. The present invention may improve the recognition rate and coverage rate of fraud and reduce credit risk.

Description

Credit anti-fraud method, system, equipment and computer readable storage medium The

Technical field

The invention relates to the field of anti-fraud, in particular to a credit anti-fraud method, system, equipment and computer-readable storage medium.

Background technique

Financial companies need to check whether credit applicants have the risk of fraud when carrying out credit review, and the accuracy and efficiency of manual review cannot be guaranteed. For this reason, major financial companies have developed anti-fraud systems, which can help finance through anti-fraud systems The organization locates and identifies fraudulent customers and events. With the rapid development of artificial intelligence, the development of anti-fraud from artificial recognition to intelligent algorithm recognition not only reduces the cost of labor, but more importantly improves the accuracy and recall rate of anti-fraud, and ensures the stability of anti-fraud Consistent with standards.

At present, the identification of fraudulent customers and events is mainly based on face recognition technology and voiceprint recognition technology, which has improved the recognition rate of fraud to a certain extent. However, fraudulent customers are driven by interests and change various fraud methods to search for anti-fraud There are loopholes in the system, and fraud is used to attempt to defraud financial institutions. The existing anti-fraud methods cannot identify all fraudulent customers and events, and the coverage rate of fraud identification is low. Therefore, how to improve fraud recognition rate and coverage rate and reduce credit risk is an urgent problem to be solved.

Summary of the invention

The main purpose of the present invention is to provide a credit anti-fraud method, system, device and computer-readable storage medium, aiming to increase fraud recognition rate and coverage rate and reduce credit risk.

To achieve the above object, the present invention provides a credit anti-fraud method, which includes the following steps:

When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.

Further, according to the risk level, the step of determining the target anti-fraud strategy includes:

Obtain the mapping table between the pre-stored risk level and the strategy identification code;

Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.

Further, after performing the anti-fraud identification operation according to the customer voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current nuclear customer, it also includes:

According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.

Further, according to the anti-fraud identification result, the step of adjusting the current risk level of the nuclear customer includes:

According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.

Further, after the step of performing the fraud risk reminding operation, the method further includes:

When the triggered core adding instruction of the current core body client is detected, the core adding issue broadcast operation is performed in the credit core body call according to the preset core adding question tree.

Further, the step of performing the auditing issue broadcast operation in the credit core call according to the preset auditing issue tree includes:

Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.

In addition, to achieve the above object, the present invention also provides a credit anti-fraud system. The credit anti-fraud system includes:

An obtaining module, used to obtain the current risk level of the customer of the nuclear nucleus when the credit nucleus call is monitored, and obtain the customer voice data collected during the credit nucleus call;

A strategy determination module for determining a target anti-fraud strategy based on the risk level, wherein the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background voice anti-fraud strategy and One or more of voice emotion anti-fraud strategies;

The anti-fraud module is used to perform an anti-fraud identification operation based on the customer voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current nuclear customer.

In addition, in order to achieve the above object, the present invention also provides a credit anti-fraud device, the credit anti-fraud device includes: a memory, a processor, and a credit anti-fraud device stored on the memory and operable on the processor Program, when the credit anti-fraud program is executed by the processor, the following steps are realized:

When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.

The present invention also provides a computer-readable storage medium on which a credit anti-fraud program is stored. When the credit anti-fraud program is executed by a processor, the following steps are implemented:

When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.

The invention provides a credit anti-fraud method, system, equipment and computer-readable storage medium. When the credit verification instruction is detected, the invention initiates an outgoing call to the corresponding terminal device, and when the outgoing call is connected, Receive the customer voice data sent by the terminal device during the call, and then determine the target anti-fraud strategy based on the risk level in the credit verification instruction, and execute the corresponding anti-fraud based on the customer voice data and the target anti-fraud strategy Recognition operation to obtain the anti-fraud identification result of the current nuclear customer, through the above-mentioned methods, in the nuclear verification process, based on the current nuclear customer's risk level, select the corresponding anti-fraud strategy, and based on the customer's voice data and anti-fraud strategy , Perform the corresponding anti-fraud identification operation, which can accurately and comprehensively identify anti-fraud, effectively improve the identification rate and coverage of fraud, and reduce credit risk.

BRIEF DESCRIPTION

FIG. 1 is a schematic diagram of a device structure of a hardware operating environment according to an embodiment of the present invention;

2 is a schematic flowchart of a first embodiment of a credit anti-fraud method of the present invention;

3 is a schematic flowchart of a second embodiment of the credit anti-fraud method of the present invention;

4 is a schematic diagram of functional modules of the first embodiment of the credit anti-fraud system of the present invention.

The implementation, functional characteristics and advantages of the present invention will be further described in conjunction with the embodiments and with reference to the drawings.

detailed description

It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not intended to limit the present invention.

As shown in FIG. 1, FIG. 1 is a schematic diagram of a device structure of a hardware operating environment according to a solution of an embodiment of the present invention.

As shown in FIG. 1, the credit anti-fraud device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to implement connection communication between these components. The user interface 1003 may include a display (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as disk storage. The memory 1005 may optionally be a storage device independent of the foregoing processor 1001.

Those skilled in the art can understand that the structure of the credit anti-fraud device shown in FIG. 1 does not constitute a limitation on the credit anti-fraud device, and may include more or fewer components than the illustration, or a combination of certain components, or different Parts arrangement.

As shown in FIG. 1, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a credit anti-fraud program.

In the credit anti-fraud device shown in FIG. 1, the network interface 1004 is mainly used to connect to the back-end server and perform data communication with the back-end server; the user interface 1003 is mainly used to connect to the client (user end) and perform data communication with the client; The processor 1001 can be used to call the credit anti-fraud program stored in the memory 1005 and perform the following steps:

When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.

Further, the processor 1001 may be used to call the credit anti-fraud program stored in the memory 1005, and further perform the following steps:

Obtain the mapping table between the pre-stored risk level and the strategy identification code;

Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.

Further, the processor 1001 may be used to call the credit anti-fraud program stored in the memory 1005, and further perform the following steps:

According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.

Further, the processor 1001 may be used to call the credit anti-fraud program stored in the memory 1005, and further perform the following steps:

According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.

Further, the processor 1001 may be used to call the credit anti-fraud program stored in the memory 1005, and further perform the following steps:

When the triggered core adding instruction of the current core body client is detected, the core adding issue broadcast operation is performed in the credit core body call according to the preset core adding question tree.

Further, the processor 1001 may be used to call the credit anti-fraud program stored in the memory 1005, and further perform the following steps:

Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.

The specific embodiments of the credit anti-fraud device of the present invention are basically the same as the specific embodiments of the credit anti-fraud method described below, and are not repeated here.

The invention provides a credit anti-fraud method.

Referring to FIG. 2, FIG. 2 is a schematic flowchart of a first embodiment of a credit anti-fraud method of the present invention.

In this embodiment, the credit anti-fraud method includes:

Step S101, when a credit verification call is monitored, the current risk level of the verification client is obtained, and the customer voice data collected during the credit verification call is acquired;

In this embodiment, the credit anti-fraud method is applied to a credit anti-fraud system, which can identify anti-fraud customers based on the credit anti-fraud system during the process of verifying customers, and determine whether the nuclear customers There is suspicion of fraud. Specifically, the electrical auditor pulls the nuclear body work order through the nuclear body front, that is, requests the nuclear body work order from the nuclear body server through the nuclear body front, and the nuclear body server issues the corresponding nuclear based on the request of the nuclear body front The body work order, and after receiving the body work order, the body front page displays the body work page. The body work page displays the body work order, outgoing control, body work process, takeover control, and check. Controls, etc. Among them, the verification process includes self-reported identity, confirmation of the convenience of customer calls, phone recording prompts, confirmation of customer information, broadcast and verification issues, and loan use reminders, etc. The verification work order contains the work order type and work order identification code , Business basic information, customer basic information and historical verification records, etc., and the basic business information includes but is not limited to review type, loan application channel, work order level and nuclear verification information, the basic customer information includes but not limited to name, Gender, date of birth, ID card number, mobile phone number and marital status.

Through the outgoing call control on the nuclear operation page, the electric nuclear operator triggers a credit verification request containing the risk level and mobile phone number, and sends the credit verification request to the nuclear verification server. The credit verification request is forwarded to the soft phone platform, and the soft verification platform initiates a credit verification call based on the mobile phone number in the credit verification request. When it is detected that the credit verification call is connected, the electric auditor or intelligent robot can communicate with The current nuclear customer communication is to make a credit nuclear phone call. During the credit nuclear phone call, the customer voice data in the nuclear phone call is collected, and the customer voice data is transmitted to the credit anti-fraud system.

When the credit anti-fraud system detects a credit verification call, it obtains the current risk level of the verification client and obtains the customer voice data collected during the credit verification call. It should be noted that the higher the risk level, the higher the customer's suspected fraud, and the lower the risk level, the lower the customer's fraud suspected. For example, the risk level includes S level, A level, B level, C level, D level, and E level, a total of six levels, among which, credit customers with risk level S are not suspected of fraud, and credit customers with risk level A Has a very low suspicion of fraud, credit customers with a risk rating of B have a low suspicion of fraud, credit customers with a risk rating of C have a high suspicion of fraud, and credit customers with a risk rating of D have a high suspicion of fraud with a risk rating of E-grade credit customers are extremely suspected of fraud.

Step S102: Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background voice anti-fraud strategy and voice emotion anti-fraud strategy One or more

In this embodiment, after obtaining the risk level of the current nuclear client, the credit anti-fraud system determines the target anti-fraud strategy of the current nuclear client based on the risk level, that is, between the pre-stored risk level and the policy identification code And then query the mapping relationship table to obtain the policy identification code corresponding to the risk level, and determine the anti-fraud strategy corresponding to the policy identification code as the target anti-fraud strategy. Among them, the target anti-fraud strategy is one or more of voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotional anti-fraud strategy. It should be noted that the mapping relationship table between the risk level and the policy identification code can be set by a person skilled in the art based on actual conditions, which is not specifically limited in this embodiment.

For example, the risk level includes six levels: S level, A level, B level, C level, D level, and E level. Voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background voice anti-fraud strategy and The corresponding policy identification codes of the voice emotion anti-fraud strategy are Pxxxx, Oxxxx, Ixxxx, Uxxxx, and Yxxxx respectively. The mapping relationship between the risk level and the policy identification code is shown in the following table:

Risk level	Strategy ID
S grade	Pxxxx
Class A	Pxxxx Oxxxx
Class B	Pxxxx Oxxxx Ixxxx
Class C	Pxxxx Oxxxx Ixxxx Uxxxx
Class D	Pxxxx Oxxxx Ixxxx Uxxxx Yxxxx
Class E	Pxxxx Oxxxx Ixxxx Uxxxx Yxxxx

As can be seen from the table above, if the risk level is S level, the target anti-fraud strategy is the voiceprint anti-fraud strategy corresponding to Pxxxx; if the risk level is A level, then the target anti-fraud strategy includes the voiceprint anti-fraud strategy corresponding to Pxxxx and Oxxxx Corresponding voice anti-fraud strategy; if the risk level is B, the target anti-fraud strategy includes the voiceprint anti-fraud strategy corresponding to Pxxxx, the voice anti-fraud strategy corresponding to Oxxxx, and the light voice anti-fraud strategy corresponding to Ixxxx; the risk level is C , The target anti-fraud strategy includes the voiceprint anti-fraud strategy corresponding to Pxxxx, the voice anti-fraud strategy corresponding to Oxxxx, the light voice anti-fraud strategy corresponding to Ixxxx, and the background sound anti-fraud strategy corresponding to Uxxxx; the risk levels are D and E , The target anti-fraud strategies include the voiceprint anti-fraud strategy corresponding to Pxxxx, the voice anti-fraud strategy corresponding to Oxxx, the light voice anti-fraud strategy corresponding to Ixxxx, the background sound anti-fraud strategy corresponding to Uxxxx, and the voice emotion anti-fraud strategy corresponding to Yxxxx .

Specifically, the voiceprint anti-fraud strategy is to perform voiceprint recognition on customer voice data to obtain the voiceprint of the current core customer, and compare the obtained voiceprint with the pre-recorded voiceprint of the current core customer in the voiceprint library. To determine whether the voiceprint is the same as the pre-recorded voiceprint of the current core customer in the voiceprint library. If the voiceprint is the same, it can be determined that the person answering the call is the person. Determine that the current nuclear customer is suspected of fraud; or, compare the recognized voiceprint with the voiceprint in the voiceprint blacklist library. If it hits, it is determined that the current nuclear customer is suspected of fraud. If it does not hit, it will not be processed. .

The voice anti-fraud strategy is to perform voice recognition on the customer's voice data, convert the customer's voice into text information, and match the text information with the currently pre-recorded related information. If it does not match, then it can be determined that the current nuclear customer exists Suspect fraud, if it does not match, it will not be processed; light voice anti-fraud strategy is to perform light voice recognition on customer voice data to determine whether there is light voice in customer voice data. If there is light voice in customer voice data, you can determine the current Suspected customers are suspected of fraud, and if there is no light voice in the customer's voice data, it will not be processed.

The background sound anti-fraud strategy is to identify the background sound in the customer's voice data, determine the type of background sound of the current nuclear customer's environment, and use the recognized background sound category and the current nuclear customer's pre-recorded background category as By comparison, if the category of the background sound recognized is different from the pre-recorded background category of the current nuclear customer, it can be determined that the current nuclear customer is suspected of fraud. If the category of the recognized background sound is different from the pre-recorded audio customer If the background categories are the same, the remaining anti-fraud strategies need to be implemented; the voice emotion anti-fraud strategy is to perform voice emotion recognition on the customer's voice data, determine the emotion category of the current nuclear customer, and determine whether the emotion category is the preset emotion category, such as Nervous, if the emotion category is the preset emotion category, it can be determined that the current nuclear customer is suspected of fraud, and if the emotion category is not the preset emotion category, it will not be processed.

Step S103, according to the customer's voice data and the target anti-fraud strategy, perform an anti-fraud identification operation to obtain the anti-fraud identification result of the current nuclear customer.

In this embodiment, after determining the customer voice data and the target anti-fraud strategy, the credit anti-fraud system performs an anti-fraud identification operation based on the customer's voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current nuclear customer. Specifically, taking the target anti-fraud strategy as voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy as examples:

The credit anti-fraud system first implements the voiceprint anti-fraud strategy for current nuclear customers to obtain the first anti-fraud recognition result of the current nuclear customers, that is, the credit anti-fraud system performs voiceprint feature recognition on the customer's voice data to obtain several voices Voiceprint feature, and input a number of voiceprint feature numbers into the voiceprint model to obtain the voiceprint of the current core user, and then determine whether the voiceprint is the same as the pre-recorded voiceprint of the current client, if the voiceprint is the same as the current client’s voiceprint If the pre-recorded voiceprints are the same, the first anti-fraud identification result of the current nuclear customer is that there is no suspect of non-self fraud. If the voiceprint is different from the pre-recorded voiceprint of the current customer, the first nuclear customer is determined to be the first The result of anti-fraud identification is suspected of non-self fraud;

Then execute the voice anti-fraud strategy on the current nuclear customer to obtain the second anti-fraud recognition result of the current nuclear customer, that is, the credit anti-fraud system performs voice recognition on the customer's voice data, converts the customer's voice data into text, and converts the text Match with the pre-entered text of the current nuclear body customer. If the text matches the pre-entered text of the current nuclear body customer, the second anti-fraud recognition result of the current nuclear body customer is determined to be that there is no suspected information fraud. If the pre-entered text of the current nuclear customer does not match, it is determined that the second anti-fraud identification result of the current nuclear customer is suspected of information fraud;

Then, execute the light voice anti-fraud strategy on the current nuclear customer to obtain the third anti-fraud recognition result of the current nuclear customer, that is, the credit anti-fraud system performs light speech recognition on the customer's voice data to determine whether the customer's voice data There is voice data that includes soft-sound sound spectrum features. If there is no voice data that includes soft-sound sound spectrum features in the customer's voice data, the third anti-fraud recognition result of the current nuclear customer is determined to be that there is no soft-spoofing suspect. If there is voice data in the customer's voice data that contains soft voice spectrum characteristics, the third anti-fraud recognition result of the current core customer is determined to be suspected of soft fraud;

Then, execute the background sound anti-fraud strategy on the current nuclear customer to obtain the fourth anti-fraud recognition result of the current nuclear customer, that is, the credit anti-fraud system performs background sound recognition on the customer's voice data to obtain the current nuclear customer's location The background sound category of the environment, and compare the background sound category with the pre-recorded background sound category of the current nuclear customer. If the background sound category is the same as the pre-recorded background sound category of the current nuclear customer, determine the current nuclear customer The fourth anti-fraud recognition result is that there is no suspect of background sound fraud. If the background sound category is different from the pre-recorded background sound category of the current nuclear body customer, the fourth anti-fraud recognition result of the current nuclear body customer is determined to be the presence of background sound Suspected of fraud;

Finally, the voice emotion anti-fraud strategy is implemented on the current nuclear customer, and the fifth anti-fraud recognition result of the current nuclear customer is obtained, that is, the credit anti-fraud system performs voice emotion recognition on the customer's voice data to obtain the voice emotion of the current nuclear customer Category, and determine whether the voice sentiment category is a preset category, such as nervousness, if the voice sentiment category is a preset category, it is determined that there is a suspected emotional fraud in the fifth anti-fraud recognition result of the current nuclear customer, if the voice emotion If the category is not the preset category, it is determined that there is no suspected emotional fraud in the fifth anti-fraud identification result of the current nuclear customer.

Synthesizing the first anti-fraud identification result, the second anti-fraud identification result, the third anti-fraud identification result, the fourth anti-fraud identification result, and the fifth anti-fraud identification result, to obtain the anti-fraud identification result of the current nuclear customer, where The results of anti-fraud identification include one or more of non-self fraud suspects, no information fraud suspects, no soft fraud suspects, no background sound fraud suspects, and no emotional fraud suspects, and non-self fraud suspects, There is a combination of one or more of suspected information fraud, suspected soft fraud, suspected background sound fraud, and suspected emotional fraud.

In specific implementation, the credit anti-fraud system can also simultaneously execute voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background voice anti-fraud strategy and voice emotional anti-fraud strategy.

In this embodiment, when the credit verification instruction is detected, the invention initiates an outgoing call to the corresponding terminal device, and when the outgoing call is connected, receives the customer voice data sent by the terminal device during the call, and then According to the risk level in the credit verification command, determine the target anti-fraud strategy, and perform the corresponding anti-fraud identification operation based on the customer's voice data and the target anti-fraud strategy to obtain the current anti-fraud identification result of the nuclear verification customer, Through the above-mentioned methods, in the verification process, based on the current risk level of the verification customer, the corresponding anti-fraud strategy is selected, and based on the customer's voice data and anti-fraud strategy, the corresponding anti-fraud identification operation is performed to accurately and comprehensively identify the anti-fraud Fraud, effectively improve the recognition rate and coverage of fraud, and reduce credit risk.

Further, referring to FIG. 3, based on the above-mentioned first implementation, a second embodiment of the credit anti-fraud method of the present invention is proposed. The difference from the foregoing embodiment is that after step S103, it further includes:

Step S104: According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

In this embodiment, after obtaining the anti-fraud identification result of the current nuclear client, the credit anti-fraud system determines whether the current nuclear client is suspected of fraud based on the anti-fraud identification result, that is, the anti-fraud identification of the current nuclear client Does the result include one or more of suspected non-person fraud, suspected information fraud, suspected soft fraud, suspected background audio fraud, and suspected emotional fraud? If the current anti-fraud identification result of the nuclear user includes one or more of non-personal fraud suspects, information fraud suspects, soft fraud suspects, background sound fraud suspects, and emotional fraud suspects, you can determine the current nuclear Personal customers are suspected of fraud, if the current anti-fraud identification results of nuclear customers are that there is no non-personal fraud suspect, no information fraud suspect, no soft fraud suspect, no background sound fraud suspect, and no emotional fraud suspect, then Make sure that current nuclear customers have no suspicions of fraud.

In step S105, if the current nuclear customer has a suspicion of fraud, the risk level of the current nuclear customer is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.

In this embodiment, if the current nuclear identity customer is suspected of fraud, the credit anti-fraud system adjusts the current nuclear identity customer's risk level based on the anti-fraud identification result, that is, increases the current nuclear identity customer's risk level, and executes a fraud risk alert operating. Specifically, based on the anti-fraud identification result, the current fraudulent customer's fraud index is determined, and it is determined whether the fraudulent index is greater than or equal to a preset threshold. Increase the risk level by one level. Among them, the higher the risk level, the greater the suspicion of fraud. For example, the risk level includes six levels: S level, A level, B level, C level, D level, and E level, and S level<A level<B level<C Level <D level <E level, that is, credit customers with risk level S are not suspected of fraud, credit customers with risk level A are extremely low, and credit customers with risk level B are low Credit customers with a rating of grade C have a higher suspicion of fraud, those with a risk rating of D have a higher suspicion of fraud, and those with a risk rating of E have a higher suspicion of fraud. It should be noted that the above-mentioned preset threshold can be set by a person skilled in the art based on actual conditions, which is not specifically limited in this embodiment.

Among them, the way to determine the fraud index is to obtain each type of fraud suspect contained in the anti-fraud identification result (non-self fraud suspect, information fraud suspect, soft fraud suspect, background sound fraud suspect and voice emotional fraud suspect), and query the pre-stored The mapping relationship table of fraud suspect types and fraud index, obtain the fraud index corresponding to the included fraud suspect type, and accumulate the fraud index corresponding to each fraud suspect type, and the total fraud index obtained is the fraud index of the current nuclear customer. It should be noted that the mapping relationship table between the fraud suspect type and the fraud index can be set by a person skilled in the art based on actual conditions, which is not specifically limited in this embodiment. Optionally, the mapping relationship between the suspected fraud type and the fraud index is shown in the following table:

Type of suspected fraud	Fraud Index
Not suspected of fraud	30
Suspected information fraud	30
Softly suspected of fraud	15
Suspected background sound fraud	15
Voice Emotional Fraud Suspected	10

Among them, when the anti-fraud identification result contains only non-person fraud suspects, the current fraud index of the nuclear self-identity customers is 30; when the anti-fraud identification result contains non-person fraud suspects and softly suspected fraud, the current fraudulent customer's fraud index is 45 ; When the anti-fraud identification result includes light fraud suspects, background sound fraud suspects, and voice emotional fraud suspects, the current fraud index of the self-identity customers is 40; when the anti-fraud identification results include non-personal fraud suspects and information fraud suspects, the current nuclear The fraud index of personal customers is 60.

In this embodiment, when the present invention determines that the current nuclear customer is suspected of fraud based on the anti-fraud identification result, the risk level of the current nuclear customer is increased, and a fraud risk reminder operation is performed to reduce the risk of dishonesty.

Further, based on the above second embodiment, a third embodiment of the credit anti-fraud method of the present invention is proposed, which differs from the foregoing embodiment in that after step S104, it further includes:

In step A, when the triggered verification instruction of the current nuclear body client is detected, the verification operation of the verification body is performed in the credit verification call according to the preset verification problem tree.

In this embodiment, after performing the fraud risk reminder operation, the electric auditor may manually trigger the current nuclear customer's nuclear addition instruction, or an intelligent robot may automatically trigger the current nuclear identity customer's nuclear addition instruction, when the credit anti-fraud system detects When the current core user's core addition instruction is triggered, the core addition question broadcast operation is performed on the outgoing call according to the preset core addition question tree, that is, each time the current answer option selected by the customer based on the current core addition question is received At the time, according to the current answer option, determine whether there is a corresponding next added nuclear question in the added nuclear question tree, if there is a corresponding next added nuclear question in the added nuclear question tree, then according to the current answer option, in the credit core During the call, the corresponding next core addition problem in the core addition problem tree is broadcast. If there is no corresponding next core addition problem in the preset core addition problem tree, the execution of the core addition problem broadcast operation is stopped. For example, the answer options of the core-adding question A are A1 and A2, respectively, and the next core-adding question corresponding to the answer option A1 in the core decision tree is the core-adding question 1, and the next option corresponding to the answer option A2 in the core decision tree The core addition question is the core addition question 2, when the answer option is the core addition question A's answer option A1, the next core addition question is the core addition question 1, otherwise the answer option is the core addition question A's answer option A2, The next nuclear issue is nuclear issue 2.

In this embodiment, when the current nucleus verification customer is suspected of fraud, the present invention can add a check to the current nucleus verification customer to further improve the accuracy rate of anti-fraud identification and the nucleus verification accuracy.

The invention also provides a credit anti-fraud system.

Referring to FIG. 4, FIG. 4 is a schematic diagram of functional modules of the first embodiment of the credit anti-fraud system of the present invention.

In this embodiment, the credit anti-fraud system includes:

The obtaining module 101 is used to obtain the current risk level of the customer of the nuclear body when the credit body call is monitored, and obtain the customer voice data collected during the credit body phone call;

The strategy determination module 102 is used to determine a target anti-fraud strategy based on the risk level, wherein the target anti-fraud strategy is a voiceprint anti-fraud strategy, a voice anti-fraud strategy, a light voice anti-fraud strategy, and a background voice anti-fraud strategy And one or more of voice emotion anti-fraud strategies;

The anti-fraud module 103 is configured to perform an anti-fraud identification operation according to the customer voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current nuclear customer.

Further, the policy determination module 102 is also used to:

Obtain the mapping table between the pre-stored risk level and the strategy identification code;

Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.

Further, the credit anti-fraud system also includes:

The judging module is used for judging whether the current nuclear customer is suspected of fraud based on the anti-fraud identification result;

The risk level adjustment module is used to adjust the current risk level of the current nuclear client based on the anti-fraud identification result if the current nuclear client is suspected of fraud;

The execution module is used to perform fraud risk reminding operations.

Further, the risk level adjustment module is also used to:

According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.

Further, the credit anti-fraud system also includes:

The core adding module is configured to perform an operation for reporting an additional core issue during the credit core call according to the preset core issue question tree when the triggered core core instruction of the current core entity is detected.

Further, the core adding module is also used to:

Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.

The specific embodiments of the credit anti-fraud system of the present invention are basically the same as the above embodiments of the credit anti-fraud method, which will not be repeated here.

In addition, an embodiment of the present invention further proposes a computer-readable storage medium on which a credit anti-fraud program is stored. When the credit anti-fraud program is executed by a processor, the following steps are performed:

When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.

Further, when the credit anti-fraud program is executed by the processor, the following steps are also performed:

Obtain the mapping table between the pre-stored risk level and the strategy identification code;

Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.

Further, when the credit anti-fraud program is executed by the processor, the following steps are also performed:

According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.

Further, when the credit anti-fraud program is executed by the processor, the following steps are also performed:

According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.

Further, when the credit anti-fraud program is executed by the processor, the following steps are also performed:

When the triggered core adding instruction of the current core body client is detected, the core adding issue broadcast operation is performed in the credit core body call according to the preset core adding question tree.

Further, when the credit anti-fraud program is executed by the processor, the following steps are also performed:

Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.

The specific embodiments of the computer-readable storage medium of the present invention are basically the same as the embodiments of the credit anti-fraud method described above, and are not repeated here.

It should be noted that in this article, the terms "include", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system that includes a series of elements includes not only those elements, It also includes other elements that are not explicitly listed, or include elements inherent to this process, method, article, or system. Without more restrictions, the element defined by the sentence "include one..." does not exclude that there are other identical elements in the process, method, article or system that includes the element.

The sequence numbers of the above embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

Through the description of the above embodiments, those skilled in the art can clearly understand that the methods in the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware, but in many cases the former is better Implementation. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product in essence or part that contributes to the existing technology, and the computer software product is stored in a storage medium (such as ROM/RAM) as described above , Disk, CD-ROM), including several instructions to enable a terminal device (which can be a mobile phone, computer, server, air conditioner, or network equipment, etc.) to perform the methods described in various embodiments of the present invention.

The above are only preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by the description and drawings of the present invention, or directly or indirectly used in other related technical fields The same reason is included in the patent protection scope of the present invention.

Claims (20)

1. A credit anti-fraud method, characterized in that the credit anti-fraud method includes the following steps:

   When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

   Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

   Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.
2. The credit anti-fraud method of claim 1, wherein the step of determining a target anti-fraud strategy based on the risk level includes:

   Obtain the mapping table between the pre-stored risk level and the strategy identification code;

   Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.
3. The credit anti-fraud method of claim 1, wherein after performing the anti-fraud identification operation according to the customer voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current self-identified customer ,Also includes:

   According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

   If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.
4. The credit anti-fraud method according to claim 3, wherein the step of adjusting the risk level of the current nuclear customer according to the anti-fraud identification result includes:

   According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

   If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.
5. The credit anti-fraud method of claim 3, wherein after the step of performing the fraud risk reminding operation, the method further comprises:

   When the triggered core adding instruction of the current core body client is detected, the core adding issue broadcast operation is performed in the credit core body call according to the preset core adding question tree.
6. The credit anti-fraud method of claim 5, wherein the step of performing the auditing issue broadcast operation in the credit auditing call according to a preset auditing question tree includes:

   Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

   If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

   If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.
7. A credit anti-fraud system, characterized in that the credit anti-fraud system includes:

   An obtaining module, used to obtain the current risk level of the customer of the nuclear nucleus when the credit nucleus call is monitored, and obtain the customer voice data collected during the credit nucleus call;

   A strategy determination module for determining a target anti-fraud strategy based on the risk level, wherein the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background voice anti-fraud strategy and One or more of voice emotion anti-fraud strategies;

   The anti-fraud module is used to perform an anti-fraud identification operation based on the customer voice data and the target anti-fraud strategy to obtain the anti-fraud identification result of the current nuclear customer.
8. The credit anti-fraud system of claim 7, wherein the policy determination module is further used to:

   Obtain the mapping table between the pre-stored risk level and the strategy identification code;

   Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.
9. The credit anti-fraud system of claim 7, wherein the credit anti-fraud system further comprises:

   The judging module is used for judging whether the current nuclear customer is suspected of fraud based on the anti-fraud identification result;

   The risk level adjustment module is used to adjust the current risk level of the current nuclear client based on the anti-fraud identification result if the current nuclear client is suspected of fraud;

   The execution module is used to perform fraud risk reminding operations.
10. The credit anti-fraud system of claim 9, wherein the risk level adjustment module is further used to:

    According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

    If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.
11. The credit anti-fraud system of claim 9, wherein the credit anti-fraud system further comprises:

    The core adding module is configured to perform an operation for reporting an additional core issue during the credit core call according to the preset core issue question tree when the triggered core core instruction of the current core entity is detected.
12. The credit anti-fraud system of claim 11, wherein the core adding module is further used to:

    Each time when receiving the current answer option selected by the customer based on the currently played core-adding question, it is determined whether there is a corresponding next core-adding question in the core-adding question tree according to the current answer option;

    If there is a corresponding next added core question in the added core question tree, the corresponding next added core question in the added core question tree is broadcasted in the credit core body call according to the current answer option;

    If there is no corresponding next core addition problem in the core addition problem tree, the execution of the core addition problem broadcast operation is stopped.
13. A credit anti-fraud device, characterized in that the credit anti-fraud device includes: a memory, a processor, and a credit anti-fraud program stored on the memory and executable on the processor, the credit anti-fraud When the program is executed by the processor, the following steps are realized:

    When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

    Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

    Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.
14. The credit anti-fraud device of claim 13, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    Obtain the mapping table between the pre-stored risk level and the strategy identification code;

    Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.
15. The credit anti-fraud device of claim 13, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

    If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.
16. The credit anti-fraud device of claim 15, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

    If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.
17. A computer-readable storage medium, characterized in that a credit anti-fraud program is stored on the computer-readable storage medium, and when the credit anti-fraud program is executed by a processor, the following steps are implemented:

    When a credit nucleus call is monitored, the current risk level of the customer with the nucleus check is obtained, and the customer voice data collected during the credit nucleus call is acquired;

    Determine the target anti-fraud strategy based on the risk level, where the target anti-fraud strategy is voiceprint anti-fraud strategy, voice anti-fraud strategy, light voice anti-fraud strategy, background sound anti-fraud strategy and voice emotion anti-fraud strategy One or more of

    Based on the customer voice data and the target anti-fraud strategy, an anti-fraud identification operation is performed to obtain the anti-fraud identification result of the current nuclear customer.
18. The computer-readable storage medium of claim 17, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    Obtain the mapping table between the pre-stored risk level and the strategy identification code;

    Query the mapping relationship table, obtain a policy identification code corresponding to the risk level, and determine an anti-fraud strategy corresponding to the policy identification code as a target anti-fraud strategy.
19. The computer-readable storage medium of claim 17, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    According to the anti-fraud identification result, determine whether the current nuclear customer has suspected fraud;

    If the current nuclear customer is suspected of fraud, the current nuclear customer's risk level is adjusted according to the anti-fraud identification result, and a fraud risk reminder operation is performed.
20. The computer-readable storage medium of claim 19, wherein when the credit anti-fraud program is executed by the processor, the following steps are further implemented:

    According to the anti-fraud identification result, determine the fraud index of the current nuclear customer, and determine whether the fraud index is greater than or equal to a preset threshold;

    If the fraud index is greater than or equal to a preset threshold, the risk level of the current nuclear customer is increased by one level.