WO2018014593A1 - 基于大数据的风险预测方法、装置、服务器及存储介质 - Google Patents

基于大数据的风险预测方法、装置、服务器及存储介质 Download PDF

Info

Publication number
WO2018014593A1
WO2018014593A1 PCT/CN2017/079529 CN2017079529W WO2018014593A1 WO 2018014593 A1 WO2018014593 A1 WO 2018014593A1 CN 2017079529 W CN2017079529 W CN 2017079529W WO 2018014593 A1 WO2018014593 A1 WO 2018014593A1
Authority
WO
WIPO (PCT)
Prior art keywords
voiceprint
risk value
user
information
voiceprint feature
Prior art date
Application number
PCT/CN2017/079529
Other languages
English (en)
French (fr)
Inventor
秦铭雪
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2018014593A1 publication Critical patent/WO2018014593A1/zh

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
    • G06Q40/03—Credit; Loans; Processing thereof
    • 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
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • 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
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/401—Transaction verification
    • G06Q20/4014—Identity check for transactions
    • G06Q20/40145—Biometric identity checks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L9/00—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
    • H04L9/32—Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols including means for verifying the identity or authority of a user of the system or for message authentication, e.g. authorization, entity authentication, data integrity or data verification, non-repudiation, key authentication or verification of credentials

Definitions

  • the present invention relates to the field of computer processing, and in particular, to a method, device, server and storage medium for risk prediction based on big data.
  • a big data based risk prediction method is provided.
  • a big data based risk prediction method including:
  • a big data based risk prediction device comprising:
  • An obtaining module configured to obtain voice information of a user
  • a voiceprint information determining module configured to determine voiceprint information of the user according to the voice information
  • An extraction module configured to extract a voiceprint feature in the voiceprint information
  • a risk value determining module configured to determine a risk value corresponding to the voiceprint feature according to the voiceprint feature
  • a prevention module for performing a corresponding preventive operation based on the risk value.
  • a server includes a memory and a processor, the memory storing instructions that, when executed by the processor, cause the processor to perform the following steps:
  • One or more non-volatile readable storage media storing computer-executable instructions, when executed by one or more processors, cause the one or more processors to perform the following steps:
  • 1 is an application environment diagram of a big data based risk prediction method in an embodiment
  • FIG. 2 is a flow chart of a risk prediction method based on big data in an embodiment
  • FIG. 3 is a flow chart of a method for predicting risk based on big data in another embodiment
  • FIG. 4 is a flow chart of a method for determining voiceprint information of a user according to voice information in an embodiment
  • FIG. 5 is a flow chart of a risk prediction method based on big data in still another embodiment
  • FIG. 6 is a structural block diagram of a big data based risk prediction apparatus in an embodiment
  • FIG. 7 is a structural block diagram of a big data based risk prediction apparatus in another embodiment
  • FIG. 8 is a structural block diagram of a big data based risk prediction apparatus in still another embodiment
  • Figure 9 is a block diagram of a server in one embodiment.
  • server 10 communicates with terminal 20 over a network.
  • the server 10 receives the operation request sent by the terminal 20, and according to the operation request, the receiving terminal 20 acquires the voice information of the user through the voice input interface, determines the voiceprint information of the user according to the voice information, and extracts the voiceprint feature in the voiceprint information according to the voiceprint.
  • the feature determines the risk value corresponding to the voiceprint feature and the corresponding preventive action based on the risk value.
  • the terminal 20 includes, but is not limited to, various personal computers, smart phones, tablet computers, notebook computers, portable wearable devices, etc., which are not enumerated here.
  • a big data-based risk prediction method is proposed, which is applied to a server as described in FIG. 1, and the method includes the following steps:
  • Step 202 Acquire voice information of the user.
  • the server first receives an operation request (such as a loan request) of the user, and then causes the terminal to display an interface for acquiring the user voice information according to the operation request of the user, and the user performs voice input according to the prompt of the interface, and then the server receives the user input.
  • Voice message In addition, in order to prevent others from pretending to use the illegally recorded voice information, a user usually randomly generates a paragraph in the user input voice interface, and asks the user to read the paragraph in a specified time (for example, 1 minute) to obtain the user. Voice message.
  • Step 204 Determine voiceprint information of the user according to the voice information.
  • the server obtains the voiceprint information of the user according to the voice information of the user.
  • Voiceprint information refers to the spectrum of sound waves carrying speech information displayed by electroacoustic instruments.
  • the process of determining voiceprint information based on voice information is the process of acquiring the user's sound wave spectrum.
  • the voice information is generally pre-processed, such as denoising, and then the pre-processed voice information is digitized to obtain voiceprint information of the user.
  • the process of digitization is the process of acquiring the spectrum of the sound wave.
  • Step 206 extracting voiceprint features in the voiceprint information.
  • the voiceprint feature in the voiceprint information is extracted by extracting the feature parameters in the voiceprint information.
  • the feature parameters include pitch feature parameters, pitch strength feature parameters, pitch length feature parameters, and timbre feature parameters.
  • Step 208 determining a risk value corresponding to the voiceprint feature according to the voiceprint feature.
  • the voiceprint big data model is established in advance, and the relationship between the voiceprint feature and the risk value is analyzed according to the voiceprint information of the past malicious behavior. Then, the risk value corresponding to the voiceprint feature of the user is obtained according to the correspondence.
  • the risk value is used to estimate the probability of a user's malicious behavior. For example, if the loan business is involved, the risk value is used to evaluate the probability that the user will not pay back on time after borrowing. If the risk value corresponding to the user is relatively high, it means that the probability that the user does not repay the loan is relatively high, and the loan amount of the user can be reduced to prevent it.
  • the similarity between the voiceprint feature of the user and the voiceprint feature with high risk is calculated. The higher the similarity, the higher the risk value of the user, and vice versa, the lower the similarity. The lower the risk value of the user.
  • Step 210 Perform corresponding prevention operations according to the risk value.
  • different measures are taken according to the risk value to perform corresponding prevention operations. For example, if the user applies for a loan business, then different loan amounts can be set according to different risk values. The higher the risk value, the lower the corresponding loan amount will be set. On the contrary, the corresponding loan amount will be larger.
  • the corresponding relationship between the risk value and the prevention operation is set in advance, and the corresponding prevention operation is performed according to the calculated risk value. According to the calculated risk value, the risk can be predicted, and the risk can be effectively evaluated. Then, according to the risk value, corresponding measures can be taken to prevent the operation, and the purpose of reducing the loss can be achieved.
  • the server receives the operation request of the user, acquires the voice information of the user according to the operation request, determines the voiceprint information of the user according to the voice information, extracts the voiceprint feature in the voiceprint information, and then according to the voiceprint.
  • the feature determines a security risk value corresponding to the voiceprint feature, and then determines whether to allow the user's operation request based on the risk value.
  • the risk prediction method is based on the voiceprint database of big data, and establishes the relationship between the voiceprint feature and the risk value. According to the voiceprint feature of the user, the risk value corresponding to the user can be obtained, and then the corresponding risk is performed according to the risk value. operating. Because the voiceprint has the characteristics of specificity and stability, by establishing the relationship between the voiceprint feature and the risk value, the risk value corresponding to the user can be conveniently obtained, so that the preventive operation can be performed in advance according to the risk value.
  • the method before the step of acquiring voice information of the user, the method further includes:
  • step 201 a voiceprint big data model is established, and the correspondence between the voiceprint feature and the risk value is analyzed.
  • the step of determining a risk value corresponding to the voiceprint feature based on the voiceprint feature comprises determining a risk value corresponding to the voiceprint feature based on a relationship between the voiceprint feature and the risk value.
  • the risk value corresponding to the voiceprint feature is determined, and then the corresponding prevention operation is performed according to the determined risk value. Specifically, it is determined whether the risk value has reached a preset threshold, and if so, a corresponding preventive operation is taken. For example, if the user requests a loan request, the risk can be prevented by reducing the loan amount.
  • the step of determining the voiceprint information of the user according to the voice information includes:
  • Step 204A Perform pre-processing on the obtained voice information.
  • the obtained voice information is first preprocessed. Because the obtained voice information generally has background noise, the obtained voice information needs to be denoised. After the denoising process, other pre-processing can also be performed. For example, the speech information can be segmented and divided into a plurality of speech segments for subsequent processing.
  • Step 204B digitizing the pre-processed voice information to obtain voiceprint information of the user.
  • the preprocessed voice information is digitized, and the digitization process is a process of converting into a sound wave spectrum, thereby obtaining voiceprint information of the user.
  • the analog information is converted into a digital signal for display to obtain the user's sound wave spectrum, and the sound wave spectrum can be further extracted by the sound wave spectrum, for example, the flat-tongue voiceprint feature, the vocal cord voiceprint feature, and the nose are extracted. Sound tone pattern and oral voice pattern features.
  • the method further includes:
  • Step 207 Match the voiceprint feature with the voiceprint in the voiceprint blacklist. If the match is successful, proceed to step 212. If the match fails, proceed to step 208.
  • the voiceprint feature in the voiceprint information is extracted, the voiceprint feature is first matched with the voiceprint in the pre-stored voiceprint blacklist. If the matching is successful, the corresponding operation request is directly rejected.
  • the establishment of a voiceprint blacklist is to record the voiceprints of people who have had malicious behaviors, and to pull the voiceprints with bad records into the blacklist.
  • the blacklist of voiceprints can effectively prevent secondary risks, that is, prevent people with bad records from continuing malicious behavior.
  • the voiceprint feature is first matched with the voiceprint in the voiceprint blacklist. If the match is successful, indicating that the user is already in the blacklist, the user's operation request is directly rejected.
  • the match fails, the user has recorded well before, then the corresponding risk value is determined according to the user's voiceprint feature.
  • voiceprints to identify whether the user has had a bad record is more convenient and efficient than the traditional simply reviewing the submitted material. Because the user's submitted data may be forged, but the user's voiceprint has the characteristics of specificity and stability, it is difficult to forge, so the user's voiceprint is matched with the voiceprint in the voiceprint blacklist. Effectively prevent secondary risks.
  • a user has not repaid the loan before, if he forges his identity information and name for refinancing, it is difficult to find the same person to borrow from the real name information, but because his voiceprint is difficult to forge, Then, after obtaining the voice information of the user, if the voiceprint feature is found to match a voiceprint feature in the blacklist, then the person can be found and the loan request is rejected.
  • step 212 the corresponding operation request is directly rejected.
  • an operation request for directly rejecting the user may be set.
  • the risk value of the user is evaluated only after confirming that the user has no bad records. If the user has a bad record, the corresponding operation request is directly rejected, which can eliminate the secondary risk and better predict the risk.
  • the step of performing a corresponding preventive action based on the risk value includes determining a preventive action corresponding to the risk value based on a correspondence between the pre-established risk value and the preventive action.
  • the correspondence between the risk value and the prevention operation is set, and the higher the risk value, the stricter the corresponding prevention operation is.
  • different loan amounts may be set according to different risk values. For example, if the risk value is between 0-10, the loan amount may be set to 100,000. The risk value is 10-20, setting the loan amount to 80,000; if the risk value is 20-30, set the loan amount to 60,000; if the risk value is 30-50, set the loan amount to 50,000; if the risk value is 60-80, set The loan amount is 30,000 yuan; if the risk value is 80-100, set the loan amount to 10,000. In other words, the higher the risk value, the less the amount of the loan can be.
  • a big data based risk prediction apparatus comprising:
  • the obtaining module 602 is configured to acquire voice information of the user.
  • the server first receives an operation request of the user (such as a loan request), and then displays an interface for obtaining the user voice information on the terminal according to the operation request of the user, and the user inputs the voice according to the prompt of the interface, and then the server receives the user.
  • Voice information entered In addition, in order to prevent others from pretending to use the illegally recorded voice information, a user usually randomly generates a paragraph in the user input voice interface, and asks the user to read the paragraph in a specified time (for example, 1 minute) to obtain the user. Voice message.
  • the voiceprint information determining module 604 is configured to determine voiceprint information of the user according to the voice information.
  • the voiceprint information of the user is obtained according to the voice information of the user.
  • Voiceprint information refers to the spectrum of sound waves carrying speech information displayed by electroacoustic instruments.
  • the process of determining voiceprint information based on voice information is the process of acquiring the user's sound wave spectrum.
  • the voice information is generally pre-processed, such as denoising, and then the pre-processed voice information is digitized to obtain the user's voiceprint information.
  • the process of digitization is the process of acquiring the spectrum of the sound wave.
  • the extraction module 606 is configured to extract a voiceprint feature in the voiceprint information.
  • the voiceprint feature in the voiceprint information is extracted by extracting the feature parameters in the voiceprint information.
  • the feature parameters include pitch feature parameters, pitch strength feature parameters, pitch length feature parameters, and timbre feature parameters.
  • the voiceprint features in the voiceprint information are extracted according to the feature parameters.
  • the risk value determining module 608 is configured to determine a risk value corresponding to the voiceprint feature according to the voiceprint feature.
  • the voiceprint big data model is established in advance, and the relationship between the voiceprint feature and the risk value is analyzed according to the voiceprint information of the past malicious behavior.
  • the risk value corresponding to the user's voiceprint feature is then obtained.
  • the risk value is used to estimate the probability of a user's malicious behavior. For example, if the loan business is involved, the risk value is used to evaluate the probability that the user will not pay back on time after borrowing. If the risk value corresponding to the user is relatively high, it means that the probability that the user does not repay the loan is relatively high, and the loan amount of the user can be reduced to prevent it.
  • the similarity between the voiceprint feature of the user and the voiceprint feature with high risk is calculated. The higher the similarity, the higher the risk value of the user, and vice versa, the lower the similarity. The lower the risk value of the user.
  • the prevention module 610 is configured to perform a corresponding prevention operation according to the risk value.
  • different measures are taken according to the risk value to perform corresponding prevention operations.
  • the corresponding relationship between the risk value and the prevention operation is set in advance, and the corresponding prevention operation is performed according to the calculated risk value.
  • the risk is predicted, and corresponding measures are taken to prevent the operation, which can effectively assess the risk, and take corresponding preventive actions to reduce the loss.
  • the risk prediction device establishes the relationship between the voiceprint feature and the risk value based on the voiceprint database of the big data, and obtains the risk value corresponding to the user according to the voiceprint feature of the user, and then performs corresponding prevention according to the risk value. operating. Because the voiceprint has the characteristics of specificity and stability, by establishing the relationship between the voiceprint feature and the risk value, the risk value corresponding to the user can be conveniently obtained, so that the preventive operation can be performed in advance according to the risk value.
  • the foregoing big data-based risk prediction apparatus further includes:
  • the module 601 is configured to establish a voiceprint big data model, and analyze the correspondence between the voiceprint feature and the risk value.
  • the risk value determining module 608 is further configured to determine a risk value corresponding to the voiceprint feature according to a relationship between the voiceprint feature and a risk value.
  • the risk value corresponding to the voiceprint feature is determined, and then the corresponding prevention operation is performed according to the determined risk value. Specifically, it is determined whether the risk value has reached a preset threshold, and if so, a corresponding preventive operation is taken. For example, if the user requests a loan request, the risk can be prevented by reducing the loan amount.
  • the voiceprint information determining module is further configured to perform pre-processing on the acquired voice information, and digitize the pre-processed voice information to obtain voiceprint information of the user.
  • the obtained voice information is first preprocessed. Because the obtained voice information generally has background noise, the obtained voice information needs to be denoised. After the denoising process, other pre-processing can also be performed. For example, the speech information can be segmented and divided into a plurality of speech segments for subsequent processing.
  • the pre-processed voice information is digitized, and the digitization process is a process of converting into a sound wave spectrum, thereby obtaining user's voiceprint information.
  • the analog information is converted into a digital signal for display to obtain the user's sound wave spectrum, and the sound wave spectrum can be further extracted by the sound wave spectrum, for example, the flat-tongue voiceprint feature, the vocal cord voiceprint feature, and the nose are extracted. Sound tone pattern and oral voice pattern features.
  • the foregoing big data-based risk prediction apparatus further includes:
  • the matching module 607 is configured to match the voiceprint feature with the voiceprint in the voiceprint blacklist. If the matching succeeds, the corresponding operation request is directly rejected. If the matching fails, the notification risk value determining module is configured according to the voiceprint feature. A risk value corresponding to the voiceprint feature is determined.
  • the voiceprint feature in the voiceprint information is extracted, the voiceprint feature is first matched with the voiceprint in the pre-stored voiceprint blacklist. If the matching is successful, the corresponding operation request is directly rejected.
  • the establishment of a voiceprint blacklist is to record the voiceprints of people who have had malicious behaviors, and to pull the voiceprints with bad records into the blacklist.
  • the blacklist of voiceprints can effectively prevent secondary risks, that is, prevent people with bad records from continuing malicious behavior.
  • the voiceprint feature is first matched with the voiceprint in the voiceprint blacklist. If the match is successful, indicating that the user is already in the blacklist, the user's operation request is directly rejected.
  • the match fails, the user has recorded well before, then the corresponding risk value is determined according to the user's voiceprint feature.
  • voiceprints to identify whether the user has had a bad record is more convenient and efficient than the traditional simply reviewing the submitted material. Because the user's submitted data may be forged, but the user's voiceprint has the characteristics of specificity and stability, it is difficult to forge, so the user's voiceprint is matched with the voiceprint in the voiceprint blacklist. Effectively prevent secondary risks.
  • a user has not repaid the loan before, if he forges his identity information and name for refinancing, it is difficult to find the same person to borrow from the real name information, but because his voiceprint is difficult to forge, Then, after obtaining the voice information of the user, if the voiceprint feature is found to match a voiceprint feature in the blacklist, then the person can be found and the loan request is rejected.
  • the prevention module is further configured to determine a prevention operation corresponding to the risk value according to a correspondence between the pre-established risk value and the prevention operation.
  • the correspondence between the risk value and the prevention operation is set, and the higher the risk value, the stricter the corresponding prevention operation is.
  • different loan amounts may be set according to different risk values. For example, if the risk value is between 0-10, the loan amount may be set to 100,000. The risk value is 10-20, setting the loan amount to 80,000; if the risk value is 20-30, set the loan amount to 60,000; if the risk value is 30-50, set the loan amount to 50,000; if the risk value is 60-80, set The loan amount is 30,000 yuan; if the risk value is 80-100, set the loan amount to 10,000. In other words, the higher the risk value, the less the amount of the loan can be.
  • the network interface may be an Ethernet card or a wireless network card.
  • the above modules may be embedded in the hardware in the processor or in the memory in the server, or may be stored in the memory in the server, so that the processor calls the corresponding operations of the above modules.
  • the processor can be a central processing unit (CPU), a microprocessor, a microcontroller, or the like.
  • a block diagram of the server 10 of FIG. 1 is illustrated, the server 10 including a processor coupled via a system bus, a non-volatile storage medium, an internal memory, and a network interface.
  • the non-volatile storage medium of the server 10 stores an operating system and computer executable instructions for implementing a big data-based risk prediction method suitable for the server 10.
  • This processor is used to provide computing and control capabilities to support the operation of the entire server.
  • the internal memory in server 10 provides an environment for the operation of an operating system and computer executable instructions in a non-volatile storage medium for network communication with the terminal. It will be understood by those skilled in the art that the structure shown in FIG.
  • server 10 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation of the server 10 to which the solution of the present application is applied.
  • the specific server 10 may It includes more or fewer components than those shown in the figures, or some components are combined, or have different component arrangements.
  • the processor when the computer executable instructions in the server of FIG. 9 are executed by the processor, causing the processor to perform the following steps to implement the above-described big data-based risk prediction method: acquiring voice information of the user; The voice information determines the voiceprint information of the user; extracts the voiceprint feature in the voiceprint information; determines a risk value corresponding to the voiceprint feature according to the voiceprint feature; and performs corresponding prevention according to the risk value operating.
  • the processor before the acquiring the voice information of the user, the processor further performs the steps of: establishing a voiceprint big data model, and analyzing a correspondence between the voiceprint feature and the risk value; Determining, according to the voiceprint feature, the risk value corresponding to the voiceprint feature comprises: determining a risk value corresponding to the voiceprint feature according to a relationship between the voiceprint feature and a risk value.
  • the determining, by the processor in the server, the voiceprint information of the user according to the voice information comprises: pre-processing the obtained voice information; and performing the pre-processed voice The information is digitized to obtain the voiceprint information of the user.
  • the processor in the server further performs the following steps: matching the voiceprint feature with the voiceprint in the voiceprint blacklist If the matching is successful, the corresponding operation request is directly rejected. If the matching fails, the risk value corresponding to the voiceprint feature is determined according to the voiceprint feature.
  • the performing the preventing operation according to the risk value performed by the processor in the server includes: determining, according to a correspondence between a pre-established risk value and a prevention operation, The preventive action corresponding to the risk value.
  • the storage medium may be a magnetic disk, an optical disk, or a read-only storage memory (Read-Only)
  • a nonvolatile storage medium such as a memory or a ROM, or a random access memory (RAM).

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Computer Security & Cryptography (AREA)
  • Finance (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Signal Processing (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Technology Law (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种基于大数据的风险预测方法包括：获取用户的语音信息（202），根据所述语音信息确定用户的声纹信息（204），提取所述声纹信息中的声纹特征（206），根据所述声纹特征确定与所述声纹特征对应的风险数值（208），根据所述风险数值进行相应的预防操作（210）。

Description

基于大数据的风险预测方法、装置、服务器及存储介质
本申请要求于2016年7月20日提交中国专利局、申请号为2016105787987、发明名称为“基于大数据的风险预测方法和装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
【技术领域】
本发明涉及计算机处理领域，特别是涉及一种基于大数据的风险预测方法、装置、服务器及存储介质。
【背景技术】
随着互联网的发展，越来越多的业务从线下转移到了线上，在网上办理借贷等业务时， 如何进行反欺诈、防恶意贷款等成为了银行等金融机构的关注重点，传统的只是对用户提交的资料的真实性进行审查，但对其贷款后是否会及时还款往往是根据工作人员的主观经验来进行判断，并不能够有效的进行风险预测，所以目前亟待需要一种能够有效对风险预测的方法来减少恶意贷款等行为带来的损失。
【发明内容】
根据本申请的各种实施例，提供一种基于大数据的风险预测方法、装置、服务器及存储介质。
一种基于大数据的风险预测方法，包括：
获取用户的语音信息；
根据所述语音信息确定用户的声纹信息；
提取所述声纹信息中的声纹特征；
根据所述声纹特征确定与所述声纹特征对应的风险数值；及
根据所述风险数值进行相应的预防操作。
一种基于大数据的风险预测装置，包括：
获取模块，用于获取用户的语音信息；
声纹信息确定模块，用于根据所述语音信息确定用户的声纹信息；
提取模块，用于提取所述声纹信息中的声纹特征；
风险数值确定模块，用于根据所述声纹特征确定与所述声纹特征对应的风险数值；及
预防模块，用于根据所述风险数值进行相应的预防操作。
一种服务器，包括存储器和处理器，所述存储器中存储有指令，所述指令被所述处理器执行时，使得所述处理器执行以下步骤：
获取用户的语音信息；
根据所述语音信息确定用户的声纹信息；
提取所述声纹信息中的声纹特征；
根据所述声纹特征确定与所述声纹特征对应的风险数值；及
根据所述风险数值进行相应的预防操作。
一个或多个存储有计算机可执行指令的非易失性可读存储介质，所述计算机可执行指令被一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
获取用户的语音信息；
根据所述语音信息确定用户的声纹信息；
提取所述声纹信息中的声纹特征；
根据所述声纹特征确定与所述声纹特征对应的风险数值；及
根据所述风险数值进行相应的预防操作。
本发明的一个或多个实施例的细节在下面的附图和描述中提出。本发明的其它特征、目的和优点将从说明书、附图以及权利要求书变得明显。
【附图说明】
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为一个实施例中基于大数据的风险预测方法的应用环境图；
图2为一个实施例中基于大数据的风险预测方法流程图；
图3为另一个实施例中基于大数据的风险预测方法流程图；
图4为一个实施例中根据语音信息确定用户的声纹信息的方法流程图；
图5为又一个实施例中基于大数据的风险预测方法流程图；
图6为一个实施例中基于大数据的风险预测装置的结构框图；
图7为另一个实施例中基于大数据的风险预测装置的结构框图；
图8为又一个实施例中基于大数据的风险预测装置的结构框图；
图9为一个实施例中服务器的框图。
【具体实施方式】
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
本申请实施例所提供的基于大数据的风险预测方法可应用于如图1所示的应用环境中。参考图1，服务器10通过网络与终端20进行通信。服务器10接收终端20发送的操作请求，根据该操作请求接收终端20通过语音录入界面获取用户的语音信息，根据语音信息确定用户的声纹信息，提取声纹信息中的声纹特征，根据声纹特征确定与声纹特征对应的风险数值，以及根据风险数值进行相应的预防操作。可以理解，终端20包括但不限于各种个人计算机、智能手机、平板电脑、笔记本电脑、便携式穿戴设备等，在此不一一列举。
如图2所示，在一个实施例中，提出了一种基于大数据的风险预测方法，应用于如图1所述的服务器中，所述方法包括以下步骤：
步骤202，获取用户的语音信息。
在本实施例中，服务器首先接收用户的操作请求（比如借贷请求），然后根据用户的操作请求使终端展示获取用户语音信息的界面，用户根据界面的提示进行语音的录入，然后服务器接收用户录入的语音信息。此外，为了防止他人用之前非法已录好的语音信息进行冒充，一般会在用户录入语音界面随机生成一段话，要求用户在规定的时间内（比如1分钟）朗读完该段话，从而获取用户的语音信息。
步骤204，根据语音信息确定用户的声纹信息。
在本实施例中，服务器获取到用户的语音信息后，根据用户的语音信息获取该用户的声纹信息。声纹信息是指用电声学仪器显示的携带言语信息的声波频谱。根据语音信息确定声纹信息的过程也就是获取用户的声波频谱的过程。具体地，服务器获取到用户的语音信息后，一般需要对该语音信息进行预处理，比如去噪声处理，然后将预处理后的语音信息进行数字化得到用户的声纹信息。数字化的过程就是获取声波频谱的过程。
步骤206，提取声纹信息中的声纹特征。
在本实施例中，确定用户的声纹信息后，通过提取声纹信息中的特征参数来提取声纹信息中的声纹特征。其中，特征参数包括音高特征参数，音强特征参数、音长特征参数、音色特征参数等。
步骤208，根据声纹特征确定与声纹特征对应的风险数值。
在本实施例中，预先建立声纹大数据模型，根据以往有过恶意行为的声纹信息分析出声纹特征与风险数值之间的关系。然后根据该对应关系获取与用户的声纹特征对应的风险数值。风险数值用于评估用户发生恶意行为的概率，比如，若是涉及借贷业务，此时风险数值就是用于评估用户借款后不按时还款的概率。若用户对应的风险数值比较高，说明该用户不还款的概率就比较高，可以通过降低对该用户的借贷额度来进行预防。具体的，通过对大量的有过不良记录的声纹信息进行大数据分析，找出其共同具有的某些特征，然后将这些特征作为参考的依据。当获取到用户的声纹特征后，计算用户的声纹特征与具有高风险的声纹特征之间的相似度，相似度越高，说明该用户的风险数值越高，反之，相似度越低，该用户的风险数值越低。
步骤210，根据风险数值进行相应的预防操作。
在本实施例中，确定完风险数值后，根据该风险数值采取不同的措施进行相应的预防操作。举例说明，若用户申请的是借贷业务，那么就可以根据不同的风险数值设置不同的借贷额度。风险数值越高，相应设置的借贷额度就会越低，反之，相应的借贷额度就会越大。预先设置风险数值与预防操作的对应关系，根据计算得到的风险数值来进行相应的预防操作。根据计算得到的风险数值对风险进行预测，可以有效的评估风险，然后根据风险数值采取相应的措施进行预防操作，可以达到减少损失的目的。
在本实施例中，服务器通过接收用户的操作请求，根据该操作请求获取用户的语音信息，根据语音信息确定用户的声纹信息，提取该声纹信息中的声纹特征，然后根据该声纹特征确定与该声纹特征对应的安全风险数值，然后根据该风险数值确定是否允许用户的操作请求。该风险预测方法基于大数据的声纹数据库，建立了声纹特征与风险数值之间的关系，根据用户的声纹特征就可以获取该用户对应的风险数值，然后根据该风险数值进行相应的预防操作。因为声纹具有特定性和稳定性的特点，通过建立声纹特征与风险数值之间的关系，可以便捷的获取用户对应的风险数值，从而可以根据该风险数值提前进行预防操作。
如图3所示，在一个实施例中，在获取用户的语音信息的步骤之前所述方法还包括：
步骤201，建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系。
在本实施例中，在获取用户的语音信息之前首先需要建立一个声纹大数据模型，根据以往的有过恶意行为的声纹信息，提取出共有的一些声纹特征，并分析出声纹特征与风险数值之间的对应关系。比如，若通过分析发现声纹细小、语速过快的人更容易贷款后不归还，那么在放贷的时候就可以降低对这类用户的贷款额度。
在一个实施例中，根据声纹特征确定与声纹特征对应的风险数值的步骤包括：根据声纹特征与风险数值之间的关系，确定与声纹特征对应的风险数值。
在本实施例中，根据预先建立的声纹特征与风险数值之间的关系，确定与该声纹特征对应的风险数值，然后根据确定的风险数值进行相应的预防操作。具体地，判断风险数值是否达到了预设的阈值，若是，则采取相应的预防操作，比如，若用户请求的是贷款请求，那么就可以通过降低贷款额度来预防风险。
如图4所示，在一个实施例中，根据语音信息确定用户的声纹信息的步骤包括：
步骤204A，对获取到的语音信息进行预处理。
在本实施例中，首先对获取到的语音信息进行预处理。因为获取到的语音信息一般都会有背景噪声，所以需要对该获取到的语音信息进行去噪处理。进行去噪处理后还可以进行其他预处理，比如，可以对语音信息进行分割处理，分成多个语音片段，便于后续处理。
步骤204B，对预处理后的语音信息进行数字化得到用户的声纹信息。
在本实施例中，对经过预处理的语音信息进行数字化，数字化的过程就是转换为声波频谱的过程，从而得到用户的声纹信息。具体的，通过进行数模转换即将模拟信息转换为数字信号进行显示来获取用户的声波频谱，通过该声波频谱可以进一步提取声纹特征，比如，提取平舌声纹特征，声带声纹特征、鼻音声纹特征和口腔声纹特征等。
如图3所示，在一个实施例中，在提取声纹信息中的声纹特征的步骤之后还包括：
步骤207，将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则进入步骤212，若匹配失败，则进入步骤208。
在本实施例中，提取声纹信息中的声纹特征后，首先将该声纹特征与预先存储的声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求。建立声纹黑名单就是将有过恶意行为的人的声纹进行记录，即将有过不良记录的声纹拉入黑名单。通过该声纹黑名单可以有效的防止二次风险，即防止有不良记录的人继续进行恶意行为。具体的，首先将声纹特征与声纹黑名单中的声纹进行匹配，如果匹配成功，说明该用户已经在黑名单内，则直接拒绝该用户的操作请求。若匹配失败，说明该用户之前记录良好，那么就继续根据用户的声纹特征确定相应的风险数值。利用声纹来识别该用户是否有过不良记录比传统的只是简单审查提交的资料更加便捷和有效。因为由于用户的提交的资料有可能有伪造，但是用户的声纹具有特定性和稳定性的特点，很难进行伪造，所以将用户的声纹与声纹黑名单中的声纹进行匹配，可以有效的防止发生二次风险。比如，若某用户之前贷款未还，如果他为了再次贷款伪造了自己的身份信息和名字，那么从实名资料就很难发现是同一个人来贷款，但是由于他的声纹是很难伪造的，那么当获取到该用户的语音信息后，若发现其声纹特征与黑名单中的某个声纹特征匹配，那么就可以发现这个人，从而拒绝他的贷款请求。
步骤212，直接拒绝相应的操作请求。
在本实施例中，若发现用户的声纹特征与之前黑名单中声纹特征一致，说明该用户之前有过不良记录，那么为了防止二次风险，可以设置直接拒绝该用户的操作请求。通过预先将用户的声纹与黑名单中的声纹进行匹配，只有确认该用户没有不良记录后再对该用户进行风险数值的评估。若该用户有不良记录，就直接拒绝相应的操作请求，这样能够排除二次风险，更好的对风险进行预测。
在一个实施例中，根据风险数值进行相应的预防操作的步骤包括：根据预先建立的风险数值与预防操作之间的对应关系，确定与风险数值对应的预防操作。
在本实施例中，为了能够有效的对风险进行预防，设置风险数值与预防操作之间的对应关系，风险数值越高，对应的预防操作对应的越严格。具体的，比如，如果是用户的操作请求为借贷，那么可以根据不同的风险数值设置不同的借贷金额，举例说明，若风险数值为0-10之间，就可以设置借贷金额为10万，如果风险数值在10-20，设置借贷金额8万；若风险数值在20-30，设置借贷金额6万；若风险数值在30-50，设置借贷金额5万；若风险数值在60-80，设置借贷金额3万；若风险数值在80-100，设置借贷金额1万。也就是说，风险数值越高，相应能够贷款的数额就越少。
如图6所示，在一个实施例中，提出了一种基于大数据的风险预测装置，所述装置包括：
获取模块602，用于获取用户的语音信息。
在本实施例中，服务器首先接收用户的操作请求（比如借贷请求），然后根据用户的操作请求在终端上展示获取用户语音信息的界面，用户根据界面的提示进行语音的录入，然后服务器接收用户录入的语音信息。此外，为了防止他人用之前非法已录好的语音信息进行冒充，一般会在用户录入语音界面随机生成一段话，要求用户在规定的时间内（比如1分钟）朗读完该段话，从而获取用户的语音信息。
声纹信息确定模块604，用于根据语音信息确定用户的声纹信息。
在本实施例中，获取到用户的语音信息后，根据用户的语音信息获取该用户的声纹信息。声纹信息是指用电声学仪器显示的携带言语信息的声波频谱。根据语音信息确定声纹信息的过程也就是获取用户的声波频谱的过程。具体的，获取到用户的语音信息后，一般需要对该语音信息进行预处理，比如去噪声处理，然后将预处理后的语音信息进行数字化得到用户的声纹信息。数字化的过程就是获取声波频谱的过程。
提取模块606，用于提取声纹信息中的声纹特征。
在本实施例中，确定用户的声纹信息后，通过提取声纹信息中的特征参数来提取声纹信息中的声纹特征。其中，特征参数包括音高特征参数，音强特征参数、音长特征参数、音色特征参数等。根据特征参数对声纹信息中的声纹特征进行提取。
风险数值确定模块608，用于根据所述声纹特征确定与所述声纹特征对应的风险数值。
在本实施例中，预先建立声纹大数据模型，根据以往有过恶意行为的声纹信息分析出声纹特征与风险数值之间的关系。然后获取与用户的声纹特征对应的风险数值。风险数值用于评估用户发生恶意行为的概率，比如，若是涉及借贷业务，此时风险数值就是用于评估用户借款后不按时还款的概率。若用户对应的风险数值比较高，说明该用户不还款的概率就比较高，可以通过降低对该用户的借贷额度来进行预防。具体的，通过对大量的有过不良记录的声纹信息进行大数据分析，找出其共同具有的某些特征，然后将这些特征作为参考的依据。当获取到用户的声纹特征后，计算用户的声纹特征与有高风险的声纹特征之间的相似度，相似度越高，说明该用户的风险数值越高，反之，相似度越低，该用户的风险数值越低。
预防模块610，用于根据风险数值进行相应的预防操作。
在本实施例中，确定完风险数值后，根据该风险数值采取不同的措施进行相应的预防操作。预先设置风险数值与预防操作的对应关系，根据计算得到的风险数值来进行相应的预防操作。根据计算得到的风险数值对风险进行预测，同时采取相应的措施进行预防操作，可以有效的评估风险，同时采取相应的预防操作可以减少损失。
在本实施例中，通过接收用户的操作请求，根据该操作请求获取用户的语音信息，根据语音信息确定用户的声纹信息，提取该声纹信息中的声纹特征，然后根据该声纹特征确定与该声纹特征对应的安全风险数值，然后根据该风险数值确定是否允许用户的操作请求。该风险预测装置基于大数据的声纹数据库，建立了声纹特征与风险数值之间的关系，根据用户的声纹特征就可以获取该用户对应的风险数值，然后根据该风险数值进行相应的预防操作。因为声纹具有特定性和稳定性的特点，通过建立声纹特征与风险数值之间的关系，可以便捷的获取用户对应的风险数值，从而可以根据该风险数值提前进行预防操作。
如图7所示，在一个实施例中，上述基于大数据的风险预测装置还包括：
建立模块601，用于建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系。
在本实施例中，在获取用户的语音信息之前首先需要建立一个声纹大数据模型，根据以往的有过恶意行为的声纹信息，提取出共有的一些声纹特征，并分析出声纹特征与风险数值之间的对应关系。比如，若通过分析发现声纹细小、语速过快的人更容易贷款后不归还，那么在放贷的时候就可以降低对这类用户的贷款额度。
风险数值确定模块608还用于根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
在本实施例中，根据预先建立的声纹特征与风险数值之间的关系，确定与该声纹特征对应的风险数值，然后根据确定的风险数值进行相应的预防操作。具体的，判断风险数值是否达到了预设的阈值，若是，则采取相应的预防操作，比如，若用户请求的是贷款请求，那么就可以通过降低贷款额度来预防风险。
在一个实施例中，声纹信息确定模块还用于对获取到的语音信息进行预处理，对预处理后的语音信息进行数字化得到用户的声纹信息。
在本实施例中，首先对获取到的语音信息进行预处理。因为获取到的语音信息一般都会有背景噪声，所以需要对该获取到的语音信息进行去噪处理。进行去噪处理后还可以进行其他预处理，比如，可以对语音信息进行分割处理，分成多个语音片段，便于后续处理。对经过预处理的语音信息进行数字化，数字化的过程就是转换为声波频谱的过程，从而得到用户的声纹信息。具体的，通过进行数模转换即将模拟信息转换为数字信号进行显示来获取用户的声波频谱，通过该声波频谱可以进一步提取声纹特征，比如，提取平舌声纹特征，声带声纹特征、鼻音声纹特征和口腔声纹特征等。
如图8所示，在一个实施例中，上述基于大数据的风险预测装置还包括：
匹配模块607，用于将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则通知风险数值确定模块根据所述声纹特征确定与所述声纹特征对应的风险数值。
在本实施例中，提取声纹信息中的声纹特征后，首先将该声纹特征与预先存储的声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求。建立声纹黑名单就是将有过恶意行为的人的声纹进行记录，即将有过不良记录的声纹拉入黑名单。通过该声纹黑名单可以有效的防止二次风险，即防止有不良记录的人继续进行恶意行为。具体的，首先将声纹特征与声纹黑名单中的声纹进行匹配，如果匹配成功，说明该用户已经在黑名单内，则直接拒绝该用户的操作请求。若匹配失败，说明该用户之前记录良好，那么就继续根据用户的声纹特征确定相应的风险数值。利用声纹来识别该用户是否有过不良记录比传统的只是简单审查提交的资料更加便捷和有效。因为由于用户的提交的资料有可能有伪造，但是用户的声纹具有特定性和稳定性的特点，很难进行伪造，所以将用户的声纹与声纹黑名单中的声纹进行匹配，可以有效的防止发生二次风险。比如，若某用户之前贷款未还，如果他为了再次贷款伪造了自己的身份信息和名字，那么从实名资料就很难发现是同一个人来贷款，但是由于他的声纹是很难伪造的，那么当获取到该用户的语音信息后，若发现其声纹特征与黑名单中的某个声纹特征匹配，那么就可以发现这个人，从而拒绝他的贷款请求。
在一个实施例中，预防模块还用于根据预先建立的风险数值与预防操作之间的对应关系，确定与风险数值对应的预防操作。
在本实施例中，为了能够有效的对风险进行预防，设置风险数值与预防操作之间的对应关系，风险数值越高，对应的预防操作对应的越严格。具体的，比如，如果是用户的操作请求为借贷，那么可以根据不同的风险数值设置不同的借贷金额，举例说明，若风险数值为0-10之间，就可以设置借贷金额为10万，如果风险数值在10-20，设置借贷金额8万；若风险数值在20-30，设置借贷金额6万；若风险数值在30-50，设置借贷金额5万；若风险数值在60-80，设置借贷金额3万；若风险数值在80-100，设置借贷金额1万。也就是说，风险数值越高，相应能够贷款的数额就越少。
上述基于大数据的风险预测装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。其中，网络接口可以是以太网卡或无线网卡等。上述各模块可以硬件形式内嵌于或独立于服务器中的处理器中，也可以以软件形式存储于服务器中的存储器中，以便于处理器调用执行以上各个模块对应的操作。该处理器可以为中央处理单元（CPU)、微处理器、单片机等。
如图9所示，在一个实施例中，示出了图1中服务器10的框图，该服务器10包括通过***总线连接的处理器、非易失性存储介质、内存储器和网络接口。其中，该服务器10的非易失性存储介质存储有操作***和计算机可执行指令，该计算机可执行指令用于实现适用于服务器10的一种基于大数据的风险预测方法。该处理器用于提供计算和控制能力，支撑整个服务器的运行。服务器10中的内存储器为非易失性存储介质中的操作***和计算机可执行指令的运行提供环境，该服务器10的网络接口用于与终端进行网络通信。本领域技术人员可以理解，图10中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的服务器10的限定，具体的服务器10可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一个实施例中，图9服务器中的所述计算机可执行指令被处理器执行时，使得处理器执行以下步骤以实现上述一种基于大数据的风险预测方法：获取用户的语音信息；根据所述语音信息确定用户的声纹信息；提取所述声纹信息中的声纹特征；根据所述声纹特征确定与所述声纹特征对应的风险数值；及根据所述风险数值进行相应的预防操作。
在一个实施例中，在所述获取用户的语音信息之前，上述处理器还执行以下步骤：建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系；所述处理器所执行的所述根据所述声纹特征确定与所述声纹特征对应的风险数值包括：根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
在一个实施例中，上述服务器中的处理器所执行的所述根据所述语音信息确定用户的声纹信息包括：对获取到的所述语音信息进行预处理；对预处理后的所述语音信息进行数字化得到所述用户的声纹信息。
在一个实施例中，在所述提取所述声纹信息中的声纹特征之后，上述服务器中的所述处理器还执行以下步骤：将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则执行根据所述声纹特征确定与所述声纹特征对应的风险数值。
在一个实施例中，上述服务器中的所述处理器所执行的所述根据所述风险数值进行相应的预防操作包括：根据预先建立的风险数值与预防操作之间的对应关系，确定与所述风险数值对应的预防操作。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，所述的存储介质可为磁碟、光盘、只读存储记忆体（Read-Only Memory，ROM）等非易失性存储介质，或随机存储记忆体（Random Access Memory，RAM）等。
以上所述实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本发明的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对本发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本发明构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。因此，本发明专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种基于大数据的风险预测方法，包括：
    获取用户的语音信息；
    根据所述语音信息确定用户的声纹信息；
    提取所述声纹信息中的声纹特征；
    根据所述声纹特征确定与所述声纹特征对应的风险数值；及
    根据所述风险数值进行相应的预防操作。
  2. 根据权利要求1所述的方法，其特征在于，在所述获取用户的语音信息之前还包括：建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系；
    所述根据所述声纹特征确定与所述声纹特征对应的风险数值包括：根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
  3. 根据权利要求1所述的方法，其特征在于，所述根据所述语音信息确定用户的声纹信息包括：
    对获取到的所述语音信息进行预处理；
    对预处理后的所述语音信息进行数字化得到所述用户的声纹信息。
  4. 根据权利要求1所述的方法，其特征在于，在所述提取所述声纹信息中的声纹特征之后还包括：
    将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则进入根据所述声纹特征确定与所述声纹特征对应的风险数值。
  5. 根据权利要求1所述的方法，其特征在于，所述根据所述风险数值进行相应的预防操作包括：
    根据预先建立的风险数值与预防操作之间的对应关系，确定与所述风险数值对应的预防操作。
  6. 一种基于大数据的风险预测装置，包括：
    获取模块，用于获取用户的语音信息；
    声纹信息确定模块，用于根据所述语音信息确定用户的声纹信息；
    提取模块，用于提取所述声纹信息中的声纹特征；
    风险数值确定模块，用于根据所述声纹特征确定与所述声纹特征对应的风险数值；及
    预防模块，用于根据所述风险数值进行相应的预防操作。
  7. 根据权利要求6所述的装置，其特征在于，所述装置还包括：
    建立模块，用于建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系；
    所述风险数值确定模块还用于根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
  8. 根据权利要求6所述的装置，其特征在于，所述声纹信息确定模块还用于对获取到的所述语音信息进行预处理，对预处理后的所述语音信息进行数字化得到所述用户的声纹信息。
  9. 根据权利要求6所述的装置，其特征在于，所述装置还包括：
    匹配模块，用于将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则通知风险数值确定模块根据所述声纹特征确定与所述声纹特征对应的风险数值。
  10. 根据权利要求6所述的装置，其特征在于，所述预防模块还用于根据预先建立的风险数值与预防操作之间的对应关系，确定与所述风险数值对应的预防操作。
  11. 一种服务器，包括存储器和处理器，所述存储器中存储有指令，所述指令被所述处理器执行时，使得所述处理器执行以下步骤：
    获取用户的语音信息；
    根据所述语音信息确定用户的声纹信息；
    提取所述声纹信息中的声纹特征；
    根据所述声纹特征确定与所述声纹特征对应的风险数值；及
    根据所述风险数值进行相应的预防操作。
  12. 根据权利要求11所述的服务器，其特征在于，在所述获取用户的语音信息之前，所述处理器还执行以下步骤：
    建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系；
    所述处理器所执行的所述根据所述声纹特征确定与所述声纹特征对应的风险数值包括：根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
  13. 根据权利要求11所述的服务器，其特征在于，所述处理器所执行的所述根据所述语音信息确定用户的声纹信息包括：
    对获取到的所述语音信息进行预处理；
    对预处理后的所述语音信息进行数字化得到所述用户的声纹信息。
  14. 根据权利要求11所述的服务器，其特征在于，在所述提取所述声纹信息中的声纹特征之后，所述处理器还执行以下步骤：
    将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则执行根据所述声纹特征确定与所述声纹特征对应的风险数值。
  15. 根据权利要求11所述的服务器，其特征在于，所述处理器所执行的所述根据所述风险数值进行相应的预防操作包括：
    根据预先建立的风险数值与预防操作之间的对应关系，确定与所述风险数值对应的预防操作。
  16. 一个或多个存储有计算机可执行指令的非易失性可读存储介质，所述计算机可执行指令被一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取用户的语音信息；
    根据所述语音信息确定用户的声纹信息；
    提取所述声纹信息中的声纹特征；
    根据所述声纹特征确定与所述声纹特征对应的风险数值；及
    根据所述风险数值进行相应的预防操作。
  17. 根据权利要求16所述的非易失性可读存储介质，其特征在于，在所述获取用户的语音信息之前，所述处理器还执行以下步骤：
    建立声纹大数据模型，分析出声纹特征与风险数值之间的对应关系；
    所述处理器所执行的所述根据所述声纹特征确定与所述声纹特征对应的风险数值包括：根据所述声纹特征与风险数值之间的关系，确定与所述声纹特征对应的风险数值。
  18. 根据权利要求16所述的非易失性可读存储介质，其特征在于，所述处理器所执行的所述根据所述语音信息确定用户的声纹信息包括：
    对获取到的所述语音信息进行预处理；
    对预处理后的所述语音信息进行数字化得到所述用户的声纹信息。
  19. 根据权利要求16所述的非易失性可读存储介质，其特征在于，在所述提取所述声纹信息中的声纹特征之后，所述处理器还执行以下步骤：
    将声纹特征与声纹黑名单中的声纹进行匹配，若匹配成功，则直接拒绝相应的操作请求，若匹配失败，则执行根据所述声纹特征确定与所述声纹特征对应的风险数值。
  20. 根据权利要求16所述的非易失性可读存储介质，其特征在于，所述处理器所执行的所述根据所述风险数值进行相应的预防操作包括：
    根据预先建立的风险数值与预防操作之间的对应关系，确定与所述风险数值对应的预防操作。
PCT/CN2017/079529 2016-07-20 2017-04-06 基于大数据的风险预测方法、装置、服务器及存储介质 WO2018014593A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610578798.7 2016-07-20
CN201610578798.7A CN107025597A (zh) 2016-07-20 2016-07-20 基于大数据的风险预测方法和装置

Publications (1)

Publication Number Publication Date
WO2018014593A1 true WO2018014593A1 (zh) 2018-01-25

Family

ID=59524411

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/079529 WO2018014593A1 (zh) 2016-07-20 2017-04-06 基于大数据的风险预测方法、装置、服务器及存储介质

Country Status (2)

Country Link
CN (1) CN107025597A (zh)
WO (1) WO2018014593A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107977776B (zh) * 2017-11-14 2021-05-11 重庆小雨点小额贷款有限公司 信息处理方法、装置、服务器及计算机可读存储介质
CN108681934A (zh) * 2018-04-25 2018-10-19 厦门快商通信息技术有限公司 一种交易平台及其不良用户的识别方法
CN109598410A (zh) * 2018-10-31 2019-04-09 平安科技（深圳）有限公司 预售风险评估方法、***、计算机装置及可读存储介质
CN113409050B (zh) * 2021-05-06 2022-05-17 支付宝(杭州)信息技术有限公司 基于用户操作判断业务风险的方法和装置

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101262524A (zh) * 2008-04-23 2008-09-10 沈阳东软软件股份有限公司 垃圾语音过滤的方法及***
US20110224986A1 (en) * 2008-07-21 2011-09-15 Clive Summerfield Voice authentication systems and methods
CN103915096A (zh) * 2014-04-15 2014-07-09 胡上杰 警务声纹识别方法
CN104881783A (zh) * 2015-05-14 2015-09-02 中国科学院信息工程研究所 电子银行账户欺诈行为及风险检测方法与***
CN105100363A (zh) * 2015-06-29 2015-11-25 小米科技有限责任公司 信息处理方法、装置及终端
CN105357006A (zh) * 2014-08-20 2016-02-24 中兴通讯股份有限公司 一种基于声纹特征进行安全认证的方法及设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101262524A (zh) * 2008-04-23 2008-09-10 沈阳东软软件股份有限公司 垃圾语音过滤的方法及***
US20110224986A1 (en) * 2008-07-21 2011-09-15 Clive Summerfield Voice authentication systems and methods
CN103915096A (zh) * 2014-04-15 2014-07-09 胡上杰 警务声纹识别方法
CN105357006A (zh) * 2014-08-20 2016-02-24 中兴通讯股份有限公司 一种基于声纹特征进行安全认证的方法及设备
CN104881783A (zh) * 2015-05-14 2015-09-02 中国科学院信息工程研究所 电子银行账户欺诈行为及风险检测方法与***
CN105100363A (zh) * 2015-06-29 2015-11-25 小米科技有限责任公司 信息处理方法、装置及终端

Also Published As

Publication number Publication date
CN107025597A (zh) 2017-08-08

Similar Documents

Publication Publication Date Title
US10249304B2 (en) Method and system for using conversational biometrics and speaker identification/verification to filter voice streams
WO2020034526A1 (zh) 保险录音的质检方法、装置、设备和计算机存储介质
WO2018014593A1 (zh) 基于大数据的风险预测方法、装置、服务器及存储介质
US10372891B2 (en) System and method for identifying special information verbalization timing with the aid of a digital computer
US10135818B2 (en) User biological feature authentication method and system
WO2017215558A1 (zh) 一种声纹识别方法和装置
WO2020062642A1 (zh) 基于区块链的电子合同签署方法、装置、设备及存储介质
WO2020207035A1 (zh) 骚扰电话拦截方法、装置、设备及存储介质
WO2013125910A1 (en) Method and system for authenticating user of a mobile device via hybrid biometics information
WO2020073495A1 (zh) 基于人工智能的复审方法、装置、设备及存储介质
WO2020253115A1 (zh) 基于语音识别的产品推荐方法、装置、设备和存储介质
WO2020147384A1 (zh) 基于区块链的安全交易方法、装置、设备及存储介质
WO2021182683A1 (ko) 워터마크를 삽입한 음성 인증 시스템 및 이에 대한 방법
CN110458662A (zh) 反欺诈风控方法及装置
TW201504839A (zh) 可攜式電子裝置及互動式人臉登入方法
WO2020006886A1 (zh) 门禁***的识别方法、装置、门禁***及存储介质
WO2023128342A1 (ko) 동형 암호화된 음성을 이용한 개인 식별 방법 및 시스템
Olalere et al. Bring your own device: security challenges and A theoretical framework for two-factor Authentication
WO2021017277A1 (zh) 一种图片截取方法、装置及计算机存储介质
WO2022050459A1 (en) Method, electronic device and system for generating record of telemedicine service
JP6733901B2 (ja) 心理分析装置、心理分析方法、およびプログラム
CN111339829B (zh) 用户身份鉴定方法、装置、计算机设备和存储介质
CN109003190B (zh) 一种核保方法、计算机可读存储介质及终端设备
JP2000306090A (ja) 個人認証装置、方法及び記録媒体
CN115688071B (zh) 一种防止智能手表信息篡改的处理方法及***

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17830232

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 08/05/2019)

122 Ep: pct application non-entry in european phase

Ref document number: 17830232

Country of ref document: EP

Kind code of ref document: A1