CN110782335B - Method, device and storage medium for processing credit data based on artificial intelligence - Google Patents
Method, device and storage medium for processing credit data based on artificial intelligence Download PDFInfo
- Publication number
- CN110782335B CN110782335B CN201910884636.XA CN201910884636A CN110782335B CN 110782335 B CN110782335 B CN 110782335B CN 201910884636 A CN201910884636 A CN 201910884636A CN 110782335 B CN110782335 B CN 110782335B
- Authority
- CN
- China
- Prior art keywords
- media stream
- overdue
- target
- loan
- collection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000012545 processing Methods 0.000 title claims abstract description 55
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 48
- 230000004044 response Effects 0.000 claims abstract description 29
- 238000005516 engineering process Methods 0.000 claims abstract description 17
- 230000002787 reinforcement Effects 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 21
- 238000009432 framing Methods 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 13
- 230000037433 frameshift Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 230000001965 increasing effect Effects 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000013461 design Methods 0.000 description 13
- 238000004590 computer program Methods 0.000 description 7
- 239000000284 extract Substances 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 230000006854 communication Effects 0.000 description 3
- 238000013480 data collection Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000003197 catalytic effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
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
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/1822—Parsing for meaning understanding
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/225—Feedback of the input speech
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Acoustics & Sound (AREA)
- Human Computer Interaction (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Technology Law (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The application relates to the field of artificial intelligence, and provides a method, a device and a storage medium for processing credit data based on artificial intelligence, wherein the method comprises the following steps: respectively determining the problem types matched with the overdue types according to the overdue types corresponding to the target users in the loan collection list, selecting target problems matched with the problem types from the question bank according to the problem types, determining the collection channels of the target users according to the overdue types, establishing session connection with the terminals bound with the target users through the collection channels, converting the target problems into first media streams according to response logic of the target problems, and sending the first media streams to the terminals of the target users to be in session with the terminals. And carrying out semantic analysis on a second media stream of the target user according to a voice recognition technology, extracting audio data related to overdue repayment from the second media stream, converting the audio data related to overdue repayment into text information, and recording the text information and a collection accelerating label into a collection accelerating platform. The scheme can improve the work efficiency of the collecting.
Description
Technical Field
The present application relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, and a storage medium for processing credit data based on artificial intelligence.
Background
At present, domestic consumption finance is being vigorously developed, and meanwhile, the number of cases of overdue loans is continuously increased. Aiming at the overdue loan users, the current collection method is mainly carried out by customer service personnel through telephone, short messages, mails or letters and the like. Specifically, the collection procedure is as follows: when the call is made by using the call-promoting platform, the call-promoting personnel needs to communicate with debtor to confirm various information, and the confirmed information is filled in the operation interface of the call-promoting platform. A lot of labor is required for the collecting staff.
However, in the communication process between the seeker and the debtor, if the voice recognition of the seeker in the telephone communication is inaccurate, the seeker may understand the intention of the wrong debtor, or the factors such as the reduced attention caused by frequent repeated recording may cause the wrong action code to be input, and finally the intention of the wrong debtor is input. Because the cashier needs to fill in a lot of information, such as action codes, cashier notes and the like, when the cashier fills in the confirmed information on the operation interface of the cashier platform, the cashier can easily input wrong information even if he understands the correct voice intention, and the operation is time-consuming and laborious.
Disclosure of Invention
The application provides a method, a device and a storage medium for processing credit data based on artificial intelligence, which can solve the problems of low efficiency and easy error of artificial induction in the prior art.
In a first aspect, the present application provides a method of processing credit data based on artificial intelligence, the method comprising:
acquiring and traversing a plurality of user loan information;
screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan collection list comprising a plurality of target users;
respectively determining the overdue types matched with the overdue types of all target users according to the overdue types corresponding to all target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to all target users, and respectively determining the collection channels of all target users in the loan collection list according to the overdue types;
establishing session connection with the terminals bound with each target user through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and sending the first media stream corresponding to the target problem to the terminal of the target user so as to perform session with the terminal of the target user; the response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal;
Receiving a second media stream sent by a terminal of a target user, carrying out semantic analysis on the second media stream according to a voice recognition technology, and extracting audio data related to overdue repayment from the second media stream;
audio data related to overdue repayment is converted into text information, and the text information and a collection accelerating label matched with the text information are input into the collection accelerating platform;
and generating a loan collection record according to the text information and the collection label, and storing the loan collection record.
In some possible designs, the semantic analysis of the second media stream according to the voice recognition technique extracts data related to overdue payouts from the second media stream, including:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
Semantic understanding (e.g., contextual understanding) the sequence of strings to identify core words in the second media stream;
matching a preset keyword with a core word in the second media stream to obtain the audio data related to overdue repayment; the preset keywords refer to keywords agreeing to repayment or having repayment tendency.
In some possible designs, after receiving the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to the voice recognition technology, the method further includes:
measuring a play decibel of the second media stream;
if the playing decibel of the second media stream is lower than the preset decibel, increasing the playing decibel of the second media stream so that the playing decibel of the second media stream is not lower than the preset decibel, wherein the preset decibel is the lowest decibel value meeting the requirement of the machine for carrying out semantic recognition on the second media stream.
In some possible designs, after receiving the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to the voice recognition technology, the method further includes:
Analyzing data in the second media stream;
if the second media stream contains noise data through analysis, performing voice enhancement processing on the voice signal of the second media stream by adopting the following formula to obtain an optimal voice signal:
subject to Y(k)＝CX(k)+n(k)+v(k)
wherein X (k) and n (k) are variables, X (k) is the optimal estimation of the state value in the Kalman filtering algorithm, n (k) is the estimation of Gaussian noise, Θ is the covariance matrix of the Gaussian noise, and v (k) is sparse noise;
the noise data is filtered from the second media stream using a kalman filter algorithm.
In some possible designs, after receiving the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to the voice recognition technology, the method further includes:
measuring a play speed of the second media stream;
if the playing speed of the second media stream is higher than the preset playing speed, the playing speed of the second media stream is reduced so that the playing speed of the second media stream is not higher than the preset playing speed.
In some possible designs, the reducing the playing speed of the second media stream includes:
reducing the playing speed of the second media stream by framing or windowing
Dividing the voice signal of the second media stream into a plurality of short segments in an overlapping segmentation mode, wherein the voice signal of the second media stream comprises a plurality of video frames, and the overlapping part of the previous frame and the next frame is frame shift;
and weighting the plurality of short segments by using a movable window with a limited length so as to reduce the playing speed of the second media stream.
In some possible designs, after the extracting the audio data related to the overdue payment from the second media stream, before converting the audio data related to the overdue payment to text information, the method further includes:
lie detection analysis is carried out on the second media stream;
if the target user is determined to have false answers, selecting reinforcement questions according to questions corresponding to the false answers, wherein the reinforcement questions are used for guiding the target user to answer questions of multiple dimensions;
sending the reinforcement question to the target user according to a question guiding mode;
receiving an answer of a target user for the reinforcement question;
and carrying out semantic analysis on the answer of the target user aiming at the reinforcement question, and replacing false answers in the data related to overdue repayment with the result of the semantic analysis.
In a second aspect, the present application provides an apparatus for processing credit data based on artificial intelligence, having a function of implementing a method for processing credit data based on artificial intelligence corresponding to the one provided in the first aspect. The functions may be implemented by hardware, or may be implemented by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the functions described above, which may be software and/or hardware.
In one possible design, the apparatus includes:
the receiving and transmitting module is used for acquiring a plurality of user loan information;
the processing module is used for traversing the plurality of user loan information acquired by the receiving and transmitting module; screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan collection list comprising a plurality of target users; respectively determining the overdue types matched with the overdue types of all target users according to the overdue types corresponding to all target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to all target users, and respectively determining the collection channels of all target users in the loan collection list according to the overdue types; establishing session connection with each target user bound terminal through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and transmitting the first media stream corresponding to the target problem to the target user terminal through the receiving-transmitting module so as to perform session with the target user terminal; the response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal;
The processing module is also used for receiving a second media stream sent by the terminal of the target user through the receiving and transmitting module, carrying out semantic analysis on the second media stream according to a voice recognition technology, and extracting audio data related to overdue repayment from the second media stream; audio data related to overdue repayment is converted into text information, and the text information and a collection accelerating label matched with the text information are input into the collection accelerating platform; and generating a loan collection record according to the text information and the collection label, and storing the loan collection record.
In some possible designs, the processing module is specifically configured to:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
semantic understanding (e.g., contextual understanding) the sequence of strings to identify core words in the second media stream;
Matching a preset keyword with a core word in the second media stream to obtain the audio data related to overdue repayment; the preset keywords refer to keywords agreeing to repayment or having repayment tendency.
In some possible designs, the processing module is further configured to, after the transceiver module receives the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to a speech recognition technique:
measuring a play decibel of the second media stream;
if the playing decibel of the second media stream is lower than the preset decibel, increasing the playing decibel of the second media stream so that the playing decibel of the second media stream is not lower than the preset decibel, wherein the preset decibel is the lowest decibel value meeting the requirement of the machine for carrying out semantic recognition on the second media stream.
In some possible designs, the processing module is further configured to, after the transceiver module receives the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to a speech recognition technique:
analyzing data in the second media stream;
if the second media stream contains noise data through analysis, performing voice enhancement processing on the voice signal of the second media stream by adopting the following formula to obtain an optimal voice signal:
subject to Y(k)＝CX(k)+n(k)+v(k)
Wherein X (k) and n (k) are variables, X (k) is the optimal estimation of the state value in the Kalman filtering algorithm, n (k) is the estimation of Gaussian noise, Θ is the covariance matrix of the Gaussian noise, and v (k) is sparse noise;
the noise data is filtered from the second media stream using a kalman filter algorithm.
In some possible designs, the processing module is further configured to, after the transceiver module receives the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to a speech recognition technique:
measuring a play speed of the second media stream;
if the playing speed of the second media stream is higher than the preset playing speed, the playing speed of the second media stream is reduced so that the playing speed of the second media stream is not higher than the preset playing speed.
In some possible designs, the processing module is specifically configured to:
reducing the playing speed of the second media stream by framing or windowing
Dividing the voice signal of the second media stream into a plurality of short segments in an overlapping segmentation mode, wherein the voice signal of the second media stream comprises a plurality of video frames, and the overlapping part of the previous frame and the next frame is frame shift;
And weighting the plurality of short segments by using a movable window with a limited length so as to reduce the playing speed of the second media stream.
In some possible designs, after the processing module extracts the audio data related to the overdue payment from the second media stream, the processing module is further configured to, before converting the audio data related to the overdue payment into text information:
lie detection analysis is carried out on the second media stream;
if the target user is determined to have false answers, selecting reinforcement questions according to questions corresponding to the false answers, wherein the reinforcement questions are used for guiding the target user to answer questions of multiple dimensions;
sending the reinforcement question to the target user according to a question guiding mode;
receiving an answer of a target user for the reinforcement question through the transceiver module;
and carrying out semantic analysis on the answer of the target user aiming at the reinforcement question, and replacing false answers in the data related to overdue repayment with the result of the semantic analysis.
In yet another aspect, the present application provides a computer device comprising at least one connected processor, a memory and a transceiver, wherein the memory is adapted to store program code, and the processor is adapted to invoke the program code in the memory to perform the method according to the first aspect.
A further aspect of the application provides a computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of the first aspect described above.
Compared with the existing mechanism, in the embodiment of the application, the problem types matched with the overdue types of all target users are respectively determined according to the overdue types corresponding to all target users in the loan collection list, the target problems matched with the problem types are selected from the problem base according to the problem types corresponding to all target users, the collection channels of all target users in the loan collection list are respectively determined according to the overdue types, session connection is established between the collection channels and the terminals bound with all target users, the target problems are converted into first media streams according to response logic of the target problems, and the first media streams corresponding to the target problems are sent to the terminals of the target users to conduct session with the terminals of the target users. And carrying out semantic analysis on a second media stream of the target user according to a voice recognition technology, extracting audio data related to overdue repayment from the second media stream, converting the audio data related to overdue repayment into text information, and recording the text information and a collection tag matched with the text information into the collection platform. Therefore, after the scheme is adopted, the machine communicates with the target user and confirms various basic information, so that the collection efficiency is improved, the collection cost is reduced, whether the machine is tired or not is not required to be considered, the accuracy of identifying the real voice intention of the target user can be improved, the loan collection record is automatically generated, and a plurality of information is not required to be filled by a collector, so that collection is intelligent, the productivity is liberated and the user experience is improved.
Drawings
FIG. 1 is a flow chart of a method for processing credit data based on artificial intelligence in an embodiment of the application;
FIG. 2 is a schematic diagram of an apparatus for processing credit data based on artificial intelligence in an embodiment of the application;
fig. 3 is a schematic structural diagram of a computer device according to an embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application. The terms first, second and the like in the description and in the claims and in the above-described figures, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or modules is not necessarily limited to those listed or explicitly listed or inherent to such process, method, article, or apparatus, but may include other steps or modules that may not be listed or inherent to such process, method, article, or apparatus, the partitioning of such modules by the present application may be by one logical partitioning, and may be implemented by other means, such as a plurality of modules may be combined or integrated in another system, or some features may be omitted, or not implemented.
The application provides a method, a device and a storage medium for processing credit data based on artificial intelligence. The scheme is executed by a collecting forcing platform, and artificial intelligence (artificial intelligence, AI) is deployed in the collecting forcing platform.
Referring to FIG. 1, a method of processing credit data based on artificial intelligence in an embodiment of the application is described below, the method comprising:
101. and acquiring and traversing the loan information of a plurality of users.
The user loan information comprises a borrower name, an identity card number, a face image, a mailbox, a mobile phone number, a loan product, a borrowing date, a borrowing amount and a repayment deadline.
102. And screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan prompting list comprising a plurality of target users.
The preset screening conditions may include an overdue unrevealed amount, an overdue day higher than a preset period, an unrevealed amount adjacent to the repayment day, and a remaining repayment day.
The overdue types include risk users, potential risk users, elder Lai users, disjunctive users, and elder Lai users. Wherein, the risk user refers to a user with overdue repayment days exceeding 90 days. A potentially risky user refers to a user whose overdue payoff days do not exceed 60 days. Elder Lai users refer to users who have accumulated up to 4 historical overdue payouts in 6 months. An out-of-contact user refers to a user that is not reachable from a contact phone in an incoming catalytic case. Elder Lai users refer to users who have not paid until the specified payment time. In order to facilitate the distinction of overdue users, an overdue type identifier is set for each overdue user. For example: the potential risk user corresponds to the potential risk type identifier L, the risk user corresponds to the overdue type identifier H, the elder Lai user corresponds to the overdue type identifier F, the disjunct user corresponds to the overdue type identifier M, and the elder Lai user corresponds to the overdue type identifier L.
In the application, only the overdue target user is taken as an example, and the method for processing the overdue target user can be referred to for the scheme of the overdue target user to be overdue target user, and the description is omitted.
103. And respectively determining the overdue types matched with the overdue types of the target users according to the overdue types corresponding to the target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to the target users, and respectively determining the collection channels of the target users in the loan collection list according to the overdue types.
The collect-urging channel comprises a telephone, a network communication account, a mailbox, a home address, a work address and a court.
For example, for a potentially risky user, the target question may be "Mr. Li, your good, your Shenzhen mountain Bank tail 6222342100009094 credit card overdue 22090.23 yuan, please pay as soon as possible before 2019-5-20 days, if there is no objection and agrees to pay to please confirm.
For elder Lai users, the target question may be "you good, when you can pay, if you do not pay, declare your overdue information to the credit centre, and show or submit a court ticket to you nationally.
For an out-of-line user, the target question may be "you good, please pay for payment before 2019-8-20 days, and if not, apply for frozen funds in personal real estate and all financial accounts in place to the court.
104. And establishing session connection with the terminals bound with each target user through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and sending the first media stream corresponding to the target problem to the terminal of the target user so as to perform session with the terminal of the target user.
The response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal.
The first media stream may be voice data or video data, and the voice data or the video data are all AI simulated voice.
In some embodiments, the response logic for converting the target question into the first media stream includes:
and encoding the target problem according to the response logic of the target problem to obtain a first media stream.
105. And receiving a second media stream sent by the terminal of the target user, carrying out semantic analysis on the second media stream according to a voice recognition technology, and extracting audio data related to overdue repayment from the second media stream.
In some embodiments, the semantic analysis of the second media stream according to the voice recognition technique extracts data related to overdue payment from the second media stream, including:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
semantic understanding (e.g., contextual understanding) the sequence of strings to identify core words in the second media stream;
and matching the preset keywords with the core words in the second media stream to obtain the audio data related to overdue repayment.
The preset keywords refer to keywords agreeing to repayment or having repayment tendency. For example, the preset keywords include "good", "OK", and "about still", etc. keywords that carry agreeing to repayment. If the audio data contains keywords of good, OK, return, etc. which carry one-way agreement to repayment, the overdue user agrees to repayment, and the overdue collection result is willing to repayment. If the voice data does not contain the keywords which are good, OK, return, etc. and carry the consent for repayment, the overdue user does not agree for repayment, and the overdue collection result is unwilling to repayment.
In some embodiments, the preset keywords may be implemented using a pre-trained hidden markov (Hidden Markov Model, HMM) model, and the keywords may be identified using the HMM model. The HMM model is a model which is pre-trained and stored in the collection platform and is used for identifying whether preset keywords such as good, OK, OK, return and the like exist in conversation voice. In this embodiment, whether the preset keywords exist in the audio data is identified by using the pre-trained HMM model, so that the identification accuracy can be improved.
In some embodiments, to further ensure accuracy of semantic analysis, after receiving the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to the speech recognition technology, play decibels of the second media stream may be further improved, play speed of the second media stream may be reduced, and noise in the second media stream may be removed. The following are introduced separately:
(1) Enhancing play decibel of second media stream
Measuring a play decibel of the second media stream;
if the playing decibel of the second media stream is lower than the preset decibel, the playing decibel of the second media stream is increased, so that the playing decibel of the second media stream is not lower than the preset decibel.
The preset decibels are the lowest decibel values meeting the requirement of the machine for carrying out semantic recognition on the second media stream.
Even if the volume of the target user speaking is low or the target user speaking is far away from the microphone, the received second media stream cannot be clearly and accurately analyzed due to the fact that the recorded user speaking (namely, the recorded user speaking is converted into the second media stream and sent to the prompting platform) is low. Therefore, after the playing decibel of the second media stream is improved, the accuracy and definition of semantic recognition can be enhanced.
(2) Reducing the playing speed of the second media stream
Measuring a play speed of the second media stream;
if the playing speed of the second media stream is higher than the preset playing speed, the playing speed of the second media stream is reduced so that the playing speed of the second media stream is not higher than the preset playing speed.
Even if the speaking speed of the target user is fast, the received second media stream cannot be clearly and accurately analyzed because of the fast speaking speed of the recorded user speaking (namely, the second media stream is converted into the second media stream and sent to the collection platform). Therefore, after the playing speed of the second media stream is reduced, the accuracy and definition of semantic recognition can be enhanced.
In some embodiments, the playing speed of the second media stream may be reduced in a framing or windowing manner. Taking the second media stream as the voice signal for example, the voice signal has short-time stationarity (the voice signal can be considered to be approximately unchanged within 10-30 ms), then, the method of framing the voice signal refers to a method of dividing the voice signal into a plurality of short segments, for example, overlapping segments, where the overlapping part of the previous frame and the subsequent frame is called frame shift, and the ratio of the frame shift to the frame length is generally 0-0.5. For example, the number of frames per second is about 33 to 100 frames, the frame length is 25ms, and the frame shift is 10ms. Framing may be achieved by weighting with a movable window of finite length.
(3) Removing noise in the second media stream
Analyzing data in the second media stream;
and if the second media stream contains noise data through analysis, filtering the noise data from the second media stream.
Even if noise exists around the target user speaking, the received second media stream cannot be clearly and accurately analyzed due to the fact that the recorded user speaking (namely, the recorded user speaking is converted into the second media stream and sent to the collection platform) comprises noise data. It can be seen that, after removing the noise data in the second media stream, semantic enhancement can be achieved, that is, accuracy and definition of semantic recognition are enhanced.
The noise may be filtered using a kalman (Kronecker) filter algorithm. The present application is not limited to the manner in which noise data is analyzed, or the manner in which noise data is filtered.
For example, the statistical characteristics of the speech signal, noise data p (k) and noise data n (k) are respectively:
E(p(k))＝q,E(n(k))＝r
E(p(k)p(j)T)＝Qδkj,E(n(k)n(j)T)＝Rδkj
wherein q and r are the average of the noise p (k) and n (k), respectively; q and R are the covariance of noise data p (k) and noise data n (k), respectively, and δkj is the Kronecker function. In the present application, speech enhancement refers to estimating an optimal speech signal X (k), which may also be referred to as optimal estimation, on the premise that the speech signal Y (k) is known. The optimization formula of the voice signal is as follows:
subject to Y(k)＝CX(k)+n(k)+v(k)
wherein X (k) and n (k) are variables, X (k) is the optimal estimation of the state value in the Kronecker filtering algorithm, n (k) is the estimation of Gaussian noise, Θ is the covariance matrix of Gaussian noise, v (k) is sparse noise, and the optimal estimation X (k) of the voice signal is obtained by solving the optimization formula.
106. And converting the audio data related to overdue repayment into text information, and recording the text information and a collection accelerating label matched with the text information into the collection accelerating platform.
The collect tag may include, among other things, a promised pay-off schedule (e.g., connect-principal-promised pay-off), a request to delay pay-off (e.g., connect-principal-promised pay-off), a weak pay-off (e.g., connect-user family-weak pay-off), etc.
In some embodiments, after extracting the audio data related to the overdue payment from the second media stream, before converting the audio data related to the overdue payment to text information, the method further includes:
lie detection analysis is carried out on the second media stream;
if the target user is determined to have false answers, selecting reinforcement questions according to questions corresponding to the false answers, wherein the reinforcement questions are used for guiding the target user to answer questions of multiple dimensions;
sending the reinforcement question to the target user according to a question guiding mode;
receiving an answer of a target user for the reinforcement question;
and carrying out semantic analysis on the answer of the target user aiming at the reinforcement question, and replacing false answers in the data related to overdue repayment with the result of the semantic analysis.
Therefore, through lie detection analysis on the second media stream of the target user, whether the target user is in the process of concealing the real repayment capability and repayment willingness can be deeply mined, the grasping degree of the real repayment capability of the target user and the repayment willingness can be improved, and accordingly the overdue repayment probability and the risk of coping with potential reimbursement are reduced.
107. And generating a loan collection record according to the text information and the collection label, and storing the loan collection record.
The loan collection record refers to the whole record of each collection and repayment, and is used for analysis, updating and management of collection personnel, for example, the collection record is conveniently searched according to conditions, or collection grades are allocated for target users. The overdue loan data collection level is determined according to the time exceeding the repayment date or the overdue loan amount, and the longer the overdue time is, the higher the overdue loan data collection level is, the larger the loan amount is, and the higher the overdue loan data collection level is.
In some embodiments, after the loan collection record is generated, user images (for example, in the form of a knowledge graph) can be drawn for each target user according to the collection record, so that the collection personnel can more intuitively analyze the characteristics of the target user, such as repayment capability, repayment credit, and the like.
Compared with the existing mechanism, in the embodiment of the application, the problem types matched with the overdue types of all target users are respectively determined according to the overdue types corresponding to all target users in the loan collection list, the target problems matched with the problem types are selected from the problem base according to the problem types corresponding to all target users, the collection channels of all target users in the loan collection list are respectively determined according to the overdue types, session connection is established between the collection channels and the terminals bound with all target users, the target problems are converted into first media streams according to response logic of the target problems, and the first media streams corresponding to the target problems are sent to the terminals of the target users to conduct session with the terminals of the target users. And carrying out semantic analysis on a second media stream of the target user according to a voice recognition technology, extracting audio data related to overdue repayment from the second media stream, converting the audio data related to overdue repayment into text information, and recording the text information and a collection tag matched with the text information into the collection platform. Therefore, after the scheme is adopted, the machine communicates with the target user and confirms various basic information, so that the collection efficiency is improved, the collection cost is reduced, whether the machine is tired or not is not required to be considered, the accuracy of identifying the real voice intention of the target user can be improved, the loan collection record is automatically generated, and a plurality of information is not required to be filled by a collector, so that collection is intelligent, the productivity is liberated and the user experience is improved.
The technical features mentioned in the embodiment or implementation corresponding to fig. 1 are also applicable to the embodiments corresponding to fig. 2 and 3 in the present application, and the details of the similar parts will not be described in detail.
The above describes a method for processing credit data based on artificial intelligence in the present application, and the following describes an apparatus for performing the above method for processing credit data based on artificial intelligence.
A schematic diagram of an apparatus 20 for processing credit data based on artificial intelligence, as shown in fig. 2, is applicable to speech intent recognition, such as a robot voice or video call. The apparatus 20 of the present embodiment is capable of implementing steps corresponding to the method of processing credit data based on artificial intelligence performed in the embodiment corresponding to fig. 1 described above. The functions implemented by the apparatus 20 may be implemented by hardware, or may be implemented by executing corresponding software by hardware. The hardware or software includes one or more modules corresponding to the functions described above, which may be software and/or hardware. The apparatus 20 may include a transceiver module 201 and a processing module 202, and the functional implementation of the processing module 202 and the obtaining module 201 may refer to operations performed in the embodiment corresponding to fig. 1, which are not described herein. The processing module 202 may be configured to control operations of transceiving, acquiring, etc. of the transceiving module 201.
In some embodiments, the transceiver module 201 may be configured to obtain a plurality of user loan information;
the processing module 202 may be configured to traverse the plurality of user loan information obtained by the transceiver module 201; screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan collection list comprising a plurality of target users; respectively determining the overdue types matched with the overdue types of all target users according to the overdue types corresponding to all target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to all target users, and respectively determining the collection channels of all target users in the loan collection list according to the overdue types; establishing session connection with the terminals bound with each target user through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and sending the first media stream corresponding to the target problem to the terminal of the target user through the receiving-transmitting module 201 so as to perform session with the terminal of the target user; the response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal;
The processing module 202 is further configured to receive a second media stream sent by the terminal of the target user through the transceiver module, perform semantic analysis on the second media stream according to a voice recognition technology, and extract audio data related to overdue payment from the second media stream; audio data related to overdue repayment is converted into text information, and the text information and a collection accelerating label matched with the text information are input into the collection accelerating platform; and generating a loan collection record according to the text information and the collection label, and storing the loan collection record.
In some embodiments, the processing module 202 is specifically configured to:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
semantic understanding (e.g., contextual understanding) the sequence of strings to identify core words in the second media stream;
Matching a preset keyword with a core word in the second media stream to obtain the audio data related to overdue repayment; the preset keywords refer to keywords agreeing to repayment or having repayment tendency.
In some embodiments, after the transceiver module 201 receives the second media stream sent by the terminal of the target user, the processing module 202 is further configured to, before performing semantic analysis on the second media stream according to the speech recognition technology:
measuring a play decibel of the second media stream;
if the playing decibel of the second media stream is lower than the preset decibel, increasing the playing decibel of the second media stream so that the playing decibel of the second media stream is not lower than the preset decibel, wherein the preset decibel is the lowest decibel value meeting the requirement of the machine for carrying out semantic recognition on the second media stream.
In some embodiments, the processing module 202 is further configured to, after the transceiver module 201 receives the second media stream sent by the terminal of the target user, before performing semantic analysis on the second media stream according to the speech recognition technology:
analyzing data in the second media stream;
if the second media stream contains noise data through analysis, performing voice enhancement processing on the voice signal of the second media stream by adopting the following formula to obtain an optimal voice signal:
subject to Y(k)＝CX(k)+n(k)+v(k)
Wherein X (k) and n (k) are variables, X (k) is the optimal estimation of the state value in the Kalman filtering algorithm, n (k) is the estimation of Gaussian noise, Θ is the covariance matrix of the Gaussian noise, and v (k) is sparse noise;
the noise data is filtered from the second media stream using a kalman filter algorithm.
In some embodiments, after the transceiver module receives the second media stream sent by the terminal of the target user, the processing module 202 is further configured to, before performing semantic analysis on the second media stream according to a speech recognition technique:
measuring a play speed of the second media stream;
if the playing speed of the second media stream is higher than the preset playing speed, the playing speed of the second media stream is reduced so that the playing speed of the second media stream is not higher than the preset playing speed.
In some embodiments, the processing module 202 is specifically configured to:
reducing the playing speed of the second media stream by framing or windowing
Dividing the voice signal of the second media stream into a plurality of short segments in an overlapping segmentation mode, wherein the voice signal of the second media stream comprises a plurality of video frames, and the overlapping part of the previous frame and the next frame is frame shift;
And weighting the plurality of short segments by using a movable window with a limited length so as to reduce the playing speed of the second media stream.
In some embodiments, after the processing module 202 extracts the audio data related to the overdue payment from the second media stream, the processing module is further configured to:
lie detection analysis is carried out on the second media stream;
if the target user is determined to have false answers, selecting reinforcement questions according to questions corresponding to the false answers, wherein the reinforcement questions are used for guiding the target user to answer questions of multiple dimensions;
sending the reinforcement question to the target user according to a question guiding mode;
receiving an answer of a target user for the reinforcement question through the transceiver module;
and carrying out semantic analysis on the answer of the target user aiming at the reinforcement question, and replacing false answers in the data related to overdue repayment with the result of the semantic analysis.
Note that, the entity device corresponding to the transceiver module 201 shown in fig. 2 is a transceiver shown in fig. 3, and the transceiver can implement a part or all of the functions of the transceiver module 201, or implement the same or similar functions as the transceiver module 201.
The physical device corresponding to the processing module 202 shown in fig. 2 is a processor shown in fig. 3, which can implement part or all of the functions of the processing module 202, or implement the same or similar functions as the processing module 202.
The apparatus 20 of the embodiment of the present application is described above in terms of modular functional entities, and a computer device is described below in terms of hardware, as shown in fig. 3, which includes: a processor, a memory, a transceiver (which may also be an input-output unit, not identified in fig. 3) and a computer program stored in the memory and executable on the processor. For example, the computer program may be a program corresponding to the method of processing credit data based on artificial intelligence in the embodiment corresponding to fig. 1. For example, when the computer device implements the functionality of apparatus 20 as shown in FIG. 2, the processor, when executing the computer program, implements the steps of the method for processing credit data based on artificial intelligence performed by apparatus 20 in the embodiment corresponding to FIG. 2 described above; alternatively, the processor may implement the functions of the modules in the apparatus 20 according to the embodiment corresponding to fig. 2 when executing the computer program. As another example, the computer program may be a program corresponding to the method of processing credit data based on artificial intelligence in the embodiment corresponding to fig. 1.
The processor may be a central processing unit (Central Processing Unit, CPU), other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like that is a control center of the computer device, connecting various parts of the overall computer device using various interfaces and lines.
The memory may be used to store the computer program and/or modules, and the processor may implement various functions of the computer device by running or executing the computer program and/or modules stored in the memory, and invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like; the storage data area may store data (such as audio data, video data, etc.) created according to the use of the cellular phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart Media Card (SMC), secure Digital (SD) Card, flash Card (Flash Card), at least one disk storage device, flash memory device, or other volatile solid-state storage device.
The transceiver may also be replaced by a receiver and a transmitter, which may be the same or different physical entities. Which are the same physical entities, may be collectively referred to as transceivers. The transceiver may be an input-output unit.
The memory may be integrated in the processor or may be provided separately from the processor.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM), comprising instructions for causing a terminal (which may be a mobile phone, a computer, a server or a network device, etc.) to perform the method according to the embodiments of the present application.
While the embodiments of the present application have been described above with reference to the drawings, the present application is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many modifications may be made thereto by those of ordinary skill in the art without departing from the spirit of the present application and the scope of the appended claims, which are to be accorded the full scope of the present application as defined by the following description and drawings, or by any equivalent structures or equivalent flow changes, or by direct or indirect application to other relevant technical fields.
Claims (9)
1. A method of processing credit data based on artificial intelligence, the method comprising:
acquiring and traversing a plurality of user loan information;
screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan collection list comprising a plurality of target users;
respectively determining the overdue types matched with the overdue types of all target users according to the overdue types corresponding to all target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to all target users, and respectively determining the collection channels of all target users in the loan collection list according to the overdue types;
establishing session connection with the terminals bound with each target user through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and sending the first media stream corresponding to the target problem to the terminal of the target user so as to perform session with the terminal of the target user; the response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal;
Receiving a second media stream sent by a terminal of a target user, carrying out semantic analysis on the second media stream according to a voice recognition technology, and extracting audio data related to overdue repayment from the second media stream;
audio data related to overdue repayment is converted into text information, and the text information and a collection accelerating label matched with the text information are input into the collection accelerating platform;
generating a loan collection record according to the text information and the collection label and storing the loan collection record;
the semantic analysis is performed on the second media stream according to a voice recognition technology, and the extracting of the data related to overdue repayment from the second media stream comprises the following steps:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
semantic understanding is carried out on the character string sequence so as to identify core words in the second media stream;
Matching a preset keyword with a core word in the second media stream to obtain the audio data related to overdue repayment; the preset keywords refer to keywords agreeing to repayment or having repayment tendency.
2. The method according to claim 1, wherein after receiving the second media stream sent by the terminal of the target user, before the semantic analysis of the second media stream according to the speech recognition technique, the method further comprises:
measuring a play decibel of the second media stream;
if the playing decibel of the second media stream is lower than the preset decibel, increasing the playing decibel of the second media stream so that the playing decibel of the second media stream is not lower than the preset decibel, wherein the preset decibel is the lowest decibel value meeting the requirement of the machine for carrying out semantic recognition on the second media stream.
3. The method according to claim 1 or claim 2, wherein after receiving the second media stream sent by the terminal of the target user, before the semantic analysis of the second media stream according to the speech recognition technique, the method further comprises:
analyzing data in the second media stream;
If the second media stream contains noise data through analysis, performing voice enhancement processing on the voice signal of the second media stream by adopting the following formula to obtain an optimal voice signal:
subject to Y(k)＝CX(k)+n(k)+v(k)
wherein X (k) and n (k) are variables, X (k) is the optimal estimation of the state value in the Kalman filtering algorithm, n (k) is the estimation of Gaussian noise, Θ is the covariance matrix of the Gaussian noise, and v (k) is sparse noise;
the noise data is filtered from the second media stream using a kalman filter algorithm.
4. The method according to claim 1 or claim 2, wherein after receiving the second media stream sent by the terminal of the target user, before the semantic analysis of the second media stream according to the speech recognition technique, the method further comprises:
measuring a play speed of the second media stream;
if the playing speed of the second media stream is higher than the preset playing speed, the playing speed of the second media stream is reduced so that the playing speed of the second media stream is not higher than the preset playing speed.
5. The method of claim 4, wherein said reducing the playback speed of the second media stream comprises:
Reducing the playing speed of the second media stream by framing or windowing
Dividing the voice signal of the second media stream into a plurality of short segments in an overlapping segmentation mode, wherein the voice signal of the second media stream comprises a plurality of video frames, and the overlapping part of the previous frame and the next frame is frame shift;
and weighting the plurality of short segments by using a movable window with a limited length so as to reduce the playing speed of the second media stream.
6. The method of claim 1 or claim 2, wherein after extracting the audio data related to the overdue payment from the second media stream, the method further comprises, prior to converting the audio data related to the overdue payment to text information:
lie detection analysis is carried out on the second media stream;
if the target user is determined to have false answers, selecting reinforcement questions according to questions corresponding to the false answers, wherein the reinforcement questions are used for guiding the target user to answer questions of multiple dimensions;
sending the reinforcement question to the target user according to a question guiding mode;
receiving an answer of a target user for the reinforcement question;
and carrying out semantic analysis on the answer of the target user aiming at the reinforcement question, and replacing false answers in the data related to overdue repayment with the result of the semantic analysis.
7. An apparatus for processing credit data based on artificial intelligence, the apparatus comprising:
the receiving and transmitting module is used for acquiring a plurality of user loan information;
the processing module is used for traversing the plurality of user loan information acquired by the receiving and transmitting module; screening out a plurality of overdue loan data from the user loan information according to preset screening conditions, classifying each overdue loan data according to overdue types, and counting target users corresponding to each overdue loan data to generate a loan collection list comprising a plurality of target users; respectively determining the overdue types matched with the overdue types of all target users according to the overdue types corresponding to all target users in the loan collection list, selecting target questions matched with the question types from a question bank according to the question types corresponding to all target users, and respectively determining the collection channels of all target users in the loan collection list according to the overdue types; establishing session connection with each target user bound terminal through the collecting channel, converting the target problem into a first media stream according to response logic of the target problem, and transmitting the first media stream corresponding to the target problem to the target user terminal through the receiving-transmitting module so as to perform session with the target user terminal; the response logic refers to a response rule that the receiving platform sends a media stream of a target problem to a terminal of a target user according to a preset response mode so as to session with the terminal;
The processing module is also used for receiving a second media stream sent by the terminal of the target user through the receiving and transmitting module, carrying out semantic analysis on the second media stream according to a voice recognition technology, and extracting audio data related to overdue repayment from the second media stream; audio data related to overdue repayment is converted into text information, and the text information and a collection accelerating label matched with the text information are input into the collection accelerating platform; generating a loan collection record according to the text information and the collection label and storing the loan collection record;
the processing module is further configured to:
carrying out sound framing on the second media stream by adopting a moving window function to obtain a plurality of sections of audio data, wherein each section of audio data is a frame;
respectively carrying out waveform transformation on each section of audio data to obtain a multidimensional vector corresponding to each section of audio data, wherein the multidimensional vector is an acoustic feature;
identifying a pronunciation sequence with the highest probability of matching with each multidimensional vector by adopting an acoustic model;
searching a character string sequence with the maximum matching probability with the pronunciation sequence by adopting a language model;
semantic understanding is carried out on the character string sequence so as to identify core words in the second media stream;
Matching a preset keyword with a core word in the second media stream to obtain the audio data related to overdue repayment; the preset keywords refer to keywords agreeing to repayment or having repayment tendency.
8. A computer device, the device comprising:
at least one processor, memory, and transceiver;
wherein the memory is for storing program code and the processor is for invoking the program code stored in the memory to perform the method of any of claims 1-6.
9. A computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any of claims 1-6.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910884636.XA CN110782335B (en) | 2019-09-19 | 2019-09-19 | Method, device and storage medium for processing credit data based on artificial intelligence |
| PCT/CN2019/118239 WO2021051592A1 (en) | 2019-09-19 | 2019-11-14 | Method, device, and storage medium for processing data on basis of artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910884636.XA CN110782335B (en) | 2019-09-19 | 2019-09-19 | Method, device and storage medium for processing credit data based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110782335A CN110782335A (en) | 2020-02-11 |
| CN110782335B true CN110782335B (en) | 2023-08-15 |
Family
ID=69384230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910884636.XA Active CN110782335B (en) | 2019-09-19 | 2019-09-19 | Method, device and storage medium for processing credit data based on artificial intelligence |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110782335B (en) |
| WO (1) | WO2021051592A1 (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111968632B (en) * | 2020-07-14 | 2024-05-10 | 招联消费金融股份有限公司 | Call voice acquisition method, device, computer equipment and storage medium |
| CN111882431A (en) * | 2020-08-04 | 2020-11-03 | 武汉众邦银行股份有限公司 | Intelligent message pushing method based on NLP deep learning |
| CN112258317B (en) * | 2020-10-30 | 2022-11-11 | 深圳壹账通智能科技有限公司 | Online credit method, device, computer equipment and medium based on artificial intelligence |
| CN113159901B (en) * | 2021-04-29 | 2024-06-04 | 天津狮拓信息技术有限公司 | Method and device for realizing financing lease business session |
| CN113724066A (en) * | 2021-08-31 | 2021-11-30 | 平安普惠企业管理有限公司 | Rights and interests allocation method, device and equipment based on semantic analysis and storage medium |
| CN114630186B (en) * | 2022-03-21 | 2023-11-07 | 河北太行农牧供应链有限公司 | Intelligent guiding method, system and computer storage medium for live webcasting |
| CN114610840A (en) * | 2022-03-23 | 2022-06-10 | 平安普惠企业管理有限公司 | Sensitive word-based accounting monitoring method, device, equipment and storage medium |
| CN114820124A (en) * | 2022-04-26 | 2022-07-29 | 平安国际融资租赁有限公司 | Method, device, equipment and medium for reducing vehicle rental overdue rate |
| CN116153330B (en) * | 2023-04-04 | 2023-06-23 | 杭州度言软件有限公司 | Intelligent telephone voice robot control method |
| CN116629456B (en) * | 2023-07-20 | 2023-10-13 | 杭银消费金融股份有限公司 | Method, system and storage medium for predicting overdue risk of service |
| CN117094817B (en) * | 2023-10-20 | 2024-02-13 | 国任财产保险股份有限公司 | Credit risk control intelligent prediction method and system |
| CN117579723A (en) * | 2023-11-22 | 2024-02-20 | 东亚银行(中国)有限公司 | Message parsing method and system |
| CN118110990A (en) * | 2024-04-19 | 2024-05-31 | 新疆凯龙清洁能源股份有限公司 | Method and system for recovering heat of flue gas of natural gas power generation device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194807A (en) * | 2017-06-29 | 2017-09-22 | 喀什博雅成信网络科技有限公司 | The intelligent collection system and method for one kind loan |
| CN109064315A (en) * | 2018-08-02 | 2018-12-21 | 平安科技（深圳）有限公司 | Overdue bill intelligence collection method, apparatus, computer equipment and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7191150B1 (en) * | 2000-02-01 | 2007-03-13 | Fair Isaac Corporation | Enhancing delinquent debt collection using statistical models of debt historical information and account events |
| CN108961095A (en) * | 2018-06-13 | 2018-12-07 | 国网福建省电力有限公司 | A kind of intelligent collection electricity charge system based on AI |
| CN109147796B (en) * | 2018-09-06 | 2024-02-09 | 平安科技（深圳）有限公司 | Speech recognition method, device, computer equipment and computer readable storage medium |
| CN109670148A (en) * | 2018-09-26 | 2019-04-23 | 平安科技（深圳）有限公司 | Collection householder method, device, equipment and storage medium based on speech recognition |
| CN109671433B (en) * | 2019-01-10 | 2023-06-16 | 腾讯科技（深圳）有限公司 | Keyword detection method and related device |
-
2019
- 2019-09-19 CN CN201910884636.XA patent/CN110782335B/en active Active
- 2019-11-14 WO PCT/CN2019/118239 patent/WO2021051592A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194807A (en) * | 2017-06-29 | 2017-09-22 | 喀什博雅成信网络科技有限公司 | The intelligent collection system and method for one kind loan |
| CN109064315A (en) * | 2018-08-02 | 2018-12-21 | 平安科技（深圳）有限公司 | Overdue bill intelligence collection method, apparatus, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021051592A1 (en) | 2021-03-25 |
| CN110782335A (en) | 2020-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110782335B (en) | Method, device and storage medium for processing credit data based on artificial intelligence | |
| CN110600017A (en) | Training method of voice processing model, voice recognition method, system and device | |
| CN110704618B (en) | Method and device for determining standard problem corresponding to dialogue data | |
| CN111696558A (en) | Intelligent outbound method, device, computer equipment and storage medium | |
| CN106709804A (en) | Interactive wealth planning consulting robot system | |
| CN112235470B (en) | Incoming call client follow-up method, device and equipment based on voice recognition | |
| CN109145050A (en) | A kind of calculating equipment | |
| CN111739537B (en) | Semantic recognition method and device, storage medium and processor | |
| CN113887214A (en) | Artificial intelligence based wish presumption method and related equipment thereof | |
| CN113726942A (en) | Intelligent telephone answering method, system, medium and electronic terminal | |
| CN110610697B (en) | Voice recognition method and device | |
| CN109120509B (en) | Information collection method and device | |
| CN113724698B (en) | Training method, device, equipment and storage medium of voice recognition model | |
| CN110675865A (en) | Method and apparatus for training hybrid language recognition models | |
| CN110379419A (en) | Phonetic feature matching process based on convolutional neural networks | |
| CN115831125A (en) | Speech recognition method, device, equipment, storage medium and product | |
| CN113282686B (en) | Association rule determining method and device for unbalanced sample | |
| CN112820298B (en) | Voiceprint recognition method and device | |
| KR20200140235A (en) | Method and device for building a target speaker's speech model | |
| CN113314103A (en) | Illegal information identification method and device based on real-time speech emotion analysis | |
| CN113220933A (en) | Method and device for classifying audio segments and electronic equipment | |
| CN110853623A (en) | Automatic financial collection urging method and device based on voice question answering | |
| CN112820316A (en) | Intelligent customer service dialogue method and system | |
| CN112116926A (en) | Audio data processing method and device and model training method and device | |
| CN113239164B (en) | Multi-round dialogue flow construction method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |