CN109509086A - The method, apparatus and storage medium of processing collection business based on artificial intelligence - Google Patents
The method, apparatus and storage medium of processing collection business based on artificial intelligence Download PDFInfo
- Publication number
- CN109509086A CN109509086A CN201811434822.5A CN201811434822A CN109509086A CN 109509086 A CN109509086 A CN 109509086A CN 201811434822 A CN201811434822 A CN 201811434822A CN 109509086 A CN109509086 A CN 109509086A
- Authority
- CN
- China
- Prior art keywords
- collection
- model data
- overdue
- data set
- scorecard
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Abstract
The embodiment of present disclosure discloses a kind of method and apparatus of processing collection business based on artificial intelligence.This method comprises: obtaining collection scorecard set；Obtain the current overdue state and parameter to be refunded of client to be assessed；One and the matched collection scorecard of client to be assessed are selected from the collection scorecard set according to the current overdue state and the parameter to be refunded；It is scored using the collection that selected collection scorecard obtains the client to be assessed；And the collection mode for being used for the client to be assessed is determined based on collection scoring.The method of the processing collection business based on artificial intelligence of embodiment according to present disclosure is by the collection scorecard suitable for different client's selections, to obtain the recommendation of corresponding collection mode, auxiliary establishes more accurate collection strategy.
Description
Technical field
Present disclosure belongs to information technology field more particularly to a kind of side of the processing collection business based on artificial intelligence
Method, device and a kind of corresponding computer readable storage medium.
Background technique
Traditional collection industry is labor-intensive industry, is limited to many factors such as time, place, weather, place
Time cost and the human cost for managing the consumption of collection business are larger, and mechanism (for example, internet financing corporation) is for control cost
Purpose, mostly based on dialogue (for example, voice, mail, short message etc.) collection.
Artificial intelligence (Artificial Intelligence), english abbreviation AI.It is research, develop for simulating,
Extend and the theory of the intelligence of extension people, method, a new technological sciences of technology and application system.With big data and people
The development of work intellectual technology, collection industry start to skill type and intelligent development.For example, some mechanisms can be by using machine
Device voice or intelligent customer service replace operator attendance.
Regardless of using which kind of collection means, how risk rating is carried out to different clients, to determine which kind of, which is matched, urges
Debit's formula, and then improving collection efficiency is all primary technical problem.
In terms of collection scorecard, collection scorecard mainly studies the history refund behavior of client in industry, for history
Then there is no form profound excavation for collection record.In addition, the A card that do not link (scorecard when application) carries out multidimensional to client
Degree modeling.Furthermore most of scorecard is all static scorecard in industry, fails iteration day by day, does not just become dynamic yet and comment
Divide card, so that more accurate carry out effective post-loan management to client.
In addition, the post-loan management in industry is concentrated on for different overdue state clients at present in terms of post-loan management
Management, early warning and risk identification after seldom being borrowed to the client that do not break a contract, for example, present collection is primarily directed to having exceeded
The fund being in arrears with is recovered by short message forms such as visiting abroad of making a phone call, is a kind of risk control of hysteresis quality by the client of phase；
Also distinctive management is not carried out to the client of same overdue state, for example, the client of same overdue state now, used in the person of urging
If art and collection method be it is identical, without a kind of directiveness method can by the client of these same overdue states into
Client's method therefor needs of row one differentiation, different attribute are had any different；Risk control is still with artificial collection after present loan
Based on, fail deeply to apply artificial intelligence technology into the various aspects in collection management and Risk-warning.
Summary of the invention
The embodiment of present disclosure provides the method, apparatus and phase of a kind of processing collection business based on artificial intelligence
The computer readable storage medium answered makes it possible to match different collection scorecards for the client of different conditions, by selected
After the collection scorecard selected has determined scoring, collection strategy can be accurately matched according to identified scoring, so as to
Better collection effect is obtained by more efficiently collection mode.
For this purpose, the first aspect of the embodiment of present disclosure proposes a kind of processing collection business based on artificial intelligence
Method, which comprises
Obtain collection scorecard set；
Obtain the current overdue state and parameter to be refunded of client to be assessed；
According to the current overdue state and the parameter to be refunded select from the collection scorecard set one and
The matched collection scorecard of client to be assessed；
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And
The collection mode for being used for the client to be assessed is determined based on collection scoring.
The second aspect of the embodiment of present disclosure proposes a kind of dress of processing collection business based on artificial intelligence
It sets, described device includes:
Processor；And
Memory makes the processor execute following operation when executed for storing instruction:
Obtain collection scorecard set；
Obtain the current overdue state and parameter to be refunded of client to be assessed；
According to the current overdue state and the parameter to be refunded select from the collection scorecard set one and
The matched collection scorecard of client to be assessed；
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And
The collection mode for being used for the client to be assessed is determined based on collection scoring.
The third aspect of the embodiment of present disclosure proposes a kind of computer readable storage medium, including computer can
It executes instruction, the computer executable instructions make described device execute the reality according to present disclosure when running in a device
The method for applying the processing collection business described in the first aspect of example based on artificial intelligence.
The method, apparatus of the processing collection business based on artificial intelligence of embodiment according to present disclosure and opposite
The computer readable storage medium answered is by the collection scorecard suitable for different client's selections, to obtain corresponding collection
The recommendation of mode, auxiliary establish more accurate collection strategy.Other advantages of present disclosure will hereinafter furtherly
It is bright.
Detailed description of the invention
It refers to the following detailed description in conjunction with the accompanying drawings, the feature, advantage and other aspects of each embodiment of present disclosure
It will be apparent, several embodiments of present disclosure are shown by way of example rather than limitation herein, attached
In figure:
The embodiment that Fig. 1 shows present disclosure can be applied to the processing collection business therein based on artificial intelligence
Exemplary environments 100 signal architecture diagram；
Fig. 2 shows the methods 200 according to the processing collection business based on artificial intelligence of the embodiment of present disclosure
Flow chart.
Fig. 3 shows the device 300 of the processing collection business based on artificial intelligence of the embodiment according to present disclosure
Schematic diagram.
Specific embodiment
Below with reference to each exemplary embodiment of attached drawing detailed description present disclosure.Flow chart and block diagram in attached drawing
Show the architecture, function and operation in the cards of the method and system of the various embodiments according to present disclosure.
It should be noted that each of flowchart or block diagram box can represent a part of a module, program segment or code, institute
The a part for stating module, program segment or code may include one or more patrolling for realizing defined in each embodiment
Collect the executable instruction of function.It should also be noted that in some alternative implementations, function marked in the box can also be with
Occur according to the sequence for being different from being marked in attached drawing.For example, two boxes succeedingly indicated can actually be substantially parallel
Ground executes or they can also be executed in a reverse order sometimes, this depends on related function.It is also noted that
, the combination of the box in each of flowchart and or block diagram box and flowchart and or block diagram can be used
The dedicated hardware based system of defined functions or operations is executed to realize, or specialized hardware and computer can be used
The combination of instruction is realized.
Term as used herein "include", "comprise" and similar terms are open terms, i.e., " including/include but
It is not limited to ", expression can also include other content.Term "based" is " being based at least partially on ".Term " one embodiment "
It indicates " at least one embodiment "；Term " another embodiment " expression " at least one other embodiment " etc..
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as part of specification.For the company between each unit in attached drawing
Line, it is only for convenient for explanation, indicate that the unit at least line both ends is in communication with each other, it is not intended that the non-line of limitation
Unit between can not communicate.
For ease of description, some terms occurred in present disclosure are illustrated below, it should be understood that the disclosure
Term used in content should be interpreted that have and its context of this specification and in relation to the meaning in field it is consistent
Meaning.
Term " client " in present disclosure refer to for meet production, personal consumption and need to buy and mention using mechanism
The user group for the service that the product or acceptance agencies of confession provide.
Term " employee " in present disclosure refers in mechanism for providing services to the user (for example, carrying out with user
Dialog interaction etc.) group.
Term " user's portrait " in present disclosure refers to the tag set for portraying user characteristics, such as user's
Age, gender, region, place city, constellation, educational background, family structure, marital status, hobby, personality, income, consumption energy
Power repays habit, repaying ability, repays wish, converse habit, affiliated purchasing group etc..
Term " user basic information " in present disclosure can include but is not limited to age of user, gender, region,
Place city, constellation, educational background, family structure, marital status, hobby, income etc..
Term " user behavior information " in present disclosure can include but is not limited to consumption information, loan information, repay
Also information (for example, time of payment etc.), call-information are (for example, the duration of call, air time section, talkthrough rate, call language
Speed, the tone and voice etc.) etc..
Term " employee's portrait " in present disclosure refers to the tag set for portraying employee's feature, such as employee
Age, region, place city, constellation, educational background, family structure, marital status, hobby, the length of service, personality, urges at gender
Receipts ability, professional quality, call habit etc..
Term " employee's essential information " in present disclosure can include but is not limited to age of user, gender, region,
Place city, constellation, educational background, family structure, marital status, hobby, length of service etc..
Term " employee's image information " in present disclosure can include but is not limited to call-information (for example, when call
Length, air time section, talkthrough rate, converse word speed, the tone and voice etc.) etc..
Term " product portrait " in present disclosure refers to the tag set of the feature for portraying product, such as product
When developing design towards purchasing group, product in effective sale when towards purchasing group etc..
Term " first ", " second " in present disclosure are only used for description reference, purpose or certain specific things, and cannot
It is interpreted as indication or suggestion relative importance, or implicitly indicates the quantity of indicated technical characteristic.
Term " multiple " in present disclosure refers to two or more.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as part of specification.For the company between each unit in attached drawing
Line, it is only for convenient for explanation, indicate that the unit at least line both ends is in communication with each other, it is not intended that the non-line of limitation
Unit between can not communicate.
Before introducing the solution that present disclosure is proposed, applicant introduces the current prior art first
Status.It is always the emphasis of Wang Dai company concern, for different overdue shapes as the important link of risk control after loan
State carries out corresponding collection strategy, current traditional collection mode is main or short message, make a phone call, the person of urging visits, general short message
For the shorter case of overdue number of days, overdue number of days, which reaches to attend a banquet after certain number of days, to be started to carry out phone collection, part for collection
Overdue serious case can arrange the person of urging to visit collection.Applicant has perfect post-loan management and collection body for storage client
System has formulated different collection strategies for C, M1, M2, M3, M3+ or more respectively respectively, in terms of personal management, focuses on echelon
Construction, matches optimal personnel for different overdue state clients.
But traditional collection mode needs to consume a large amount of manpower with the increase of portfolio, exceedes simultaneously for same
The different client of phase status attribute fails the optimal collection method for having tactic, and such collection is usually associated with the continuous of cost
Increase, brings more cost pressures to company, collection effect also fails to be significantly increased.
Collection and risk control fail to carry out each client the attributive analysis of science and be directed to after traditional loan
Property risk control, applicant, which establishes big data and borrows rear platform, for these mass datas passes through big data analysis, text
It excavates, voice semantic analysis establishes the collection scorecard for each storage client, in combination with the people currently rapidly developed
Work intellectual technology allows machine to replace manually going to solve some basic pre- collections and short-term overdue collection is reminded, mid-term
Overdue collection intervention, and the depth of promise breaking client are excavated, thus after having built a set of intelligence loan for each client
Risk platform fundamentally realizes post-loan management and Risk-warning to each storage client, is improving the same of collection efficiency
When also achieve reduction to company operation cost.
Specifically, the post-loan management in industry is concentrated on for different overdue state visitors at present in terms of post-loan management
The management at family, early warning and risk identification after seldom being borrowed to the client that do not break a contract, for example, present collection is primarily directed to
Through overdue client, the fund being in arrears with recovered by short message forms such as visiting abroad of making a phone call, is a kind of risk of hysteresis quality
Control；Also distinctive management is not carried out to the client of same overdue state, for example, the client of same overdue state now, urges
Art and collection method are identical if member is used, can be by these same overdue states without a kind of method of directiveness
Client carries out a differentiation, and client's method therefor needs of different attribute are had any different；Risk control is still with people after present loan
Based on work collection, fail deeply to apply artificial intelligence technology into the various aspects in collection management and Risk-warning.
On the other hand, in terms of collection scorecard, collection scorecard mainly studies the history refund behavior of client in industry,
For history collection record, then there is no form profound excavation.In addition, the A card that do not link (scorecard when application) is to client
Carry out various dimensions modeling.Furthermore most of scorecard is all static scorecard in industry, fails iteration day by day, also just formed not
Dynamic grading card, so that more accurate carry out effective post-loan management to client.
Based on the above status, present disclosure applicant proposed solution explained in greater detail below, the solutions
Scheme can be applied among the scene of processing collection business shown in FIG. 1, and the embodiment that Fig. 1 shows present disclosure can be with
The signal architecture diagram of exemplary environments 100 applied to the processing collection business therein based on artificial intelligence.
Exemplary environments 100 include the employee 101-103 of mechanism (for example, internet financing corporation), service process platform
110.Employee 101-103 can be connected to service process platform 110 by wired or wireless way.Service process platform 110 can
To handle various businesses related to user.In exemplary environments 110, service process platform 110 is connected to 111 (example of network
Such as, wired or wireless communication network), and network 111 is via communication link 112 (for example, wired or wireless link) and mechanism
The user 121-123 serviced is connected.Service process platform 110 can provide such as speech processing module, mail treatment mould
The dialog process modules such as block, short message processing module enable the employee 101-103 of mechanism by these modules, via network
111 interact formula dialogue (for example, voice, mail, short message etc.) with the user 121-123 that mechanism is serviced.As user 121-
When user in 123 fails product (for example, the financial product) for repaying mechanism on time, mechanism can handle associated with the product
Collection business.As previously mentioned, mechanism is mostly urged with dialogue (for example, voice, mail, short message etc.) for the purpose of control cost
Based on receipts, wherein the employee in the employee 101-103 of mechanism is enable to interact formula dialogue via network 111 and the user
To carry out collection.It should be appreciated that the quantity of employee and user shown in Fig. 1 are only signals rather than limit, can be any
Quantity.
Fig. 2 shows the methods 200 of the processing collection business based on artificial intelligence of the embodiment of foundation present disclosure
Flow chart.It can be seen from the figure that this method 200 includes at least following five steps, i.e., it will first in method and step 210
Obtain collection scorecard set；Next, obtaining the current overdue state of client to be assessed and to also in method and step 220
Money parameter；Then, it in method and step 230, is commented according to the current overdue state and the parameter to be refunded from the collection
Divide selection one and the matched collection scorecard of client to be assessed in card set；Next, in method and step 240,
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And finally, in method and step 250,
The collection mode for being used for the client to be assessed is determined based on collection scoring.
Thus, it is possible to different clients pass through different collection scorecards to score, so that scoring is more
Accurately reflect the actual state of assessed client；In addition, the scoring will be used in the determination of subsequent collection strategy, thus
The collection strategy for more meeting the client can be formulated, collection effect is improved.
Acquired collection scorecard set obtains in the following manner in Fig. 2, specifically, model will be obtained first
Data acquisition system.
More particularly establishing model data set, that is, current data with existing is subjected to Classifying Sum, it is main
Include the following categories information:
A, client's essential information, for example, extract client apply loan when information, mainly include monthly income, the age, gender,
City of residence, industry, educational background etc..
B, client refunds record: according to refund behavior of the client before model execution, the amount of money is gone back in extraction the field moon, when refund
Between, odd corpus ratio, for the first time overdue issue etc., while establishing refund and recording derivative field about 80, such as: history maximum is overdue
Maximum overdue 6 monthly variety ratios of number of days of number of days/access moon MOB, 3 months etc.
C, client's collection records: the person of urging can form corresponding collection record in collection system when carrying out collection to client,
Extract these text informations, comprising: Bounced check number, promise refund number, preceding collection customer status etc., meanwhile, in the person of urging
When with client connection, speech recognition technology and natural language processing (NLP:Natural Language are utilized
Processing) technology extracts the keyword that client answers, and matches currently existing susceptibility dictionary, forms the conduct of sensitive word field
Variable, these sensitive word fields include positive sensitive word field, neutral sensitive word field, passive sensitive word field, each field
Different labelings is had below, if some label is hit in the answer of client, that is, is classified as certain below some sensitive word
One kind, for example: the person of urging is with the interacting of client, and the person of urging asks when client refunds, and client answers: " out of funds ", that is, hitting
" be not intended to refund " label under passive sensitive word；Meanwhile according to these labels, more derivative variables, such as " mistake can also be established
Three months clients are gone to be not intended to refund number " etc., form more data dimensions
D, it client association side's information: utilizes knowledge mapping (Knowledge Graph), finds out strongly connected graph, and it is marked
Note comes out, and forms high wind danger relating dot, is based on dynamic network analysis, forms the strong related information of dynamic client and extracts, establishes
Client association information field, as client and black intermediary contact it is strong and weak (for example: when client A is related with B in knowledge mapping
System's transmitting, B and C have relation transmission, and C is certain black intermediary, then it is strong and weak to give relationship by the number of levels in knowledge mapping
Coefficient), client region venture entrepreneur (for example: it is often real in some position of zone of action and black intermediary according to client A
Body has relation transmission, then it is assumed that client A has default risk), the main method of knowledge mapping building at present is neural network.
Next, this method further includes data cleansing and processing step, data cleansing is carried out to the model data set
And by the model data set after cleaning according to whether overdue be divided into not overdue model data set and overdue model data set.
Cross validation method is folded using K to be respectively divided into not overdue model data set and overdue model data set and not overdue mould
Associated first training set of type data acquisition system and the first verifying collection and the second instruction associated with overdue model data set
Practice collection and the second verifying collection.
Specifically, carrying out data cleansing, screening, processing based on the model data set in former approach step.First
Need the division that modeling object is carried out to model data set: i.e. modeling data is integrated in tradition modeling, uses data mining and machine
The method of device study establishes collection scorecard；This modeling breaks tradition and models large and complete integral analysis, carries out to client
Divide group.Such as client can be divided into normal clients, and overdue 0-7 days clients, overdue 8-30 days clients.For normal clients, meter
Calculate its odd corpus ratio, by normal clients according to odd corpus ratio cut partition be two major classes；For overdue client, it is close to calculate it
Overdue client is divided two major classes according to returned money rate by three months returned money rates.Six formed for different objective groups are calculated based on above
Cover scorecard system.
Preferably, it for normal clients, predicts the following generation Default Probability, precisely estimating for the potential loss of client will be turned
It is melted into the marking of model, auxiliary collection department formulates pre- Anti-lost strategy.For overdue 0-7 days clients, predict client's future also
Money probability will be converted to model marking to precisely estimating for the potential refund probability of client, and auxiliary collection department, which formulates, urges back strategy.
For overdue 8-30 days clients, client's future refund probability is predicted, precisely excavate potential high probability refund client, assist collection
Department's fast accurate emphasis collection height scoring client.Modeling time window is making loans before 6 months, and modeling sample is pressed at random
It samples according to certain proportion, using k-fold cross validation (k- folds cross validation) method to model partition training
Collection collects with verifying, finds the super ginseng value so that model generalization best performance.
Finally, being based respectively on first training set and the first verifying collection using logistic regression method and XGBoost method
The first collection scorecard associated with the not overdue model data set is obtained, and is based on second training set and the
Two verifying collection obtain the second collection scorecard associated with the overdue model data set.
Specifically, after cleaning, processing and the screening for completing data, for there is the variable of independence with decision tree, card
Square branch mailbox does grouping and calculates, and rejects the small variable of iv value, to remaining variable with logistic regression (Logistic Regression:
LR) and XGBoost algorithm determines the variable for eventually entering into model, determines that the effect of model is excellent by AUC and KS index
It is bad, according to credit scoring marking logic, determine the grouping score of each variable, and go the accurate of verifying model result with test set
Property, ultimately form the comprehensive score and risk class for collection object.
Preferably, this method 200 can also include being scored according to updated model data set first collection
Card and the second collection scorecard are iterated, to obtain updated first collection scorecard and updated second collection
Scorecard.
Specifically, the variable and each group score value that are obtained according to abovementioned steps, determine scorecard marking logic, and dispose
In regulation engine, daily iterative model formulates the collection strategy for being directed to different collection objects according to model result, realizes
The effective communication of the person of urging and client and the real-time update of data and feedback form urging for a set of various dimensions based on dynamic data
Receive scorecard.
Above-mentioned model data set includes essential information associated with client, record of refunding, collection record and/or closes
Connection side's information.And collection record includes Bounced check number, promises to undertake refund number and/or a preceding collection customer status.Wherein, described
Affiliated party's information includes the third-party information being determined as by knowledge mapping with the client association.
It is further preferable that this method can also include the following steps (not shown in FIG.):
The overdue model data set is further divided into the first overdue model data according to the length of overdue time
Set and the second overdue model data set；And
The described first overdue model data set, which is based respectively on, using logistic regression method and XGBoost method obtains third
Collection scorecard, and the 4th collection scorecard is obtained based on the described second overdue model data set.
It is further preferable that this method can also include the following steps (not shown in FIG.):
Not overdue model data set is divided into the first not overdue model data set and second according to odd corpus ratio
Not overdue model data set；
Overdue model data set is divided into the first overdue model data collection according to the length of returned money rate and overdue time
Conjunction, the second overdue model data set, the overdue model data set of third and the 4th overdue model data set；And
It is obtained respectively and the described first not overdue model data set, institute using logistic regression method and XGBoost method
State the second not overdue model data set, the first overdue model data set, the second overdue model data set, institute
State the overdue model data set of third and the 4th overdue associated six collection scorecards of model data set.
Traditional collection scorecard depends primarily on the essential information of client and record of refunding, present disclosure are urged in tradition
On the basis for receiving scorecard, the collection record of structuring, and the non-structured dynamic using the crawl of NLP technology are increased
Collection record, meanwhile, present disclosure utilizes the analysis of knowledge mapping unlike traditional scorecard, increases for being urged
The extraction of the real-time related information of object not only utilizes the behavior for being urged object to predict its following refund wish, more
By knowledge mapping, the refund wish for being urged its future of affiliated person's information prediction of object is utilized.And in above six models,
We are excluded without deviation caused by customer grouping and noise, depth excavates client's by modeling respectively to six class clients
The communication data of history refund record and client and collection personnel, so that model is more accurate, auxiliary establishes more accurate plan
Slightly.
After having obtained above-mentioned accurately scorecard, the applicant of present disclosure expects in the following manner utilizing
Obtained scorecard, to reduce operations risks and save collection cost.
Specifically, carrying out different phase after above-mentioned collection scorecard is established to client and targetedly intelligently urging
It receives.For example, the hospitable family of collection scorecard in normal clients: robot refund in N-3 days is reminded, the collection scoring in normal clients
Block bad client: short message is reminded within N-7 days, and N-5 days robots, which refund, to be reminded, and repayment date inner machine people's collection in N+3 days utilizes current city
Existing voice technology on field provides dialect collection for the borrower of different regions.The beginning that do not refund after repayment date N+3 days
Artificial collection.The client preferential collection low for score, and the person of urging for arranging collection ability strong carries out collection, the high client of score
Client's collection a little later that can be higher than score, can match the slightly weak person of urging of collection ability.Repayment date is answered in N representative among the above.
And for the hospitable family of collection scorecard in overdue client: N-3 days robots, which refund, to be reminded, N+1 days inner machine people sides of repayment date
Say collection, the artificial collection of the beginning that do not refund after repayment date N+1 days.Collection scorecard bad client in overdue client: continue people
Work collection.The client preferential collection low for score, and the person of urging for arranging collection ability strong carries out collection, the high client of score can
With the client a little later collection higher than score, the slightly weak person of urging of collection ability can be matched.The new overdue number of refund & is formed daily
According to, and enter collection big data platform, and then the daily iteration of model can be carried out.
In addition, the system according to present disclosure can also establish history collection voice platform, history collection voice is integrated
Data.Specifically, speech analysis can be carried out by the voice data of collecting robot people and client and the person of urging and client,
Speech analysis model is established, optimal words art is found out, rate is urged back in promotion.Optionally, intelligent quality inspection platform can be established.It is specific and
Speech, current collection quality inspection carry out quality inspection mainly by way of artificial playback, there is very strong hysteresis quality and one-sidedness, people
Working medium inspection platform mainly advises speech database by establishing non-conjunction, and real-time recording is connect with attending a banquet, and is carrying out electricity once attending a banquet
Sensitive word is touched when talking about collection or non-compliance words art occurs, is sounded an alarm at once.Preferably, by speech recognition and
Voice semantic analysis technology establishes Risk-warning platform after intelligence is borrowed.
Additionally or alternatively, the above method can be by computer program product, i.e. computer readable storage medium is real
It is existing.Computer program product may include computer readable storage medium, containing for executing each of present disclosure
The computer-readable program instructions of aspect.Computer readable storage medium, which can be, can keep and store by instruction execution equipment
The tangible device of the instruction used.Computer readable storage medium for example can be but not limited to storage device electric, magnetic storage is set
Standby, light storage device, electric magnetic storage apparatus, semiconductor memory apparatus or above-mentioned any appropriate combination.It is computer-readable
The more specific example (non exhaustive list) of storage medium includes: portable computer diskette, hard disk, random access memory
(RAM), read-only memory (ROM), erasable programmable read only memory (EPROM or flash memory), static random access memory
(SRAM), Portable compressed disk read-only memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding
Equipment, the punch card for being for example stored thereon with instruction or groove internal projection structure and above-mentioned any appropriate combination.Here
Used computer readable storage medium is not interpreted as instantaneous signal itself, such as radio wave or other Free propagations
Electromagnetic wave, the electromagnetic wave (for example, the light pulse for passing through fiber optic cables) propagated by waveguide or other transmission mediums or pass through
The electric signal of electric wire transmission.
Fig. 3 shows the device 300 of the processing collection business based on artificial intelligence of the embodiment according to present disclosure
Schematic diagram.It should be appreciated that the method 200 of the processing collection business based on artificial intelligence in Fig. 2 may be implemented in device 300
Function.As can be seen from Figure 3 the device 300 of the processing collection business based on artificial intelligence includes processor 301 and memory
302.Processor 301 can be central processing unit (CPU), microcontroller, specific integrated circuit (ASIC), digital signal processor
(DSP), it field programmable gate array (FPGA) or other programmable logic device or is configured as realizing present disclosure
One or more integrated circuits of embodiment.Memory 302 may include volatile memory, also may include non-volatile deposit
Reservoir, ROM, RAM, mobile disk, disk, CD and USB flash disk etc..Make when storing instruction execution in the memory 302
Processor 301 executes following operation:
Obtain collection scorecard set；
Obtain the current overdue state and parameter to be refunded of client to be assessed；
According to the current overdue state and the parameter to be refunded select from the collection scorecard set one and
The matched collection scorecard of client to be assessed；
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And
The collection mode for being used for the client to be assessed is determined based on collection scoring.
In one embodiment according to present disclosure, obtaining collection scorecard set further comprises:
Obtain model data set；
Data cleansing is carried out and by the model data set after cleaning according to whether overdue point to the model data set
For not overdue model data set and overdue model data set；
Using K fold cross validation method respectively will not overdue model data set and overdue model data set be divided into
Associated first training set of not overdue model data set and the first verifying collection and associated with overdue model data set
The second training set and second verifying collection；And
Using logistic regression method and XGBoost method be based respectively on first training set and first verifying collection obtain with
The associated first collection scorecard of not overdue model data set, and based on second training set and the second verifying
Collection obtains the second collection scorecard associated with the overdue model data set.
In one embodiment according to present disclosure, the method also includes:
The first collection scorecard and the second collection scorecard are carried out according to updated model data set
Iteration, to obtain updated first collection scorecard and updated second collection scorecard.
In one embodiment according to present disclosure, the model data set includes associated with client basic
Information, record of refunding, collection record and/or affiliated party's information.
In one embodiment according to present disclosure, the collection record includes Bounced check number, promises to undertake refund number
An and/or preceding collection customer status.
In one embodiment according to present disclosure, affiliated party's information includes being determined as by knowledge mapping
With the third-party information of the client association.
In one embodiment according to present disclosure, execute the processor
It operates below:
The overdue model data set is further divided into the first overdue model data according to the length of overdue time
Set and the second overdue model data set；And
The described first overdue model data set, which is based respectively on, using logistic regression method and XGBoost method obtains third
Collection scorecard, and the 4th collection scorecard is obtained based on the described second overdue model data set.
In one embodiment according to present disclosure, execute the processor
It operates below:
Not overdue model data set is divided into the first not overdue model data set and second according to odd corpus ratio
Not overdue model data set；
Overdue model data set is divided into the first overdue model data collection according to the length of returned money rate and overdue time
Conjunction, the second overdue model data set, the overdue model data set of third and the 4th overdue model data set；And
It is obtained respectively and the described first not overdue model data set, institute using logistic regression method and XGBoost method
State the second not overdue model data set, the first overdue model data set, the second overdue model data set, institute
State the overdue model data set of third and the 4th overdue associated six collection scorecards of model data set.
In general, the various example embodiments of present disclosure can in hardware or special circuit, software, firmware, patrol
Volume, or any combination thereof in implement.Some aspects can be implemented within hardware, and other aspect can by controller,
Implement in the firmware or software that microprocessor or other calculating equipment execute.When the various aspects diagram of the embodiment of present disclosure
Or when being described as block diagram, flow chart or using other certain graphical representations, it will be understood that box described herein, device, system, skill
Art or method can be used as unrestricted example in hardware, software, firmware, special circuit or logic, common hardware or control
Implement in device or other calculating equipment or its certain combination.
It should be noted that although being referred to several modules or unit of device in the detailed description above, this stroke
It point is only exemplary rather than enforceable.In fact, according to the embodiment of present disclosure, it is above-described two or more
The feature and function of module can embody in a module.Conversely, the feature and function of an above-described module can
It is to be embodied by multiple modules with further division.
The foregoing is merely the embodiment alternative embodiments of present disclosure, are not limited to the implementation of present disclosure
Example, for those skilled in the art, the embodiment of present disclosure can have various modifications and variations.It is all in the disclosure
Within the spirit and principle of the embodiment of content, made any modification, equivalence replacement, improvement etc. should be included in the disclosure
Within the protection scope of the embodiment of content.
Although describing the embodiment of present disclosure by reference to several specific embodiments, it should be appreciated that, this public affairs
The embodiment for opening content is not limited to disclosed specific embodiment.The embodiment of present disclosure is intended to cover in appended right
It is required that spirit and scope in included various modifications and equivalent arrangements.Scope of the appended claims meet broadest solution
It releases, to include all such modifications and equivalent structure and function.
Claims (17)
1. a kind of method of the processing collection business based on artificial intelligence, which is characterized in that the described method includes:
Obtain collection scorecard set；
Obtain the current overdue state and parameter to be refunded of client to be assessed；
According to the current overdue state and the parameter to be refunded selected from the collection scorecard set one with it is described
The matched collection scorecard of client to be assessed；
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And
The collection mode for being used for the client to be assessed is determined based on collection scoring.
2. the method according to claim 1, wherein acquisition collection scorecard set further comprises:
Obtain model data set；
Data cleansing is carried out and by the model data set after cleaning according to whether overdue be divided into not to the model data set
Overdue model data set and overdue model data set；
Cross validation method is folded using K not overdue model data set and overdue model data set are divided into and are not exceeded respectively
Associated first training set of phase model data set and the first verifying collection and associated with overdue model data set the
Two training sets and the second verifying collection；And
Using logistic regression method and XGBoost method be based respectively on first training set and first verifying collection obtain with it is described
The not overdue associated first collection scorecard of model data set, and obtained based on second training set and the second verifying collection
The second collection scorecard associated with the overdue model data set.
3. according to the method described in claim 2, it is characterized in that, the method also includes:
The first collection scorecard and the second collection scorecard are iterated according to updated model data set,
To obtain updated first collection scorecard and updated second collection scorecard.
4. according to the method in claim 2 or 3, which is characterized in that the model data set includes associated with client
Essential information, refund record, collection record and/or affiliated party's information.
5. according to the method described in claim 4, it is characterized in that, collection record includes Bounced check number, promises to undertake refund time
A several and/or preceding collection customer status.
6. according to the method described in claim 4, it is characterized in that, affiliated party's information includes being determined by knowledge mapping
For the third-party information with the client association.
7. according to the method in claim 2 or 3, which is characterized in that the method also includes:
The overdue model data set is further divided into the first overdue model data set according to the length of overdue time
With the second overdue model data set；And
The described first overdue model data set, which is based respectively on, using logistic regression method and XGBoost method obtains third collection
Scorecard, and the 4th collection scorecard is obtained based on the described second overdue model data set.
8. according to the method in claim 2 or 3, which is characterized in that the method also includes:
Not overdue model data set is divided into the first not overdue model data set and second according to odd corpus ratio not exceed
Phase model data set；
Overdue model data set is divided into the first overdue model data set, according to the length of returned money rate and overdue time
Two overdue model data sets, the overdue model data set of third and the 4th overdue model data set；And
It is obtained respectively and the described first not overdue model data set, described the using logistic regression method and XGBoost method
Two not overdue model data sets, the first overdue model data set, the second overdue model data set, described
Three overdue model data sets and the 4th overdue associated six collection scorecards of model data set.
9. a kind of device of the processing collection business based on artificial intelligence characterized by comprising
Processor；And
Memory makes the processor execute following operation when executed for storing instruction:
Obtain collection scorecard set；
Obtain the current overdue state and parameter to be refunded of client to be assessed；
According to the current overdue state and the parameter to be refunded selected from the collection scorecard set one with it is described
The matched collection scorecard of client to be assessed；
It is scored using the collection that selected collection scorecard obtains the client to be assessed；And
The collection mode for being used for the client to be assessed is determined based on collection scoring.
10. device according to claim 9, which is characterized in that obtaining collection scorecard set further comprises:
Obtain model data set；
Data cleansing is carried out and by the model data set after cleaning according to whether overdue be divided into not to the model data set
Overdue model data set and overdue model data set；
Cross validation method is folded using K not overdue model data set and overdue model data set are divided into and are not exceeded respectively
Associated first training set of phase model data set and the first verifying collection and associated with overdue model data set the
Two training sets and the second verifying collection；And
Using logistic regression method and XGBoost method be based respectively on first training set and first verifying collection obtain with it is described
The not overdue associated first collection scorecard of model data set, and obtained based on second training set and the second verifying collection
The second collection scorecard associated with the overdue model data set.
11. device according to claim 10, which is characterized in that the method also includes:
The first collection scorecard and the second collection scorecard are iterated according to updated model data set,
To obtain updated first collection scorecard and updated second collection scorecard.
12. device described in 0 or 11 according to claim 1, which is characterized in that the model data set includes related to client
The essential information of connection, record of refunding, collection record and/or affiliated party's information.
13. device according to claim 12, which is characterized in that the collection record includes Bounced check number, promises to undertake and refund
Number and/or a preceding collection customer status.
14. device according to claim 12, which is characterized in that affiliated party's information includes true by knowledge mapping
It is set to the third-party information with the client association.
15. device described in 0 or 11 according to claim 1, which is characterized in that also make the place when executed
It manages device and executes following operation:
The overdue model data set is further divided into the first overdue model data set according to the length of overdue time
With the second overdue model data set；And
The described first overdue model data set, which is based respectively on, using logistic regression method and XGBoost method obtains third collection
Scorecard, and the 4th collection scorecard is obtained based on the described second overdue model data set.
16. device described in 0 or 11 according to claim 1, which is characterized in that also make the place when executed
It manages device and executes following operation:
Not overdue model data set is divided into the first not overdue model data set and second according to odd corpus ratio not exceed
Phase model data set；
Overdue model data set is divided into the first overdue model data set, according to the length of returned money rate and overdue time
Two overdue model data sets, the overdue model data set of third and the 4th overdue model data set；And
It is obtained respectively and the described first not overdue model data set, described the using logistic regression method and XGBoost method
Two not overdue model data sets, the first overdue model data set, the second overdue model data set, described
Three overdue model data sets and the 4th overdue associated six collection scorecards of model data set.
17. a kind of computer readable storage medium, including computer executable instructions, the computer executable instructions are in device
Described device is made to execute the processing collection according to any one of claim 1 to 8 based on artificial intelligence when middle operation
The method of business.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811434822.5A CN109509086A (en) | 2018-11-28 | 2018-11-28 | The method, apparatus and storage medium of processing collection business based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811434822.5A CN109509086A (en) | 2018-11-28 | 2018-11-28 | The method, apparatus and storage medium of processing collection business based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109509086A true CN109509086A (en) | 2019-03-22 |
Family
ID=65750991
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811434822.5A Pending CN109509086A (en) | 2018-11-28 | 2018-11-28 | The method, apparatus and storage medium of processing collection business based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109509086A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110414714A (en) * | 2019-06-28 | 2019-11-05 | 上海上湖信息技术有限公司 | Returned money probability forecasting method and device, the readable storage medium storing program for executing of overdue event |
| CN110619434A (en) * | 2019-09-17 | 2019-12-27 | 北谷电子有限公司上海分公司 | Method and device for predicting repayment probability, electronic equipment and storage medium |
| CN110704496A (en) * | 2019-09-02 | 2020-01-17 | 山东科技大学 | Proposed aggregation method based on asymmetric response |
| CN110796430A (en) * | 2019-10-08 | 2020-02-14 | 中国建设银行股份有限公司 | Overdue transaction data processing method and device |
| CN111062518A (en) * | 2019-11-22 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | Method, device and storage medium for processing hastening service based on artificial intelligence |
| CN111061948A (en) * | 2019-11-25 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | User label recommendation method and device, computer equipment and storage medium |
| CN111192136A (en) * | 2019-12-24 | 2020-05-22 | 中信百信银行股份有限公司 | Credit service collection method and device, electronic equipment and storage medium |
| CN111311397A (en) * | 2020-02-13 | 2020-06-19 | 上海凯岸信息科技有限公司 | Scheme for improving voice call-out robot collection fee collection rate by combining scoring card model and ABtest |
| CN111383093A (en) * | 2020-03-06 | 2020-07-07 | 北京网众共创科技有限公司 | Intelligent overdue bill collection method and system |
| CN111401845A (en) * | 2020-03-17 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Service processing method and device |
| CN111461505A (en) * | 2020-03-18 | 2020-07-28 | 黑龙江正泽信服务外包有限公司 | Artificial labeling system for intelligent matching of collection urging information |
| CN111553574A (en) * | 2020-04-16 | 2020-08-18 | 上海诚收信息科技有限公司 | Case allocation method and device, electronic device and computer-readable storage medium |
| CN111581252A (en) * | 2020-05-04 | 2020-08-25 | 上海维信荟智金融科技有限公司 | Dynamic collection urging method and system based on multi-dimensional information data |
| CN111986018A (en) * | 2019-05-22 | 2020-11-24 | 财付通支付科技有限公司 | Bill collection prompting method and device based on preset collection prompting system and electronic equipment |
| CN112785419A (en) * | 2021-01-22 | 2021-05-11 | 中信银行股份有限公司 | Prompt receipt action recommendation method and device |
| CN113269632A (en) * | 2021-05-27 | 2021-08-17 | 深圳市蓬莱产业科技有限公司 | Label library construction and money return efficiency method based on bad asset disposal efficiency |
| CN117476011A (en) * | 2023-12-28 | 2024-01-30 | 杭州度言软件有限公司 | Method and system for identifying object to be induced and received based on voice signal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296391A (en) * | 2016-08-08 | 2017-01-04 | 联动优势科技有限公司 | A kind of assessment exceeds the time limit the method and apparatus of probability |
| CN106952155A (en) * | 2017-03-08 | 2017-07-14 | 深圳前海纵腾金融科技服务有限公司 | A kind of collection method and device based on credit scoring |
| CN108830653A (en) * | 2018-06-13 | 2018-11-16 | 国网福建省电力有限公司 | A kind of intelligent collection electricity charge method based on AI |
-
2018
- 2018-11-28 CN CN201811434822.5A patent/CN109509086A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296391A (en) * | 2016-08-08 | 2017-01-04 | 联动优势科技有限公司 | A kind of assessment exceeds the time limit the method and apparatus of probability |
| CN106952155A (en) * | 2017-03-08 | 2017-07-14 | 深圳前海纵腾金融科技服务有限公司 | A kind of collection method and device based on credit scoring |
| CN108830653A (en) * | 2018-06-13 | 2018-11-16 | 国网福建省电力有限公司 | A kind of intelligent collection electricity charge method based on AI |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111986018A (en) * | 2019-05-22 | 2020-11-24 | 财付通支付科技有限公司 | Bill collection prompting method and device based on preset collection prompting system and electronic equipment |
| CN110414714A (en) * | 2019-06-28 | 2019-11-05 | 上海上湖信息技术有限公司 | Returned money probability forecasting method and device, the readable storage medium storing program for executing of overdue event |
| CN110704496A (en) * | 2019-09-02 | 2020-01-17 | 山东科技大学 | Proposed aggregation method based on asymmetric response |
| CN110704496B (en) * | 2019-09-02 | 2023-01-24 | 山东科技大学 | Proposed aggregation method based on asymmetric response |
| CN110619434A (en) * | 2019-09-17 | 2019-12-27 | 北谷电子有限公司上海分公司 | Method and device for predicting repayment probability, electronic equipment and storage medium |
| CN110796430A (en) * | 2019-10-08 | 2020-02-14 | 中国建设银行股份有限公司 | Overdue transaction data processing method and device |
| CN111062518A (en) * | 2019-11-22 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | Method, device and storage medium for processing hastening service based on artificial intelligence |
| CN111062518B (en) * | 2019-11-22 | 2023-06-09 | 成都铂锡金融信息技术有限公司 | Method, device and storage medium for processing collect-promoting business based on artificial intelligence |
| CN111061948A (en) * | 2019-11-25 | 2020-04-24 | 成都铂锡金融信息技术有限公司 | User label recommendation method and device, computer equipment and storage medium |
| CN111061948B (en) * | 2019-11-25 | 2023-05-23 | 成都铂锡金融信息技术有限公司 | User tag recommendation method and device, computer equipment and storage medium |
| CN111192136A (en) * | 2019-12-24 | 2020-05-22 | 中信百信银行股份有限公司 | Credit service collection method and device, electronic equipment and storage medium |
| CN111311397A (en) * | 2020-02-13 | 2020-06-19 | 上海凯岸信息科技有限公司 | Scheme for improving voice call-out robot collection fee collection rate by combining scoring card model and ABtest |
| CN111383093A (en) * | 2020-03-06 | 2020-07-07 | 北京网众共创科技有限公司 | Intelligent overdue bill collection method and system |
| CN111401845A (en) * | 2020-03-17 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Service processing method and device |
| CN111401845B (en) * | 2020-03-17 | 2022-07-08 | 支付宝(杭州)信息技术有限公司 | Service processing method and device |
| CN111461505A (en) * | 2020-03-18 | 2020-07-28 | 黑龙江正泽信服务外包有限公司 | Artificial labeling system for intelligent matching of collection urging information |
| CN111553574A (en) * | 2020-04-16 | 2020-08-18 | 上海诚收信息科技有限公司 | Case allocation method and device, electronic device and computer-readable storage medium |
| CN111581252A (en) * | 2020-05-04 | 2020-08-25 | 上海维信荟智金融科技有限公司 | Dynamic collection urging method and system based on multi-dimensional information data |
| CN112785419A (en) * | 2021-01-22 | 2021-05-11 | 中信银行股份有限公司 | Prompt receipt action recommendation method and device |
| CN112785419B (en) * | 2021-01-22 | 2023-05-05 | 中信银行股份有限公司 | Method and device for recommending collecting action |
| CN113269632A (en) * | 2021-05-27 | 2021-08-17 | 深圳市蓬莱产业科技有限公司 | Label library construction and money return efficiency method based on bad asset disposal efficiency |
| CN117476011A (en) * | 2023-12-28 | 2024-01-30 | 杭州度言软件有限公司 | Method and system for identifying object to be induced and received based on voice signal |
| CN117476011B (en) * | 2023-12-28 | 2024-03-01 | 杭州度言软件有限公司 | Method and system for identifying object to be induced and received based on voice signal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109509086A (en) | The method, apparatus and storage medium of processing collection business based on artificial intelligence | |
| CN110796470B (en) | Data analysis system for market subject supervision and service | |
| CN107424070A (en) | A kind of loan user credit ranking method and system based on machine learning | |
| CN106407999A (en) | Rule combined machine learning method and system | |
| CN109767312B (en) | Credit evaluation model training and evaluation method and device | |
| CN109829631A (en) | A kind of business risk early warning analysis method and system based on memory network | |
| EP1922678A2 (en) | Apparatus and method for simulating an analytic value chain | |
| CN110276679A (en) | A kind of network individual credit fraud detection method towards deep learning | |
| CN110532398A (en) | Family's map method for auto constructing based on multitask united NNs model | |
| CN104616198A (en) | P2P (peer-to-peer) network lending risk prediction system based on text analysis | |
| CN108022016A (en) | A kind of Prediction of Stock Price method and system based on artificial intelligence | |
| CN109063921A (en) | Optimized treatment method, device, computer equipment and the medium of customer risk early warning | |
| CN111199469A (en) | User payment model generation method and device and electronic equipment | |
| KR101799395B1 (en) | Server, method and terminal for loan judgment | |
| CN109523185A (en) | The method, apparatus and storage medium of collection scorecard are generated based on artificial intelligence | |
| Zhu et al. | Loan default prediction based on convolutional neural network and LightGBM | |
| TWM583089U (en) | Smart credit risk assessment system | |
| CN109635289A (en) | Entry classification method and audit information abstracting method | |
| CN110738565A (en) | Real estate finance artificial intelligence composite wind control model based on data set | |
| CN110503481A (en) | Method and device is determined based on the target user of graph theory model | |
| Mittal et al. | A study on credit risk assessment in banking sector using data mining techniques | |
| CN114462742A (en) | Risk prompting method, device, equipment and computer storage medium | |
| Dash et al. | An empirical study for customer relationship management in banking sector using machine learning techniques | |
| KR102519878B1 (en) | Apparatus, method and recording medium storing commands for providing artificial-intelligence-based risk management solution in credit exposure business of financial institution | |
| Lin et al. | Research on Credit Big Data Algorithm Based on Logistic Regression |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190322 |