CN106651588A - Underwriting method and apparatus for logistics insurance policy - Google Patents
Underwriting method and apparatus for logistics insurance policy Download PDFInfo
- Publication number
- CN106651588A CN106651588A CN201610988595.5A CN201610988595A CN106651588A CN 106651588 A CN106651588 A CN 106651588A CN 201610988595 A CN201610988595 A CN 201610988595A CN 106651588 A CN106651588 A CN 106651588A
- Authority
- CN
- China
- Prior art keywords
- subjectinsured
- information
- core
- factor
- learning model
- 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
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013136 deep learning model Methods 0.000 claims abstract description 123
- 239000013598 vector Substances 0.000 claims abstract description 48
- 108090000623 proteins and genes Proteins 0.000 claims description 71
- 238000012549 training Methods 0.000 claims description 39
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000012856 packing Methods 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 11
- 230000001143 conditioned effect Effects 0.000 claims description 9
- 230000008676 import Effects 0.000 claims description 8
- 238000013135 deep learning Methods 0.000 claims description 6
- 108010000499 Thromboplastin Proteins 0.000 claims description 3
- 102000002262 Thromboplastin Human genes 0.000 claims description 3
- AGVAZMGAQJOSFJ-WZHZPDAFSA-M cobalt(2+);[(2r,3s,4r,5s)-5-(5,6-dimethylbenzimidazol-1-yl)-4-hydroxy-2-(hydroxymethyl)oxolan-3-yl] [(2r)-1-[3-[(1r,2r,3r,4z,7s,9z,12s,13s,14z,17s,18s,19r)-2,13,18-tris(2-amino-2-oxoethyl)-7,12,17-tris(3-amino-3-oxopropyl)-3,5,8,8,13,15,18,19-octamethyl-2 Chemical group [Co+2].N#[C-].[N-]([C@@H]1[C@H](CC(N)=O)[C@@]2(C)CCC(=O)NC[C@@H](C)OP(O)(=O)O[C@H]3[C@H]([C@H](O[C@@H]3CO)N3C4=CC(C)=C(C)C=C4N=C3)O)\C2=C(C)/C([C@H](C\2(C)C)CCC(N)=O)=N/C/2=C\C([C@H]([C@@]/2(CC(N)=O)C)CCC(N)=O)=N\C\2=C(C)/C2=N[C@]1(C)[C@@](C)(CC(N)=O)[C@@H]2CCC(N)=O AGVAZMGAQJOSFJ-WZHZPDAFSA-M 0.000 claims description 3
- BTCSSZJGUNDROE-UHFFFAOYSA-N gamma-aminobutyric acid Chemical group NCCCC(O)=O BTCSSZJGUNDROE-UHFFFAOYSA-N 0.000 claims description 3
- 238000012502 risk assessment Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000012797 qualification Methods 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001131 transforming effect 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
- G06Q40/08—Insurance
-
- 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
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
-
- 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
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The embodiment of the invention discloses an underwriting method for a logistics insurance policy and aims to solve the problems that due to the fact that existing logistics insurance policies must be manually underwritten by the underwriting personnel in the prior art, the underwriting accuracy rate is influenced by the professional levels and the working experiences of the underwriting personnel and the underwriting efficiency is difficult to increase. According to the embodiments of the invention, the method comprises the steps of acquiring a target logistics insurance policy and the information of a target offerer; according to the information of the risk subject matter of the target logistics insurance policy and the information of an insurance applicant, calculating the first underwriting factor of the risk subject matter; according to a preset vector model, extracting vectors out of the information of the risk subject matter and the information of the target offerer so as to obtain the second underwriting factor of the risk subject matter; inputting the first underwriting factor and the second underwriting factor of the risk subject matter into a pre-trained deep learning model; acquiring the output of the deep learning model and adopting the output as the underwriting result of the target logistics insurance policy. The embodiment of the invention further provides an underwriting apparatus for a logistics insurance policy.
Description
Technical field
The present invention relates to financial services industry, more particularly to a kind of core guarantor's method and apparatus of logistics insurance declaration form.
Background technology
As logistic industry is more and more flourishing, the various risks that logistic industry faces also extraordinarily are being amplified.In order to reduce thing
The tremendous influence that stream risk is brought to enterprise, increasing enterprise insures to related logistics business.Therefore, for guarantor
For dangerous company, how efficiently the risk of this kind of logistics business of accurate evaluation just seems of crucial importance.
At present, insurance company typically carries out risk assessment and report by the underwriter of specialty to the declaration form of logistics insurance
Valency, the core that underwriter draws protects result and will directly affect the subjectinsured quotation of declaration form and the air control of insurance company management.
However, core guarantor's accuracy rate of logistics insurance declaration form is often close with the professional standards of underwriter and working experience
Correlation, different underwriters may show that different cores protect result to same part logistics insurance declaration form so that logistics insurance declaration form
Core protect accuracy rate be subject to extreme influence；In addition, artificial risk assessment is carried out by underwriter, in the premise for ensureing accuracy rate
Under be often difficult to improve core protect efficiency, in the face of substantial amounts of logistics insurance declaration form core protect task when, considerably increase core guaranteed cost.
The content of the invention
The core for embodiments providing a kind of logistics insurance declaration form protects method and apparatus, it is possible to increase logistics insurance is protected
Single core protects efficiency and accuracy rate.
A kind of core of logistics insurance declaration form provided in an embodiment of the present invention protects method, including：
Obtain target logistics insurance policy and target offers people's information；
According to the first of the subjectinsured information of the target logistics insurance policy and insurer's information calculation risk target
Core protects the factor；
Enter row vector to the subjectinsured information and the target offers people information according to default vector model to extract,
Obtain the second subjectinsured core and protect the factor；
First core is protected the factor and second core protects the factor as input input to the good depth of training in advance
Practise model；
The output for obtaining the deep learning model protects result as the core of the target logistics insurance policy.
Alternatively, the deep learning model is obtained by following steps training in advance：
Obtain the logistics insurance declaration form and offerer's information corresponding with the logistics insurance declaration form as sample；
According to the subjectinsured information and the first sample of insurer's information calculation risk target of the logistics insurance declaration form
The factor；
The subjectinsured information and offerer's information of the logistics insurance declaration form are entered according to default vector model
Row vector is extracted, and obtains the second subjectinsured sample factor；
Using the first sample factor and the second sample factor as input input to the deep learning model, obtain
To the output of the deep learning model；
Using the output for obtaining as target, the hidden layer parameter of the deep learning model is adjusted, obtained with minimizing
The output and the logistics insurance declaration form core protect result between error；
If the error meets pre-conditioned, it is determined that the current deep learning model is the deep learning for training
Model.
Alternatively, the subjectinsured information includes rising for subjectinsured type, means of transportation, manner of packing and transport
Point endpoint information；
The subjectinsured information and insurer's information calculation risk target according to the target logistics insurance policy
First core is protected the factor and is specifically included：
The corresponding target type benchmark loss rate of the subjectinsured type is inquired about as the first low layer Dynamic gene；
It is determined that the second low layer Dynamic gene corresponding with insurer's information；
It is determined that the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation；
It is determined that the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing；
It is determined that the 5th low layer Dynamic gene corresponding with the starting and terminal point information of the subjectinsured transport；
To the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer Dynamic gene, the adjustment of the 4th low layer
The factor and the 5th low layer Dynamic gene carry out pretreatment, obtain the first subjectinsured core and protect the factor.
Alternatively, the subjectinsured information includes subjectinsured insurance coverage, protection amount, history loss ratio, market ring
Environment information and commercial competition information, the target offers people information includes insurance company's information of target offers people；
It is described that row vector is entered to the subjectinsured information and the target offers people information according to default vector model
Extract, obtain subjectinsured the second core guarantor's factor and specifically include：
It is determined that the first numerical value corresponding with the subjectinsured insurance coverage；
It is determined that with the subjectinsured insured amount corresponding second value；
It is determined that third value corresponding with the subjectinsured history loss ratio；
It is determined that the 4th numerical value corresponding with the subjectinsured market environment information；
It is determined that the 5th numerical value corresponding with the subjectinsured commercial competition information；
It is determined that the 6th numerical value corresponding with insurance company's information of the target offers people；
First numerical value, second value, third value, the 4th numerical value, the 5th numerical value and the 6th numerical value are imported default
Vector model, obtain subjectinsured second core and protect the factor.
Alternatively, the output for obtaining the deep learning model protects knot as the core of the target logistics insurance policy
Fruit specifically includes：
Whether the output for judging the deep learning model meets default threshold condition；
If the output of the deep learning model is unsatisfactory for default threshold condition, by the target logistics insurance policy
Corresponding input and output feed back to the deep learning model as negative sample, to correct the hidden layer of the deep learning model
Parameter；
If the output of the deep learning model meets default threshold condition, it is determined that the deep learning model it is defeated
Go out the core guarantor's result for the target logistics insurance policy, and by the corresponding input of the target logistics insurance policy and export true
It is set to the newly-increased positive sample for training the deep learning model.
A kind of core protection device of logistics insurance declaration form provided in an embodiment of the present invention, including：
Data obtaining module, for obtaining target logistics insurance policy and target offers people's information；
First core protects factor computing module, for according to the subjectinsured information of the target logistics insurance policy and insuring
First core of people's information calculation risk target protects the factor；
Second core protect factor acquisition module, for according to default vector model to the subjectinsured information and the mesh
Mark offerer's information enters row vector extraction, obtains the second subjectinsured core and protects the factor；
Input module, the factor is protected as input input to instruction in advance for first core to be protected the factor and second core
The deep learning model perfected；
Core protects result acquisition module, and the output for obtaining the deep learning model is protected as the target logistics insurance
Single core protects result.
Alternatively, the deep learning model with lower module training in advance by being obtained：
Sample acquisition module, for obtaining as the logistics insurance declaration form of sample and corresponding with the logistics insurance declaration form
Offerer's information；
First sample factor computing module, for being believed according to the subjectinsured information of the logistics insurance declaration form and insurer
The first sample factor of breath calculation risk target；
Second sample factor acquisition module, for the risk mark according to default vector model to the logistics insurance declaration form
Information and offerer's information enter row vector extraction, obtain the second subjectinsured sample factor；
Learning model training module, for the first sample factor and the second sample factor to be put into as input
To the deep learning model, the output of the deep learning model is obtained；
Hidden layer parameter adjustment module, as target, the deep learning model is adjusted for using the output for obtaining
Hidden layer parameter, with the error that the core for minimizing the output and logistics insurance declaration form for obtaining is protected between result；
Learning model confirms module, if meeting pre-conditioned for the error, it is determined that the current deep learning
Model is the deep learning model for training.
Alternatively, the subjectinsured information includes rising for subjectinsured type, means of transportation, manner of packing and transport
Point endpoint information；
First core is protected factor computing module and is specifically included：
Factor I unit, for inquiring about the corresponding target type benchmark loss rate of the subjectinsured type as
One low layer Dynamic gene；
Factor Ⅱ unit, for determining the second low layer Dynamic gene corresponding with insurer's information；
Factor III unit, for determining the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation；
CA++ unit, for determining the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing；
Accelerator factor unit, for determining the 5th low layer corresponding with the starting and terminal point information of the subjectinsured transport
Dynamic gene；
Factor pretreatment unit, for adjusting to the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer
Integral divisor, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene carry out pretreatment, obtain the first subjectinsured core
Protect the factor.
Alternatively, the subjectinsured information includes subjectinsured insurance coverage, protection amount, history loss ratio, market ring
Environment information and commercial competition information, the target offers people information includes insurance company's information of target offers people；
Second core is protected factor acquisition module and is specifically included：
First numerical value unit, for determining the first numerical value corresponding with the subjectinsured insurance coverage；
Second value unit, for determining and the subjectinsured insured amount corresponding second value；
Third value unit, for determining third value corresponding with the subjectinsured history loss ratio；
4th numerical value unit, for determining the 4th numerical value corresponding with the subjectinsured market environment information；
5th numerical value unit, for determining the 5th numerical value corresponding with the subjectinsured commercial competition information；
6th numerical value unit, for determining the 6th numerical value corresponding with insurance company's information of the target offers people；
Vector model import unit, for by first numerical value, second value, third value, the 4th numerical value, the 5th to count
Value and the 6th numerical value import default vector model, obtain the second subjectinsured core and protect the factor.
Alternatively, the core is protected result acquisition module and is specifically included：
Output judging unit, for judging whether the output of the deep learning model meets default threshold condition；
Negative sample feedback unit, if the output for the deep learning model is unsatisfactory for default threshold condition, will
The corresponding input of the target logistics insurance policy and output feed back to the deep learning model as negative sample, to correct
State the hidden layer parameter of deep learning model；
Positive sample feedback unit, if the output for the deep learning model meets default threshold condition, by institute
State the corresponding input of target logistics insurance policy and output feeds back to the deep learning model as positive sample, it is described to instruct
The hidden layer parameter learning of deep learning model；
Core protects result determining unit, if the output for the deep learning model meets default threshold condition, really
The fixed deep learning model is output as the core of the target logistics insurance policy and protects result.
As can be seen from the above technical solutions, the embodiment of the present invention has advantages below：
In the embodiment of the present invention, first, target logistics insurance policy and target offers people's information are obtained；Then, according to
The subjectinsured information of the target logistics insurance policy and the first core of insurer's information calculation risk target protect the factor；According to
Default vector model enters row vector extraction to the subjectinsured information and the target offers people information, obtains the risk
Second core of target protects the factor；Then, first core is protected the factor and second core and protects the factor as input input to pre-
The deep learning model for first training；Finally, the output for obtaining the deep learning model is protected as the target logistics insurance
Single core protects result.In embodiments of the present invention, according to the relevant information and target offers people's information of target logistics insurance policy
It is calculated the first core and protects the factor and the second core guarantor's factor, and using the two factors as the good deep learning model of training in advance
Input, so as to the output of deep learning model that will be obtained protects result as core, using deep learning model thing is automatically obtained
The core of stream insurance policy is protected, it is to avoid core protects the problem that result is affected by the professional standards and working experience of underwriter, carries
The core of high logistics insurance declaration form protects accuracy rate；Also, core is greatly improved on the premise of accuracy rate is ensured and protects efficiency, in face
When protecting task to substantial amounts of logistics insurance declaration form core, can also be rapidly completed core and protect task, reduce core guaranteed cost.
Description of the drawings
Fig. 1 is that a kind of core of logistics insurance declaration form in the embodiment of the present invention protects method one embodiment flow chart；
Fig. 2 is the step of a kind of core of logistics insurance declaration form in Fig. 1 correspondence embodiments protects method 102 in an application scenarios
Under schematic flow sheet；
Fig. 3 is the step of a kind of core of logistics insurance declaration form in Fig. 1 correspondence embodiments protects method 103 in an application scenarios
Under schematic flow sheet；
Fig. 4 is the step of a kind of core of logistics insurance declaration form in Fig. 1 correspondence embodiments protects method 104 in an application scenarios
Under schematic flow sheet；
Fig. 5 is a kind of core protection device one embodiment structure chart of logistics insurance declaration form in the embodiment of the present invention.
Specific embodiment
The core for embodiments providing a kind of logistics insurance declaration form protects method and apparatus, protects for solving existing logistics
Dangerous declaration form carries out artificial nucleus guarantor by underwriter, and core protects the professional standards and working experience that accuracy rate is subject to underwriter
Affect and be difficult to improve the problem that core protects efficiency.
To enable goal of the invention, feature, the advantage of the present invention more obvious and understandable, below in conjunction with the present invention
Accompanying drawing in embodiment, is clearly and completely described, it is clear that disclosed below to the technical scheme in the embodiment of the present invention
Embodiment be only a part of embodiment of the invention, and not all embodiment.Based on the embodiment in the present invention, this area
All other embodiment that those of ordinary skill is obtained under the premise of creative work is not made, belongs to protection of the present invention
Scope.
Fig. 1 is referred to, a kind of core of logistics insurance declaration form is protected method one embodiment and included in the embodiment of the present invention：
101st, target logistics insurance policy and target offers people's information are obtained；
In the present embodiment, it is possible, firstly, to obtain target logistics insurance policy and target offers people's information.
It is understood that above-mentioned target logistics insurance policy includes the subjectinsured information of logistics insurance, insurer's letter
The information related to declaration form such as breath, also, can also be obtained from data base by the number of policy of the target logistics insurance policy
To the other information related to the target logistics insurance policy.
Offerer, refers generally to the insurance company that certain logistics insurance policy is offered, accepted insurance.Above-mentioned target offers people
Information can include that insurance company's information, the preference of core guarantor and the business of quotation is related to the preference matching degree of core guarantor etc.
Information.
102nd, according to the subjectinsured information of the target logistics insurance policy and insurer's information calculation risk target
First core protects the factor；
After target logistics insurance policy is got, can be according to the subjectinsured letter of the target logistics insurance policy
First core of breath and insurer's information calculation risk target protects the factor.
Further, the subjectinsured information can include subjectinsured type, means of transportation, manner of packing and fortune
Defeated starting and terminal point information.Fig. 2 shows the step of a kind of core of logistics insurance declaration form in Fig. 1 correspondence embodiments protects method 102
Schematic flow sheet under an application scenarios, above-mentioned steps 102 specifically can include：
201st, inquire about the corresponding target type benchmark loss rate of the subjectinsured type as the first low layer adjust because
Son；
202nd, the second low layer Dynamic gene corresponding with insurer's information is determined；
203rd, the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation is determined；
204th, the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing is determined；
205th, the 5th low layer Dynamic gene corresponding with the starting and terminal point information of the subjectinsured transport is determined；
206th, to the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer Dynamic gene, the 4th low layer
Dynamic gene and the 5th low layer Dynamic gene carry out pretreatment, obtain the first subjectinsured core and protect the factor.
For above-mentioned steps 201, the target type benchmark loss rate is that statistics is obtained in insurance system or insurance industry
It is some type of subjectinsured accept insurance after unit insured amount insurance indemnity rate, it is subjectinsured for certain that it is insurance company
Logistics insurance declaration form carries out the important parameter offered with risk assessment.The corresponding target type benchmark loss of the subjectinsured type
Rate can pass through the statistical table of inquiry benchmark loss rate and obtain.It should be noted that for the insurance company that difference is accepted insurance,
The corresponding target type benchmark loss rate of same subjectinsured type may be differed, or even subjectinsured classification also not phase
Together, specifically can be set according to practical situations.
For above-mentioned steps 202, insurer's information can include the qualification and type of insurer.The insurance of quotation is public
Department can be directed to the insurer of different qualifications or type and be divided into several class, insurer's correspondence of each class it is default one it is low
Layer Dynamic gene.
For above-mentioned steps 203, it is to be understood that the subjectinsured means of transportation of logistics insurance declaration form is to declaration form
Risk assessment has extremely important effect, and different means of transportation are to there is different degrees of risk.The insurance company of quotation
Several class can be divided into for subjectinsured several modes, the means of transportation correspondence of each class it is default one it is low
Layer Dynamic gene.
For above-mentioned steps 204, it is to be understood that the subjectinsured manner of packing of logistics insurance declaration form is to declaration form
Risk assessment has extremely important effect, and different manner of packings are to there is different degrees of risk.The insurance company of quotation
Several class can be divided into for subjectinsured Different Package mode, the manner of packing correspondence of each class it is default one it is low
Layer Dynamic gene.
For above-mentioned steps 205, it is to be understood that the subjectinsured starting and terminal point information of logistics insurance declaration form is to protecting
Single risk assessment has extremely important effect, and different starting and terminal point information are to there is different degrees of risk.For example, rise
The distance between point terminal is nearer, then risk is relatively lower, and the distance between starting and terminal point is more remote, then risk is relatively higher.
The insurance company of quotation can be divided into several class, the starting point of each class for subjectinsured different starting and terminal point information
Endpoint information correspondence presets a low layer Dynamic gene.
It should be noted that not limiting sequencing between above-mentioned steps 201～205, for example, step 202 can be in step
Perform before either step in rapid 201,203～205, it is also possible to perform behind either step in step 201,203～205,
Even can simultaneously perform with either step in step 201,203～205.
For above-mentioned steps 206, the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer tune are being determined
After integral divisor, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene, can to the first low layer Dynamic gene, second
Low layer Dynamic gene, the 3rd low layer Dynamic gene, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene carry out pretreatment, obtain
The factor is protected to the first subjectinsured core.It is understood that specifically, X=ψ (a1, a2, a3, a4, a5) can be adopted
Pretreatment is carried out, wherein, a1～a5 represents respectively the first low layer Dynamic gene, the second low layer Dynamic gene, the adjustment of the 3rd low layer
The factor, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene, and ψ is then transforming function transformation function.
In the present embodiment, protect the factor to represent the risk mark of the target logistics insurance policy using the first core obtained above
Benchmark loss rate, by using the first core protect the factor as deep learning model input, using the deep learning for training
The internal relation that the subjectinsured benchmark loss rate implied in model and core are protected between result, so as to by deep learning model
Output protects result as close to accurate core.
103rd, row vector is entered to the subjectinsured information and the target offers people information according to default vector model
Extract, obtain the second subjectinsured core and protect the factor；
After target logistics insurance policy and target offers people's information is got, can be according to default vector model pair
The subjectinsured information and the target offers people information enter row vector extraction, obtain subjectinsured second core protect because
Son.
Further, the subjectinsured information can include subjectinsured insurance coverage, protection amount, history loss ratio,
Market environment information and commercial competition information, the target offers people information includes insurance company's information of target offers people.Fig. 3
Show 103 streams under an application scenarios of the step of a kind of core of logistics insurance declaration form in Fig. 1 correspondence embodiments protects method
Journey schematic diagram, as shown in figure 3, above-mentioned steps 103 specifically can include：
301st, the first numerical value corresponding with the subjectinsured insurance coverage is determined；
302nd, determine and the subjectinsured insured amount corresponding second value；
303rd, third value corresponding with the subjectinsured history loss ratio is determined；
304th, the 4th numerical value corresponding with the subjectinsured market environment information is determined；
305th, the 5th numerical value corresponding with the subjectinsured commercial competition information is determined；
306th, the 6th numerical value corresponding with insurance company's information of the target offers people is determined；
307th, by first numerical value, second value, third value, the 4th numerical value, the 5th numerical value and the 6th numerical value are imported
Default vector model, obtains the second subjectinsured core and protects the factor.
For above-mentioned steps 301, the insurance coverage can include principal clause, Additional Terms, the limit of logistics insurance declaration form
The information with regard to declaration form insurance coverage such as volume.For the logistics insurance declaration form of different insurance kinds, insurance company can set in advance
The different insurance coverage of fixed different the first numerical value correspondence, the first numerical value is bigger, then it represents that insurance coverage is wider.
For above-mentioned steps 302, can be directly using subjectinsured insured amount numerical value as second value, it is also possible to by
Insurance company presets the different protection amount interval of different second value correspondences, when subjectinsured protection amount falls into certain protection amount
When interval, then the insured amount corresponding second value is can determine that.
For above-mentioned steps 303, it is to be understood that for insurance company, that what is accepted insurance subjectinsured is respectively provided with
Corresponding history compensates information and history loss ratio.Subjectinsured history loss ratio can directly be determined for third value, example
If subjectinsured history loss ratio is 30%, then corresponding third value is 0.3；Can also be preset not by insurance company
The different loss ratio of same third value correspondence is interval, when subjectinsured history loss ratio (can be average) falls into certain compensation
When the rate of paying is interval, then the corresponding third value of history loss ratio is can determine that.
For above-mentioned steps 304, for it is same it is subjectinsured for, different market environments are equally to subjectinsured report
Valency and risk assessment are impacted.Above-mentioned market environment information can be the market environment for certain class is subjectinsured
Scoring.Therefore, it can in advance different market environment information be investigated, arranged and analyzed, so as to for different market
Environmental information presets corresponding the 4th different numerical value.
For above-mentioned steps 305, in the same manner, for it is same it is subjectinsured for, different commercial competitions are equally to risk mark
Quotation and risk assessment impact.Above-mentioned commercial competition information can be for certain class is subjectinsured, to business
The scoring of the severity of industry competition.Therefore, it can in advance different commercial competition information be investigated, arranged and analyzed,
So as to preset corresponding the 5th different numerical value for different commercial competition information.
For above-mentioned steps 306, it is to be understood that for the insurance company residing for different offerers, can correspond to
The 6th different numerical value are set used as " scoring " of insurance company's information, in general, the product of the 6th numerical value and insurance company
Board strength is related, can be that the brand strength of insurance company is stronger specifically, then corresponding 6th numerical value is bigger, and insures public
The brand strength of department is weaker, then corresponding 6th numerical value is less.
It should be noted that not limiting sequencing between above-mentioned steps 301～306, for example, step 302 can be in step
Perform before either step in rapid 301,303～306, it is also possible to perform behind either step in step 301,303～306,
Even can simultaneously perform with either step in step 301,303～306.
For above-mentioned steps 307, default vector model can specifically be expressed as Y=(b1, b2, b3, b4, b5, b6), its
In, b1～b6 represents respectively first numerical value, second value, third value, the 4th numerical value, the 5th numerical value and the 6th numerical value.
Further, by first numerical value, second value, third value, the 4th numerical value, the 5th numerical value and the 6th numerical value import pre-
If vector model, the vector for obtaining can also be standardized, obtain described subjectinsured after standardization
Second core protects the factor.
The 104th, first core protected the factor and second core protects the factor as input input to the good depth of training in advance
Degree learning model；
Protect the factor and after second core protects the factor first core is obtained, can first core protect the factor with
Second core protects the factor as input input to the good deep learning model of training in advance.
Further, as shown in figure 4, the deep learning model can be obtained by following steps training in advance：
401st, the logistics insurance declaration form and offerer's information corresponding with the logistics insurance declaration form as sample is obtained；
402nd, according to the first of the subjectinsured information of the logistics insurance declaration form and insurer's information calculation risk target
The sample factor；
403rd, the subjectinsured information and offerer letter according to default vector model to the logistics insurance declaration form
Cease into row vector and extract, obtain the second subjectinsured sample factor；
404th, using the first sample factor and the second sample factor as input input to the deep learning mould
Type, obtains the output of the deep learning model；
405th, the output for obtaining is adjusted into the hidden layer parameter of the deep learning model as target, to minimize
The error that the output for obtaining and the core of the logistics insurance declaration form are protected between result；
If the 406, the error meets pre-conditioned, it is determined that the current deep learning model is the depth for training
Learning model.
For above-mentioned steps 401, it is necessary first to obtain as sample logistics insurance declaration form and with the logistics insurance
The corresponding offerer's information of declaration form.In the present embodiment, above-mentioned sample can be positive sample and/or negative sample, in general, right
When deep learning model is trained, can be trained using a large amount of good samples of labelling in advance.And follow-up positive sample and negative
Sample can be used for the deep learning model to training and be verified and corrected.
For above-mentioned steps 402, get as sample logistics insurance declaration form and with the logistics insurance declaration form
After corresponding offerer's information, wind can be calculated according to the subjectinsured information of the logistics insurance declaration form and insurer's information
The first sample factor of dangerous target.Wherein the circular of the first sample factor is similar with above-mentioned steps 102, specifically may be used
So that with reference to the concrete calculating process that the factor is protected with regard to the core of step 102 first, here is omitted.
For above-mentioned steps 403, get as sample logistics insurance declaration form and with the logistics insurance declaration form
After corresponding offerer's information, can according to default vector model to the subjectinsured information of the logistics insurance declaration form and
Offerer's information enters row vector extraction, obtains the second subjectinsured sample factor.The wherein tool of the second sample factor
Body computational methods are similar with above-mentioned steps 103, specifically may be referred to protect specifically calculating for the factor with regard to the core of step 103 second
Journey, here is omitted.
For above-mentioned steps 404, output and the first sample factor and the second sample as sample of the deep learning model
The input of this factor is corresponding.Therefore, in general, in the training process of deep learning model, the deep learning model
Output protects result and there is more or less discrepancy, namely error with the actual core of the logistics insurance declaration form as sample.
For above-mentioned steps 405, the hidden layer parameter for constantly adjusting the deep learning model can be passed through so that described defeated
Go out the error minimize protected between result with the core of the logistics insurance declaration form, successively exercise supervision study to deep learning model
And training.It is understood that the deep learning model includes N number of hidden layer, wherein N >=1, the concrete numerical value of N can basis
Practical situation sets.When being trained to deep learning model, the parameter for adjusting this N number of hidden layer can be passed through, be realized to working as
The value of the output of front deep learning model is adjusted, and it is contrasted during adjustment and protects result with the core of logistics insurance declaration form
Between error, as cause error minimum as possible.Error is less, then it represents that current deep learning model training effect is better,
Conversely, then training effect is poorer.
For above-mentioned steps 406, it is to be understood that above-mentioned steps 401～405 can be repeatedly performed, using substantial amounts of
Sample is trained to deep learning model.After training, when the error meets pre-conditioned, then it is considered that current depth
Learning model has completed training.This is pre-conditioned for example can be：It is trained using M sample, wherein K sample correspondence
Output and the error protected between result of core be less than 10%, and K/M is more than or equal to 50%, namely the error for meeting certain condition
The ratio of corresponding sample exceedes the threshold value of setting, then it is considered that error meets pre-conditioned.
105th, the output for obtaining the deep learning model protects result as the core of the target logistics insurance policy.
First core is being protected the factor and second core guarantor's factor as input input to the good depth of training in advance
After learning model, the output of the deep learning model can be got, then be protected the output as the target logistics
The core of dangerous declaration form protects result.
Further, above-mentioned steps 105 specifically can include：Whether the output for judging the deep learning model meets pre-
If threshold condition, if it is not, then using the target logistics insurance policy it is corresponding input and output feed back to institute as negative sample
Deep learning model is stated, to correct the hidden layer parameter of the deep learning model, if, it is determined that the deep learning model
The core for being output as the target logistics insurance policy protects result, and by the corresponding input of the target logistics insurance policy and output
It is defined as the newly-increased positive sample for training the deep learning model.
It is understood that for the output obtained using deep learning model, can protect to the target logistics in advance
The core of dangerous declaration form is protected result and is estimated, for example, can protect result to the core of the logistics insurance declaration form of a certain type according to core guarantor
Data statisticss are carried out, the core for obtaining the logistics insurance declaration form of this type protects result average, then protects result average in the core
On the basis of set the threshold range for fluctuating up and down, such as ± 10%, you can form the threshold value of the logistics insurance declaration form of this type
Condition.When the output of deep learning model is unsatisfactory for the threshold condition, then it is assumed that the output protects result error with normal core
It is larger, then the corresponding input of the target logistics insurance policy and output are fed back to into deep learning model as negative sample
In, to be modified to the hidden layer parameter of deep learning model.
In addition, when the output of the deep learning model meets default threshold condition, then not only can determine described
Deep learning model is output as the core of the target logistics insurance policy and protects result, and can protect the target logistics insurance
Single corresponding input and output are defined as the newly-increased positive sample for training the deep learning model, increase deep learning mould
Type is used for the positive sample quantity of training, improves the training completeness of deep learning model.
In the present embodiment, first, target logistics insurance policy and target offers people's information are obtained；Then, according to described
The subjectinsured information of target logistics insurance policy and the first core of insurer's information calculation risk target protect the factor；According to default
Vector model the subjectinsured information and the target offers people information are entered row vector extraction, obtain described subjectinsured
The second core protect the factor；Then, first core is protected the factor and second core protects the factor as input input to instruction in advance
The deep learning model perfected；Finally, the output of the deep learning model is obtained as the target logistics insurance policy
Core protects result.In the present embodiment, it is calculated according to the relevant information and target offers people's information of target logistics insurance policy
First core protects the factor and the second core and protects the factor, and using the two factors as the good deep learning model of training in advance input,
So as to the output of the deep learning model that will be obtained protects result as core, it is automatically obtained logistics insurance using deep learning model and protects
Single core is protected, it is to avoid core protects the problem that result is affected by the professional standards and working experience of underwriter, improves logistics
The core of insurance policy protects accuracy rate；Also, core is greatly improved on the premise of accuracy rate is ensured and protects efficiency, in the face of substantial amounts of
When logistics insurance declaration form core protects task, can also be rapidly completed core and protect task, reduce core guaranteed cost.
The core for essentially describing a kind of logistics insurance declaration form above protects method, below by a kind of core of logistics insurance declaration form
Protection device is described in detail.
Fig. 5 shows a kind of core protection device one embodiment structure chart of logistics insurance declaration form in the embodiment of the present invention.
In the present embodiment, a kind of core protection device of logistics insurance declaration form includes：
Data obtaining module 501, for obtaining target logistics insurance policy and target offers people's information；
First core protects factor computing module 502, for according to the subjectinsured information of the target logistics insurance policy and
First core of insurer's information calculation risk target protects the factor；
Second core protects factor acquisition module 503, for according to default vector model to the subjectinsured information and institute
State target offers people's information and enter row vector extraction, obtain the second subjectinsured core and protect the factor；
Input module 504, the factor is protected as input input to pre- for first core to be protected the factor and second core
The deep learning model for first training；
Core protects result acquisition module 505, and the output for obtaining the deep learning model is protected as the target logistics
The core of dangerous declaration form protects result.
Further, the deep learning model can be by being obtained with lower module training in advance：
Sample acquisition module, for obtaining as the logistics insurance declaration form of sample and corresponding with the logistics insurance declaration form
Offerer's information；
First sample factor computing module, for being believed according to the subjectinsured information of the logistics insurance declaration form and insurer
The first sample factor of breath calculation risk target；
Second sample factor acquisition module, for the risk mark according to default vector model to the logistics insurance declaration form
Information and offerer's information enter row vector extraction, obtain the second subjectinsured sample factor；
Learning model training module, for the first sample factor and the second sample factor to be put into as input
To the deep learning model, the output of the deep learning model is obtained；
Hidden layer parameter adjustment module, as target, the deep learning model is adjusted for using the output for obtaining
Hidden layer parameter, with the error that the core for minimizing the output and logistics insurance declaration form for obtaining is protected between result；
Learning model confirms module, if meeting pre-conditioned for the error, it is determined that the current deep learning
Model is the deep learning model for training.
Further, the subjectinsured information can include subjectinsured type, means of transportation, manner of packing and fortune
Defeated starting and terminal point information；
First core protects factor computing module specifically can be included：
Factor I unit, for inquiring about the corresponding target type benchmark loss rate of the subjectinsured type as
One low layer Dynamic gene；
Factor Ⅱ unit, for determining the second low layer Dynamic gene corresponding with insurer's information；
Factor III unit, for determining the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation；
CA++ unit, for determining the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing；
Accelerator factor unit, for determining the 5th low layer corresponding with the starting and terminal point information of the subjectinsured transport
Dynamic gene；
Factor pretreatment unit, for adjusting to the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer
Integral divisor, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene carry out pretreatment, obtain the first subjectinsured core
Protect the factor.
Further, the subjectinsured information can include subjectinsured insurance coverage, protection amount, history loss ratio,
Market environment information and commercial competition information, the target offers people information can include that the insurance company of target offers people believes
Breath；
Second core is protected factor acquisition module and is specifically included：
First numerical value unit, for determining the first numerical value corresponding with the subjectinsured insurance coverage；
Second value unit, for determining and the subjectinsured insured amount corresponding second value；
Third value unit, for determining third value corresponding with the subjectinsured history loss ratio；
4th numerical value unit, for determining the 4th numerical value corresponding with the subjectinsured market environment information；
5th numerical value unit, for determining the 5th numerical value corresponding with the subjectinsured commercial competition information；
6th numerical value unit, for determining the 6th numerical value corresponding with insurance company's information of the target offers people；
Vector model import unit, for by first numerical value, second value, third value, the 4th numerical value, the 5th to count
Value and the 6th numerical value import default vector model, obtain the second subjectinsured core and protect the factor.
Further, the core is protected result acquisition module and specifically can be included：
Output judging unit, for judging whether the output of the deep learning model meets default threshold condition；
Negative sample feedback unit, if the output for the deep learning model is unsatisfactory for default threshold condition, will
The corresponding input of the target logistics insurance policy and output feed back to the deep learning model as negative sample, to correct
State the hidden layer parameter of deep learning model；
Positive sample feedback unit, if the output for the deep learning model meets default threshold condition, by institute
State the corresponding input of target logistics insurance policy and output feeds back to the deep learning model as positive sample, it is described to instruct
The hidden layer parameter learning of deep learning model；
Core protects result determining unit, if the output for the deep learning model meets default threshold condition, really
The fixed deep learning model is output as the core of the target logistics insurance policy and protects result.
Those skilled in the art can be understood that, for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, may be referred to the corresponding process in preceding method embodiment, will not be described here.
In several embodiments provided herein, it should be understood that disclosed system, apparatus and method can be with
Realize by another way.For example, device embodiment described above is only schematic, for example, the unit
Divide, only a kind of division of logic function can have other dividing mode, such as multiple units or component when actually realizing
Can with reference to or be desirably integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or
The coupling each other for discussing or direct-coupling or communication connection can be the indirect couplings by some interfaces, device or unit
Close or communicate to connect, can be electrical, mechanical or other forms.
The unit as separating component explanation can be or may not be it is physically separate, it is aobvious as unit
The part for showing can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
On NE.Some or all of unit therein can according to the actual needs be selected to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.Above-mentioned integrated list
Unit both can be realized in the form of hardware, it would however also be possible to employ the form of SFU software functional unit is realized.
If the integrated unit is realized using in the form of SFU software functional unit and as independent production marketing or used
When, during a computer read/write memory medium can be stored in.Based on such understanding, technical scheme is substantially
The part for contributing to prior art in other words or all or part of the technical scheme can be in the form of software products
Embody, the computer software product is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server, or network equipment etc.) performs the complete of each embodiment methods described of the invention
Portion or part steps.And aforesaid storage medium includes：USB flash disk, portable hard drive, read only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can store journey
The medium of sequence code.
The above, above example only to illustrate technical scheme, rather than a limitation；Although with reference to front
State embodiment to be described in detail the present invention, it will be understood by those within the art that：It still can be to front
State the technical scheme described in each embodiment to modify, or equivalent is carried out to which part technical characteristic；And these
Modification is replaced, and does not make the spirit and scope of the essence disengaging various embodiments of the present invention technical scheme of appropriate technical solution.
Claims (10)
1. a kind of core of logistics insurance declaration form protects method, it is characterised in that include：
Obtain target logistics insurance policy and target offers people's information；
Protected according to the subjectinsured information of the target logistics insurance policy and the first core of insurer's information calculation risk target
The factor；
Enter row vector to the subjectinsured information and the target offers people information according to default vector model to extract, obtain
The second subjectinsured core protects the factor；
First core is protected the factor and second core protects the factor as input input to the good deep learning mould of training in advance
Type；
The output for obtaining the deep learning model protects result as the core of the target logistics insurance policy.
2. logistics insurance declaration form according to claim 1 core protect method, it is characterised in that the deep learning model by
Following steps training in advance is obtained：
Obtain the logistics insurance declaration form and offerer's information corresponding with the logistics insurance declaration form as sample；
According to the subjectinsured information and the first sample factor of insurer's information calculation risk target of the logistics insurance declaration form；
According to default vector model the subjectinsured information and offerer's information of the logistics insurance declaration form are carried out to
Amount is extracted, and obtains the second subjectinsured sample factor；
Using the first sample factor and the second sample factor as input input to the deep learning model, institute is obtained
State the output of deep learning model；
Using the output for obtaining as target, the hidden layer parameter of the deep learning model is adjusted, to minimize the institute for obtaining
State the error between output and core guarantor's result of the logistics insurance declaration form；
If the error meets pre-conditioned, it is determined that the current deep learning model is the deep learning mould for training
Type.
3. the core of logistics insurance declaration form according to claim 1 protects method, it is characterised in that the subjectinsured packet
Include the starting and terminal point information of subjectinsured type, means of transportation, manner of packing and transport；
The first of the subjectinsured information and insurer's information calculation risk target according to the target logistics insurance policy
Core is protected the factor and is specifically included：
The corresponding target type benchmark loss rate of the subjectinsured type is inquired about as the first low layer Dynamic gene；
It is determined that the second low layer Dynamic gene corresponding with insurer's information；
It is determined that the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation；
It is determined that the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing；
It is determined that the 5th low layer Dynamic gene corresponding with the starting and terminal point information of the subjectinsured transport；
To the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer Dynamic gene, the 4th low layer Dynamic gene
Pretreatment is carried out with the 5th low layer Dynamic gene, the first subjectinsured core is obtained and is protected the factor.
4. the core of logistics insurance declaration form according to claim 1 protects method, it is characterised in that the subjectinsured packet
Include subjectinsured insurance coverage, protection amount, history loss ratio, market environment information and commercial competition information, the target offers
People's information includes insurance company's information of target offers people；
It is described to enter row vector extraction to the subjectinsured information and the target offers people information according to default vector model,
Obtain subjectinsured the second core guarantor's factor to specifically include：
It is determined that the first numerical value corresponding with the subjectinsured insurance coverage；
It is determined that with the subjectinsured insured amount corresponding second value；
It is determined that third value corresponding with the subjectinsured history loss ratio；
It is determined that the 4th numerical value corresponding with the subjectinsured market environment information；
It is determined that the 5th numerical value corresponding with the subjectinsured commercial competition information；
It is determined that the 6th numerical value corresponding with insurance company's information of the target offers people；
By first numerical value, second value, third value, the 4th numerical value, the 5th numerical value and the 6th numerical value import it is default to
Amount model, obtains the second subjectinsured core and protects the factor.
5. the core of logistics insurance declaration form according to any one of claim 1 to 4 protects method, it is characterised in that described to obtain
The output for taking the deep learning model is specifically included as core guarantor's result of the target logistics insurance policy：
Whether the output for judging the deep learning model meets default threshold condition；
If the output of the deep learning model is unsatisfactory for default threshold condition, by target logistics insurance policy correspondence
Input and output feed back to the deep learning model as negative sample, with correct the deep learning model hidden layer join
Number；
If the output of the deep learning model meets default threshold condition, it is determined that the deep learning model is output as
The core of the target logistics insurance policy protects result, and the corresponding input of the target logistics insurance policy and output are defined as
For training the newly-increased positive sample of the deep learning model.
6. a kind of core protection device of logistics insurance declaration form, it is characterised in that include：
Data obtaining module, for obtaining target logistics insurance policy and target offers people's information；
First core protects factor computing module, for being believed according to the subjectinsured information of the target logistics insurance policy and insurer
First core of breath calculation risk target protects the factor；
Second core protect factor acquisition module, for according to default vector model to the subjectinsured information and the target report
Valency people's information enters row vector extraction, obtains the second subjectinsured core and protects the factor；
Input module is good to training in advance as input input for first core to be protected the factor and second core guarantor's factor
Deep learning model；
Core protects result acquisition module, for obtaining the output of the deep learning model as the target logistics insurance policy
Core protects result.
7. the core protection device of logistics insurance declaration form according to claim 6, it is characterised in that the deep learning model by
Obtained with lower module training in advance：
Sample acquisition module, for obtaining logistics insurance declaration form and report corresponding with the logistics insurance declaration form as sample
Valency people's information；
First sample factor computing module, based on subjectinsured information and insurer's information according to the logistics insurance declaration form
The subjectinsured first sample factor；
Second sample factor acquisition module, for the subjectinsured letter according to default vector model to the logistics insurance declaration form
Breath and offerer's information enter row vector extraction, obtain the second subjectinsured sample factor；
Learning model training module, for the first sample factor and the second sample factor to be put into institute as input
Deep learning model is stated, the output of the deep learning model is obtained；
Hidden layer parameter adjustment module, as target, the hidden layer of the deep learning model is adjusted for using the output for obtaining
Parameter, with the error that the core for minimizing the output and logistics insurance declaration form for obtaining is protected between result；
Learning model confirms module, if meeting pre-conditioned for the error, it is determined that the current deep learning model
For the deep learning model for training.
8. the core protection device of logistics insurance declaration form according to claim 6, it is characterised in that the subjectinsured packet
Include the starting and terminal point information of subjectinsured type, means of transportation, manner of packing and transport；
First core is protected factor computing module and is specifically included：
Factor I unit is low as first for inquiring about the corresponding target type benchmark loss rate of the subjectinsured type
Layer Dynamic gene；
Factor Ⅱ unit, for determining the second low layer Dynamic gene corresponding with insurer's information；
Factor III unit, for determining the 3rd low layer Dynamic gene corresponding with the subjectinsured means of transportation；
CA++ unit, for determining the 4th low layer Dynamic gene corresponding with the subjectinsured manner of packing；
Accelerator factor unit, for determining the 5th low layer adjustment corresponding with the starting and terminal point information of the subjectinsured transport
The factor；
Factor pretreatment unit, for the first low layer Dynamic gene, the second low layer Dynamic gene, the 3rd low layer adjustment because
Son, the 4th low layer Dynamic gene and the 5th low layer Dynamic gene carry out pretreatment, obtain the first subjectinsured core protect because
Son.
9. the core protection device of logistics insurance declaration form according to claim 6, it is characterised in that the subjectinsured packet
Include subjectinsured insurance coverage, protection amount, history loss ratio, market environment information and commercial competition information, the target offers
People's information includes insurance company's information of target offers people；
Second core is protected factor acquisition module and is specifically included：
First numerical value unit, for determining the first numerical value corresponding with the subjectinsured insurance coverage；
Second value unit, for determining and the subjectinsured insured amount corresponding second value；
Third value unit, for determining third value corresponding with the subjectinsured history loss ratio；
4th numerical value unit, for determining the 4th numerical value corresponding with the subjectinsured market environment information；
5th numerical value unit, for determining the 5th numerical value corresponding with the subjectinsured commercial competition information；
6th numerical value unit, for determining the 6th numerical value corresponding with insurance company's information of the target offers people；
Vector model import unit, for by first numerical value, second value, third value, the 4th numerical value, the 5th numerical value and
6th numerical value imports default vector model, obtains the second subjectinsured core and protects the factor.
10. the core protection device of the logistics insurance declaration form according to any one of claim 6 to 9, it is characterised in that the core
Protect result acquisition module to specifically include：
Output judging unit, for judging whether the output of the deep learning model meets default threshold condition；
Negative sample feedback unit, if the output for the deep learning model is unsatisfactory for default threshold condition, will be described
The corresponding input of target logistics insurance policy and output feed back to the deep learning model as negative sample, to correct the depth
The hidden layer parameter of degree learning model；
Positive sample feedback unit, if the output for the deep learning model meets default threshold condition, by the mesh
The corresponding input of mark logistics insurance declaration form and output feed back to the deep learning model as positive sample, to instruct the depth
The hidden layer parameter learning of learning model；
Core protects result determining unit, if the output for the deep learning model meets default threshold condition, it is determined that institute
State core guarantor's result that deep learning model is output as the target logistics insurance policy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610988595.5A CN106651588A (en) | 2016-11-09 | 2016-11-09 | Underwriting method and apparatus for logistics insurance policy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610988595.5A CN106651588A (en) | 2016-11-09 | 2016-11-09 | Underwriting method and apparatus for logistics insurance policy |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106651588A true CN106651588A (en) | 2017-05-10 |
Family
ID=58806853
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610988595.5A Pending CN106651588A (en) | 2016-11-09 | 2016-11-09 | Underwriting method and apparatus for logistics insurance policy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106651588A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108241730A (en) * | 2017-10-12 | 2018-07-03 | 平安科技（深圳）有限公司 | Core protects test method, application server and computer readable storage medium |
| WO2019169768A1 (en) * | 2018-03-06 | 2019-09-12 | 平安科技（深圳）有限公司 | Centralized insurance policy auditing method, electronic device, and readable storage medium |
| CN110874702A (en) * | 2018-09-04 | 2020-03-10 | 菜鸟智能物流控股有限公司 | Model training method and device in logistics sorting scene and electronic equipment |
| CN110880149A (en) * | 2019-11-29 | 2020-03-13 | 上海商汤智能科技有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN112418745A (en) * | 2019-08-22 | 2021-02-26 | 天津五八到家科技有限公司 | Policy generation method, device, equipment and storage medium |
| CN118037461A (en) * | 2024-04-11 | 2024-05-14 | 中国平安财产保险股份有限公司四川分公司 | Property insurance information verification method, device, equipment and storage medium |
-
2016
- 2016-11-09 CN CN201610988595.5A patent/CN106651588A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108241730A (en) * | 2017-10-12 | 2018-07-03 | 平安科技（深圳）有限公司 | Core protects test method, application server and computer readable storage medium |
| WO2019071956A1 (en) * | 2017-10-12 | 2019-04-18 | 平安科技（深圳）有限公司 | Underwriting testing method, application server and computer readable storage medium |
| WO2019169768A1 (en) * | 2018-03-06 | 2019-09-12 | 平安科技（深圳）有限公司 | Centralized insurance policy auditing method, electronic device, and readable storage medium |
| CN110874702A (en) * | 2018-09-04 | 2020-03-10 | 菜鸟智能物流控股有限公司 | Model training method and device in logistics sorting scene and electronic equipment |
| CN110874702B (en) * | 2018-09-04 | 2023-05-23 | 菜鸟智能物流控股有限公司 | Model training method and device under logistics sorting scene and electronic equipment |
| CN112418745A (en) * | 2019-08-22 | 2021-02-26 | 天津五八到家科技有限公司 | Policy generation method, device, equipment and storage medium |
| CN110880149A (en) * | 2019-11-29 | 2020-03-13 | 上海商汤智能科技有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN118037461A (en) * | 2024-04-11 | 2024-05-14 | 中国平安财产保险股份有限公司四川分公司 | Property insurance information verification method, device, equipment and storage medium |
| CN118037461B (en) * | 2024-04-11 | 2024-06-18 | 中国平安财产保险股份有限公司四川分公司 | Property insurance information verification method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600417A (en) | Underwriting method and device of property insurance policies | |
| CN106651588A (en) | Underwriting method and apparatus for logistics insurance policy | |
| CN106600419A (en) | Method and device for underwriting on public liability insurance policy | |
| CN107067157A (en) | Business risk appraisal procedure, device and air control system | |
| KR20060013543A (en) | Method and system for predicting attrition customers | |
| CN106469276A (en) | The kind identification method of data sample and device | |
| CN110310119A (en) | Extendible capacity storage method towards the transaction of transregional piece of chain | |
| CN110430008B (en) | Civil aviation flight telegraph checking system and method based on cyclic redundancy check algorithm | |
| CN113379530A (en) | User risk determination method and device and server | |
| CN113902037A (en) | Abnormal bank account identification method, system, electronic device and storage medium | |
| US11669806B2 (en) | Retirement score calculator | |
| CN109410070A (en) | Core protects data processing method and system | |
| CN107908480A (en) | Wages computational methods, application server and computer-readable recording medium | |
| US20220414125A1 (en) | Systems and Methods for Computer Modeling Using Incomplete Data | |
| CN116416073A (en) | Risk control method, system, device and storage medium | |
| CN107678844A (en) | Breakpoint intelligence shunt method, electronic installation and computer-readable storage medium | |
| CN107066411A (en) | Data transmission method, device and computer-readable recording medium | |
| Hirano et al. | Comparing actual and simulated HFT traders' behavior for agent design | |
| CN106384280A (en) | Artificial intelligent accounting system and computing method thereof | |
| CN108492169A (en) | Risk Modeling method and system based on credit card approval scene are realized | |
| CN114638501A (en) | Business data processing method and device, computer equipment and storage medium | |
| CN117557290A (en) | E-business operation and maintenance management method and system based on machine learning | |
| CN107862576A (en) | A kind of method and platform of the more shop management of cooperation businessman | |
| CN115293867A (en) | Financial reimbursement user portrait optimization method, device, equipment and storage medium | |
| WO2021227303A1 (en) | Unmanned aerial vehicle signal suppression device scheduling method and apparatus, electronic device, and 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170510 |