CN116523661A - Claim settlement method, device, equipment and storage medium based on artificial intelligence - Google Patents
Claim settlement method, device, equipment and storage medium based on artificial intelligence Download PDFInfo
- Publication number
- CN116523661A CN116523661A CN202310492664.3A CN202310492664A CN116523661A CN 116523661 A CN116523661 A CN 116523661A CN 202310492664 A CN202310492664 A CN 202310492664A CN 116523661 A CN116523661 A CN 116523661A
- Authority
- CN
- China
- Prior art keywords
- settlement
- policy
- knowledge graph
- neural network
- deep neural
- 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 49
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 83
- 238000012549 training Methods 0.000 claims abstract description 27
- 230000002787 reinforcement Effects 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 22
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 230000009471 action Effects 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000009193 crawling Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/092—Reinforcement learning
-
- 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
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Educational Administration (AREA)
- Evolutionary Computation (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Finance (AREA)
- Health & Medical Sciences (AREA)
- Accounting & Taxation (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- Animal Behavior & Ethology (AREA)
- Technology Law (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The application relates to the field of artificial intelligence and financial insurance, and provides an artificial intelligence-based claim settlement method, device, equipment and storage medium, wherein the method comprises the following steps: acquiring a knowledge graph of the claim; generating a plurality of sample data according to the claim knowledge graph; based on an DQN algorithm, performing reinforcement learning training on a preset first deep neural network by using a plurality of sample data to obtain a claim settlement decision model; and acquiring an enterprise knowledge graph corresponding to the to-be-claiming policy, inputting the enterprise knowledge graph into the claim settlement decision model for decision processing, and obtaining a target claim settlement policy of the to-be-claiming policy. The method improves the efficiency and accuracy of the policy claims. The present application also relates to the field of blockchain, and the storage medium may store data created from the use of blockchain nodes.
Description
Technical Field
The present disclosure relates to the field of artificial intelligence, and in particular, to an artificial intelligence-based claim settlement method, apparatus, device, and storage medium.
Background
Currently, enterprise knowledge maps include information such as various attributes, risk factors, and historical data of an enterprise, and various attributes, risk factors, and historical data of an associated enterprise. In the field of financial insurance, insurance companies may evaluate risk and decision risk management policies for an enterprise based on an enterprise knowledge graph. For example, in the case that an enterprise applies for claims to a policy, a claims settlement person can determine a claims settlement policy through the policy corresponding to the enterprise knowledge graph, but the enterprise knowledge graph is complex, the claims settlement person needs to spend more time to make a decision, the claims settlement efficiency is low, and the claims settlement person has difficulty in finding the invisible risk of the enterprise through the enterprise knowledge graph, so that the claims settlement is inaccurate and losses are brought to the enterprise. Therefore, how to improve the efficiency and accuracy of policy claims is a current urgent problem to be solved.
Disclosure of Invention
The embodiment of the application provides an artificial intelligence-based claim settlement method, device, equipment and storage medium, aiming at improving the efficiency and accuracy of policy claim settlement.
In a first aspect, embodiments of the present application provide an artificial intelligence based claim settlement method, including:
acquiring a claim settlement knowledge graph, wherein the claim settlement knowledge graph comprises attribute information, relation information and claim settlement results of a plurality of entities involved in a plurality of claim settlement policy;
generating a plurality of sample data according to the claim knowledge graph, wherein the sample data comprises attribute information, relation information and claim settlement results of a plurality of entities related to the claim settlement policy;
based on an DQN algorithm, performing reinforcement learning training on a preset first deep neural network by using the plurality of sample data to obtain a claim settlement decision model;
and acquiring an enterprise knowledge graph corresponding to the to-be-claiming policy, and inputting the enterprise knowledge graph into the claim settlement decision model to perform decision processing so as to obtain a target claim settlement policy of the to-be-claiming policy.
In a second aspect, embodiments of the present application further provide an artificial intelligence based claim settlement apparatus, the claim settlement apparatus including:
the system comprises an acquisition module, a calculation module and a calculation module, wherein the acquisition module is used for acquiring a claim settlement knowledge graph, wherein the claim settlement knowledge graph comprises attribute information, relation information and claim settlement results of a plurality of entities involved in a plurality of claim settlement plan insurance policies;
The sample generation module is used for generating a plurality of sample data according to the claim knowledge graph, wherein the sample data comprises attribute information, relation information and claim result of a plurality of entities related to the claim settlement policy;
the model training module is used for performing reinforcement learning training on a preset first deep neural network by using the plurality of sample data based on an DQN algorithm to obtain a claim settlement decision model;
the acquisition module is also used for acquiring an enterprise knowledge graph corresponding to the claim policy to be settled;
and the claim settlement decision module is used for inputting the enterprise knowledge graph into the claim settlement decision model for decision processing to obtain the target claim settlement strategy of the claim policy to be settled.
In a third aspect, embodiments of the present application further provide a computer device, the computer device including a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein the computer program when executed by the processor implements the artificial intelligence based claim method of the first aspect.
In a fourth aspect, embodiments of the present application further provide a storage medium having a computer program stored thereon, where the computer program, when executed by a processor, implements the artificial intelligence based claim settlement method according to the first aspect.
The embodiment of the application provides an artificial intelligence-based claim settlement method, device, equipment and storage medium, the claim settlement method generates a plurality of sample data based on a claim settlement knowledge graph, each sample data comprises attribute information, relation information and claim settlement results of a plurality of entities related to a claim settlement policy, then the method uses the generated plurality of sample data to carry out reinforcement learning training on a deep neural network based on a DQN algorithm (an algorithm based on the deep neural network and reinforcement learning), and a claim settlement decision model is obtained, and as the claim settlement decision model has learned how to adopt optimal claim settlement strategies under different conditions, the corresponding enterprise knowledge graph can be input into the claim settlement decision model for decision processing under the condition that the claim settlement policy needs to be carried out, so that the target claim settlement policy of the claim settlement policy is obtained, and the claim settlement policy is determined without the need of the claim settlement staff to observe the enterprise knowledge graph, thereby greatly improving the efficiency and accuracy of the claim settlement policy.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an artificial intelligence based claim settlement method according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of sub-steps of the artificial intelligence based claim settlement method of FIG. 1;
FIG. 3 is a schematic block diagram of an artificial intelligence based claim settlement device provided in an embodiment of the application;
FIG. 4 is a schematic block diagram of a sub-module of the artificial intelligence based claim settlement apparatus of FIG. 3;
fig. 5 is a schematic block diagram of a computer device according to an embodiment of the present application.
The realization, functional characteristics and advantages of the present application will be further described with reference to the embodiments, referring to the attached drawings.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
The flow diagrams depicted in the figures are merely illustrative and not necessarily all of the elements and operations/steps are included or performed in the order described. For example, some operations/steps may be further divided, combined, or partially combined, so that the order of actual execution may be changed according to actual situations.
Currently, enterprise knowledge maps include information such as various attributes, risk factors, and historical data of an enterprise, and various attributes, risk factors, and historical data of an associated enterprise. In the field of financial insurance, insurance companies may evaluate risk and decision risk management policies for an enterprise based on an enterprise knowledge graph. For example, in the case that an enterprise applies for claims to a policy, a claims settlement person can determine a claims settlement policy through the policy corresponding to the enterprise knowledge graph, but the enterprise knowledge graph is complex, the claims settlement person needs to spend more time to make a decision, the claims settlement efficiency is low, and the claims settlement person has difficulty in finding the invisible risk of the enterprise through the enterprise knowledge graph, so that the claims settlement is inaccurate and losses are brought to the enterprise. Therefore, how to improve the efficiency and accuracy of policy claims is a current urgent problem to be solved.
In order to solve the above problems, the embodiments of the present application provide a method, an apparatus, a computer device and a storage medium for claim settlement based on artificial intelligence, where the method for claim settlement generates a plurality of sample data based on a claim settlement knowledge graph, and each sample data includes attribute information, relationship information and claim settlement results of a plurality of entities related to a claim settlement policy, and then uses the generated plurality of sample data to perform reinforcement learning training on a deep neural network based on DQN algorithm (an algorithm based on deep neural network and reinforcement learning), so as to obtain a claim settlement decision model.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results. Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions. The computer device may be a server or a terminal device, and the server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, a content delivery network (Content Delivery Network, CDN), and basic cloud computing services such as big data and an artificial intelligence platform.
Some embodiments of the present application are described in detail below with reference to the accompanying drawings. The following embodiments and features of the embodiments may be combined with each other without conflict.
Referring to fig. 1, fig. 1 is a schematic flow chart of an artificial intelligence-based claim settlement method according to an embodiment of the present application.
As shown in fig. 1, the claim settlement method includes steps S101 to S104.
And S101, acquiring a knowledge graph of the claim settlement.
In this embodiment, the claim knowledge graph is established according to a large number of enterprise claim settlement data related to the claim settlement policy, and the claim knowledge graph includes attribute information, relationship information and claim settlement results of a plurality of entities related to the claim settlement policy. The method comprises the steps of setting up a policy for settling a claim, wherein the policy for settling the claim is a policy for settling the claim, the data for settling the claim comprise policy information, enterprise information, a policy for settling the claim and a result for settling the claim, the policy for settling the claim comprises a policy number, a balance, an applicant, a insured person, a beneficiary and the like, the enterprise information comprises names of the enterprises, attribute information and names and attribute information of related enterprises, the attribute information comprises registered addresses, registered time, registered capital, stakeholders, industries, scale, blacklist enterprises and the like of the enterprises, the policy for settling the claim can comprise agreeing to settling the claim or refusing to settle the claim, and the result for settling the claim can comprise success of settling the claim or failure of settling the claim and the like.
In some embodiments, before step S101, further comprising: crawling enterprise claim settlement data corresponding to the claim settlement policy from a claim settlement server by using a preset crawler program, and storing the crawled enterprise claim settlement data in a database; and when the total number of the enterprise claim data stored in the database is greater than or equal to the set number, building a claim settlement knowledge graph according to all the enterprise claim data in the database, and storing the claim settlement knowledge graph. The preset crawler program may be written by a developer based on actual situations, which is not specifically limited in the embodiment of the present application. By crawling a sufficient amount of enterprise claim settlement data to establish a claim settlement knowledge graph, the comprehensiveness of the claim settlement knowledge graph can be improved.
And S102, generating a plurality of sample data according to the claim knowledge graph.
In this embodiment, the sample data includes attribute information, relationship information, and a result of claims of a plurality of entities involved in one claim settlement policy. The relationship information may include a relationship between a plurality of entities involved in a claimant policy and a relationship between a plurality of attributes of each entity.
In some embodiments, as shown in fig. 2, step S102 includes: substep S1021 to substep S1024.
Substep S1021, randomly determining a target entity in the claim knowledge-graph.
In this embodiment, the target entity may be any entity in the claim knowledge graph, and the target entity may be an enterprise or a person, which is not specifically limited in this embodiment of the present application.
In some embodiments, the manner of randomly determining the target entity in the claim knowledge-graph may be: generating a random number in a preset range, and determining the entity with the same identification code as the random number in the claim knowledge graph as a target entity. Each entity in the claim knowledge graph is allocated with an identification code, the identification code is used for uniquely identifying the entity in the claim knowledge graph, the maximum numerical value in the preset range is smaller than or equal to the maximum identification code in the claim knowledge graph, the preset range can be set based on actual conditions, and the embodiment of the application is not limited specifically. The random determination of the target entity from the claim knowledge graph can be realized through the random number and the identification code of the entity.
In some embodiments, the manner of randomly determining the target entity in the claim knowledge-graph may be: randomly selecting a preset policy number from a preset policy number library as a target policy number, wherein the preset policy number library comprises policy numbers corresponding to each claim settlement policy related to the claim knowledge graph; and acquiring an identification code group associated with the target policy number, and determining any entity of which the identification code in the claim knowledge graph is in the identification code group as a target entity. The identification code group associated with the target policy number comprises identification codes of a plurality of entities related to the claim settlement policy corresponding to the target policy number.
Sub-step S1022, obtaining attribute information, relationship information and claim result of a plurality of entities corresponding to the claim settlement policy associated with the target entity from the claim knowledge graph as one sample data.
For example, if the entity with the identifier 20 is the target entity, the claim settlement policy associated with the entity with the identifier 20 is the claim settlement policy a, and the plurality of entities corresponding to the claim settlement policy a include the entity with the identifier 20, the entity with the identifier 22, the entity with the identifier 26, the entity with the identifier 27, and the entity with the identifier 30, the attribute information and the relationship information of the entity with the identifier 20, the attribute information and the relationship information of the entity with the identifier 22, the attribute information and the relationship information of the entity with the identifier 26, the attribute information and the relationship information of the entity with the identifier 27, and the claim result corresponding to the claim settlement policy a may be obtained from the claim knowledge graph.
Substep S1023, counting the number of sample data.
In this embodiment, if the number of sample data is smaller than the preset number, the step S1021 is executed again, that is, the target entity is randomly determined in the claim knowledge graph, and if the number of sample data reaches the preset number, the step S1024 is executed to stop generating sample data. The preset number is smaller than or equal to the total number of the claims settlement policy used for building the claims knowledge graph, and can be set based on actual conditions, which is not particularly limited in the embodiment of the present application. For example, the preset number is 5000.
And step S103, based on the DQN algorithm, performing reinforcement learning training on a preset first deep neural network by using a plurality of sample data to obtain a claim settlement decision model.
In this embodiment, the DQN algorithm, i.e. Deep Q-network, refers to a Q-learning algorithm based on Deep learning, which maintains a Q-table, uses a table to store rewards obtained by taking action a in each state s, i.e. state-cost function Q (s, a), and the DQN algorithm replaces the table with a Deep neural network, and the rest is the same.
In some embodiments, based on the DQN algorithm, the reinforcement learning training is performed on the preset first deep neural network using a plurality of sample data, and the manner of obtaining the claim decision model may be: selecting a learning sample from a plurality of sample data, wherein the learning sample comprises a state and a claim settlement result, and the state is attribute information and relation information of a plurality of entities; inputting the state into a first deep neural network for claim settlement decision processing to obtain the action value (Q value) of each claim settlement decision, and determining a predicted claim settlement strategy according to the action value of each claim settlement decision; executing the predicted claim settlement strategy in the claim settlement knowledge graph to obtain a predicted claim settlement result, and determining target rewards according to the predicted claim settlement result and the claim settlement result in the learning sample; determining a target value according to the target reward, a next learning sample of the plurality of sample data and a preset second deep neural network, wherein the first deep neural network and the second deep neural network have the same structure; determining a loss value of the first deep neural network according to the target value and the action value of the predicted claim settlement result; when the loss value is larger than a preset loss value, updating parameters of the first deep neural network according to the target value and the action value of the predicted claim settlement result, and continuing to perform reinforcement learning training on the first deep neural network by using a next learning sample until the loss value of the first deep neural network is smaller than or equal to the preset loss value.
In some embodiments, the manner in which the target prize is determined based on the predicted claim result and the claim result in the learning sample may be: under the condition that the predicted claim settlement result is the same as the claim settlement result in the learning sample, taking the preset positive rewards as target rewards; and taking the preset negative rewards as target rewards in the case that the predicted claim settlement results are different from the claim settlement results in the learning sample. The preset positive rewards and negative rewards may be set by the user based on actual situations, which are not specifically limited in the embodiment of the present application.
In some embodiments, the determining the target value based on the target reward, a next learning sample of the plurality of sample data, and the preset second deep neural network may be: inputting the state in the next learning sample into a second deep neural network for decision processing to obtain the action value of each claim settlement decision; and determining a target value according to the target rewards and the maximum action value. The target rewards and the maximum action values can be accumulated, the accumulated result is determined to be a target value, the product of the target rewards and the preset coefficients can be calculated to obtain gain values, the gain values and the maximum action values are accumulated, and the accumulated result is determined to be the target value.
For example, a learning sample C1 is selected and denoted as (s C1 ，p C1 )，s C1 For learning the shape in sample C1State, P C1 For the result of claim settlement in the learning sample C1, the next learning sample C2 is selected and recorded as (s C2 ，p C2 )，s C2 For the state in the next learning sample C2, p C2 For the result of the claim in the next learning sample C2, the state s is calculated C1 Inputting a first deep neural network (evaluation network) to obtain Q value of each claim settlement decision (action), and selecting claim settlement decision a with maximum Q value by using greedy strategy C1 Executing the claim decision a in a claim knowledge graph (environment) C1 Obtaining a predicted result p of claim settlement, and obtaining the result p of claim settlement according to the predicted result p of claim settlement and the learning sample C1 C1 Determining a target prize r C1 State s C2 Inputting the second deep neural network for processing to obtain Q value of each claim settlement decision, and obtaining the maximum Q value and target rewards r C1 And determining a target value, calculating a loss value of the first deep neural network according to the target value and the Q value of the predicted claim settlement result p, and updating parameters of the first deep neural network when the loss value is larger than a preset loss value so that the Q value of the predicted claim settlement result p is as close to the target value as possible, and continuing reinforcement learning training on the first deep neural network by using a next learning sample C2.
In some embodiments, during reinforcement learning training of the first deep neural network, the state in the learning sample, the state in the next learning sample, the target rewards, and the predictive claim settlement strategy are stored as one experience to an experience pool; sampling a plurality of experiences from the experience pool at preset interval time, and updating parameters of the first deep neural network according to the plurality of experiences. For example, the four-tuple (s C1 ，a C1 ，r C1 ，s C2 ) As an experience is stored in an experience pool s C1 To learn the state in sample C1, s C2 R for the state in the next learning sample C2 C1 Awarding for the goal, a C1 Is state s C1 Corresponding claim settlement decision a C1 . By updating the parameters of the first deep neural network in an empirical playback mode, the variance of parameter updating can be reduced, and the convergence rate can be improved.
In some embodiments, determining a number of updates of the first deep neural network during reinforcement learning training of the first deep neural network; and when the updating times are integer multiples of a preset value, updating the parameters of the second deep neural network to the current parameters of the first deep neural network. The preset value may be set based on actual situations, which is not specifically limited in the embodiment of the present application. For example, the preset value is 10, that is, every 10 updates of the first deep neural network, the parameters of the second deep neural network are updated to the current parameters of the first deep neural network. In the learning process, only the parameters of the first deep neural network are updated in real time, the parameters of the second deep neural network are updated in a timing manner, and the second deep neural network is used for calculating the target value, so that the target value reported in a period of time when the second deep neural network is unchanged is ensured to be relatively fixed, and the stability of reinforcement learning is improved.
And step S104, acquiring an enterprise knowledge graph corresponding to the to-be-claiming policy, inputting the enterprise knowledge graph into the claim settlement decision model for decision processing, and obtaining the target claim settlement policy of the to-be-claiming policy.
In this embodiment, the policy to be claiming is a policy to be claiming, the enterprise knowledge graph includes attribute information and relationship information of a plurality of entities, the attribute information includes registration address, registration time, registration capital, stakeholders, industries to which the enterprise belongs, scale, whether the enterprise is a blacklist enterprise, and the relationship information includes relationships between the plurality of entities and relationships between the plurality of attributes of each entity. The enterprise knowledge graph corresponding to the claim policy to be settled is a knowledge graph established in advance according to enterprise information.
In some embodiments, the manner of inputting the enterprise knowledge graph into the claim settlement decision model to perform decision processing to obtain the target claim settlement policy of the claim policy to be settled may be: inputting the enterprise knowledge graph into a claim settlement decision model for decision processing to obtain the output probability of each claim settlement strategy; and determining the claim settlement strategy corresponding to the maximum output probability as the target claim settlement strategy of the claim policy to be settled.
According to the claim settlement method provided by the embodiment, the plurality of sample data are generated based on the claim settlement knowledge graph, each sample data comprises attribute information, relation information and claim settlement results of a plurality of entities related to a claim settlement policy, then the deep neural network is subjected to reinforcement learning training based on the DQN algorithm (an algorithm based on the deep neural network and reinforcement learning), so that a claim settlement decision model is obtained, and as the claim settlement decision model has learned how to adopt the optimal claim settlement policy under different conditions, the corresponding enterprise knowledge graph can be input into the claim settlement decision model for decision processing under the condition that the claim settlement policy is required to be carried out on the claim settlement policy, so that the target claim policy of the claim settlement policy is obtained, and the claim settlement policy is determined without observing the enterprise knowledge graph by the claim settlement staff, thereby greatly improving the efficiency and accuracy of the claim settlement policy.
Referring to fig. 3, fig. 3 is a schematic block diagram of an artificial intelligence-based claim settlement device according to an embodiment of the present application.
As shown in fig. 3, the artificial intelligence based claim settlement apparatus 200 includes:
an obtaining module 210, configured to obtain a claim knowledge graph, where the claim knowledge graph includes attribute information, relationship information, and claim results of a plurality of entities related to a plurality of claim settlement policy;
the sample generation module 220 is configured to generate a plurality of sample data according to the claim knowledge graph, where the sample data includes attribute information, relationship information and claim result of a plurality of entities related to the claim settlement policy;
the model training module 230 is configured to perform reinforcement learning training on a preset first deep neural network by using the plurality of sample data based on an DQN algorithm, so as to obtain a claim settlement decision model;
the obtaining module 210 is further configured to obtain an enterprise knowledge graph corresponding to the claim policy to be resolved;
and the claim settlement decision module 240 is configured to input the enterprise knowledge graph into the claim settlement decision model for decision processing, so as to obtain the target claim settlement policy of the claim policy to be settled.
In some embodiments, as shown in fig. 4, the sample generation module 220 includes:
A determining submodule 221, configured to randomly determine a target entity in the claim knowledge graph;
an obtaining sub-module 222, configured to obtain attribute information, relationship information and claim result of a plurality of entities corresponding to the claim settlement policy associated with the target entity from the claim knowledge graph as one sample data;
the statistics sub-module 223 is configured to count the number of the sample data, determine, by the determining sub-module 221, a target entity in the claim knowledge graph if the number is less than a preset number, and stop generating the sample data if the number reaches the preset number.
In some embodiments, the determining submodule 221 is further configured to:
generating a random number in a preset range, and determining an entity with the same identification code in the claim knowledge graph as the random number as a target entity.
In some embodiments, the determining submodule 221 is further configured to:
randomly selecting a preset policy number from a preset policy number library as a target policy number, wherein the preset policy number library comprises policy numbers corresponding to each claim settlement policy related to the claim knowledge graph;
and acquiring an identification code group associated with the target insurance policy number, and determining any entity of the identification code in the identification code group in the claim knowledge graph as a target entity.
In some embodiments, the model training module 230 is further configured to:
selecting a learning sample from the plurality of sample data, wherein the learning sample comprises a state and a claim settlement result, and the state is attribute information and relation information of a plurality of entities;
inputting the state into the first deep neural network to perform claim settlement decision processing to obtain the action value of each claim settlement decision, and determining a predicted claim settlement strategy according to the action value of each claim settlement decision;
executing the predicted claim settlement strategy in the claim settlement knowledge graph to obtain a predicted claim settlement result, and determining a target reward according to the predicted claim settlement result and the claim settlement result;
determining a target value according to the target reward, a next learning sample of the sample data and a preset second deep neural network, wherein the first deep neural network and the second deep neural network have the same structure;
determining a loss value of the first deep neural network according to the target value and the action value of the predicted claim result;
when the loss value is larger than a preset loss value, updating parameters of the first deep neural network according to the target value and the action value of the predicted claim result;
And continuing reinforcement learning training on the first deep neural network by using the next learning sample until the loss value of the first deep neural network is smaller than or equal to a preset loss value.
In some embodiments, the model training module 230 is further configured to:
storing the state in the learning sample, the state in the next learning sample, the target prize, and the predictive claim settlement strategy as one experience to an experience pool;
sampling a plurality of experiences from the experience pool at preset interval time, and updating parameters of the first deep neural network according to the plurality of experiences.
In some embodiments, the model training module 230 is further configured to:
determining the update times of the first deep neural network;
and when the updating times are integral multiples of a preset value, updating the parameters of the second deep neural network to the current parameters of the first deep neural network.
It should be noted that, for convenience and brevity of description, specific working processes of the above-described apparatus and modules and units may refer to corresponding processes in the foregoing embodiment of the claim settlement method based on artificial intelligence, which are not described herein again.
The apparatus provided by the above embodiments may be implemented in the form of a computer program which may be run on a computer device as shown in fig. 5.
Referring to fig. 5, fig. 5 is a schematic block diagram of a computer device according to an embodiment of the present application. The computer device may be a server or a terminal.
As shown in fig. 5, the computer device includes a processor, a memory, and a network interface connected by a system bus, wherein the memory may include a storage medium and an internal memory.
The storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any of a variety of artificial intelligence based claims.
The processor is used to provide computing and control capabilities to support the operation of the entire computer device.
The network interface is used for network communication such as transmitting assigned tasks and the like. It will be appreciated by those skilled in the art that the structure shown in fig. 5 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.
It should be appreciated that the processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), field-programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. Wherein the general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
Wherein in an embodiment the processor is configured to run a computer program stored in the memory to implement the steps of:
acquiring a claim settlement knowledge graph, wherein the claim settlement knowledge graph comprises attribute information, relation information and claim settlement results of a plurality of entities involved in a plurality of claim settlement policy;
generating a plurality of sample data according to the claim knowledge graph, wherein the sample data comprises attribute information, relation information and claim settlement results of a plurality of entities related to the claim settlement policy;
based on an DQN algorithm, performing reinforcement learning training on a preset first deep neural network by using the plurality of sample data to obtain a claim settlement decision model;
And acquiring an enterprise knowledge graph corresponding to the to-be-claiming policy, and inputting the enterprise knowledge graph into the claim settlement decision model to perform decision processing so as to obtain a target claim settlement policy of the to-be-claiming policy.
In an embodiment, the processor, when implementing generating a plurality of sample data from the claim knowledge-graph, is configured to implement:
randomly determining a target entity in the claim knowledge graph;
acquiring attribute information, relation information and claim settlement results of a plurality of entities corresponding to the claim settlement policy associated with the target entity from the claim settlement knowledge graph as one sample of data;
and counting the number of the sample data, if the number is smaller than the preset number, returning to the step of determining the target entity in the claim knowledge graph, and if the number reaches the preset number, stopping generating the sample data.
In an embodiment, the processor, when implementing randomly determining a target entity in the claim knowledge-graph, is configured to implement:
generating a random number in a preset range, and determining an entity with the same identification code in the claim knowledge graph as the random number as a target entity.
In an embodiment, the processor, when implementing randomly determining a target entity in the claim knowledge-graph, is configured to implement:
Randomly selecting a preset policy number from a preset policy number library as a target policy number, wherein the preset policy number library comprises policy numbers corresponding to each claim settlement policy related to the claim knowledge graph;
and acquiring an identification code group associated with the target insurance policy number, and determining any entity of the identification code in the identification code group in the claim knowledge graph as a target entity.
In an embodiment, when implementing the DQN algorithm and performing reinforcement learning training on the preset deep neural network by using the plurality of sample data, the processor is configured to implement:
selecting a learning sample from the plurality of sample data, wherein the learning sample comprises a state and a claim settlement result, and the state is attribute information and relation information of a plurality of entities;
inputting the state into the first deep neural network to perform claim settlement decision processing to obtain the action value of each claim settlement decision, and determining a predicted claim settlement strategy according to the action value of each claim settlement decision;
executing the predicted claim settlement strategy in the claim settlement knowledge graph to obtain a predicted claim settlement result, and determining a target reward according to the predicted claim settlement result and the claim settlement result;
Determining a target value according to the target reward, a next learning sample of the sample data and a preset second deep neural network, wherein the first deep neural network and the second deep neural network have the same structure;
determining a loss value of the first deep neural network according to the target value and the action value of the predicted claim result;
when the loss value is larger than a preset loss value, updating parameters of the first deep neural network according to the target value and the action value of the predicted claim result;
and continuing reinforcement learning training on the first deep neural network by using the next learning sample until the loss value of the first deep neural network is smaller than or equal to a preset loss value.
In an embodiment, the processor is further configured to implement the steps of:
storing the state in the learning sample, the state in the next learning sample, the target prize, and the predictive claim settlement strategy as one experience to an experience pool;
sampling a plurality of experiences from the experience pool at preset interval time, and updating parameters of the first deep neural network according to the plurality of experiences.
In an embodiment, the processor is further configured to implement the steps of:
determining the update times of the first deep neural network;
and when the updating times are integral multiples of a preset value, updating the parameters of the second deep neural network to the current parameters of the first deep neural network.
It should be noted that, for convenience and brevity of description, specific working processes of the computer device described above may refer to corresponding processes in the foregoing embodiment of the artificial intelligence-based claim settlement method, which are not described herein again.
From the above description of embodiments, it will be apparent to those skilled in the art that the present application may be implemented in software plus a necessary general purpose hardware platform. Based on such understanding, the technical solutions of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to perform the methods described in the embodiments or some parts of the embodiments of the present application.
Embodiments of the present application also provide a storage medium having a computer program stored thereon, where the computer program includes program instructions, and a method implemented when the program instructions are executed may refer to various embodiments of an artificial intelligence based claim settlement method of the present application.
The storage medium may be volatile or nonvolatile. The storage medium may be an internal storage unit of the computer device according to the foregoing embodiment, for example, a hard disk or a memory of the computer device. The storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the computer device.
Further, the storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created from the use of blockchain nodes, and the like.
The blockchain referred to in the application is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The Blockchain (Blockchain), which is essentially a decentralised database, is a string of data blocks that are generated by cryptographic means in association, each data block containing a batch of information of network transactions for verifying the validity of the information (anti-counterfeiting) and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.
It is to be understood that the terminology used in the description of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should also be understood that the term "and/or" as used in this specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations. It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments. While the invention has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made and equivalents will be apparent to those skilled in the art without departing from the scope of the invention. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method of claim settlement based on artificial intelligence, comprising:
acquiring a claim settlement knowledge graph, wherein the claim settlement knowledge graph comprises attribute information, relation information and claim settlement results of a plurality of entities involved in a plurality of claim settlement policy;
generating a plurality of sample data according to the claim knowledge graph, wherein the sample data comprises attribute information, relation information and claim settlement results of a plurality of entities related to the claim settlement policy;
based on an DQN algorithm, performing reinforcement learning training on a preset first deep neural network by using the plurality of sample data to obtain a claim settlement decision model;
and acquiring an enterprise knowledge graph corresponding to the to-be-claiming policy, and inputting the enterprise knowledge graph into the claim settlement decision model to perform decision processing so as to obtain a target claim settlement policy of the to-be-claiming policy.
2. The method of claim 1, wherein generating a plurality of sample data from the claim knowledge-graph comprises:
randomly determining a target entity in the claim knowledge graph;
acquiring attribute information, relation information and claim settlement results of a plurality of entities corresponding to the claim settlement policy associated with the target entity from the claim settlement knowledge graph as one sample of data;
and counting the number of the sample data, if the number is smaller than the preset number, returning to the step of determining the target entity in the claim knowledge graph, and if the number reaches the preset number, stopping generating the sample data.
3. The method of claim 2, wherein randomly determining target entities in the claim knowledge-graph comprises:
generating a random number in a preset range, and determining an entity with the same identification code in the claim knowledge graph as the random number as a target entity.
4. The method of claim 2, wherein randomly determining target entities in the claim knowledge-graph comprises:
randomly selecting a preset policy number from a preset policy number library as a target policy number, wherein the preset policy number library comprises policy numbers corresponding to each claim settlement policy related to the claim knowledge graph;
And acquiring an identification code group associated with the target insurance policy number, and determining any entity of the identification code in the identification code group in the claim knowledge graph as a target entity.
5. The method according to any one of claims 1-4, wherein the performing reinforcement learning training on a preset deep neural network by using the plurality of sample data based on DQN algorithm to obtain a claim decision model includes:
selecting a learning sample from the plurality of sample data, wherein the learning sample comprises a state and a claim settlement result, and the state is attribute information and relation information of a plurality of entities;
inputting the state into the first deep neural network to perform claim settlement decision processing to obtain the action value of each claim settlement decision, and determining a predicted claim settlement strategy according to the action value of each claim settlement decision;
executing the predicted claim settlement strategy in the claim settlement knowledge graph to obtain a predicted claim settlement result, and determining a target reward according to the predicted claim settlement result and the claim settlement result;
determining a target value according to the target reward, a next learning sample of the sample data and a preset second deep neural network, wherein the first deep neural network and the second deep neural network have the same structure;
Determining a loss value of the first deep neural network according to the target value and the action value of the predicted claim result;
when the loss value is larger than a preset loss value, updating parameters of the first deep neural network according to the target value and the action value of the predicted claim result;
and continuing reinforcement learning training on the first deep neural network by using the next learning sample until the loss value of the first deep neural network is smaller than or equal to a preset loss value.
6. The method of claim 5, further comprising:
storing the state in the learning sample, the state in the next learning sample, the target prize, and the predictive claim settlement strategy as one experience to an experience pool;
sampling a plurality of experiences from the experience pool at preset interval time, and updating parameters of the first deep neural network according to the plurality of experiences.
7. The method of claim 5, further comprising:
determining the update times of the first deep neural network;
and when the updating times are integral multiples of a preset value, updating the parameters of the second deep neural network to the current parameters of the first deep neural network.
8. An artificial intelligence based claim settlement device, characterized in that the claim settlement device comprises:
the system comprises an acquisition module, a calculation module and a calculation module, wherein the acquisition module is used for acquiring a claim settlement knowledge graph, wherein the claim settlement knowledge graph comprises attribute information, relation information and claim settlement results of a plurality of entities involved in a plurality of claim settlement plan insurance policies;
the sample generation module is used for generating a plurality of sample data according to the claim knowledge graph, wherein the sample data comprises attribute information, relation information and claim result of a plurality of entities related to the claim settlement policy;
the model training module is used for performing reinforcement learning training on a preset first deep neural network by using the plurality of sample data based on an DQN algorithm to obtain a claim settlement decision model;
the acquisition module is also used for acquiring an enterprise knowledge graph corresponding to the claim policy to be settled;
and the claim settlement decision module is used for inputting the enterprise knowledge graph into the claim settlement decision model for decision processing to obtain the target claim settlement strategy of the claim policy to be settled.
9. A computer device comprising a processor, a memory, and a computer program stored on the memory and executable by the processor, wherein the computer program when executed by the processor implements the artificial intelligence based claim method of any one of claims 1 to 7.
10. A storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements the artificial intelligence based claim method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310492664.3A CN116523661A (en) | 2023-05-04 | 2023-05-04 | Claim settlement method, device, equipment and storage medium based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310492664.3A CN116523661A (en) | 2023-05-04 | 2023-05-04 | Claim settlement method, device, equipment and storage medium based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116523661A true CN116523661A (en) | 2023-08-01 |
Family
ID=87389876
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310492664.3A Pending CN116523661A (en) | 2023-05-04 | 2023-05-04 | Claim settlement method, device, equipment and storage medium based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116523661A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117391313A (en) * | 2023-12-12 | 2024-01-12 | 广东正迪科技股份有限公司 | Intelligent decision method, system, equipment and medium based on AI |
| CN117934177A (en) * | 2024-03-22 | 2024-04-26 | 湖南多层次商保科技有限公司 | Method and system for constructing insurance intelligent responsibility determination model |
-
2023
- 2023-05-04 CN CN202310492664.3A patent/CN116523661A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117391313A (en) * | 2023-12-12 | 2024-01-12 | 广东正迪科技股份有限公司 | Intelligent decision method, system, equipment and medium based on AI |
| CN117391313B (en) * | 2023-12-12 | 2024-04-30 | 广东正迪科技股份有限公司 | Intelligent decision method, system, equipment and medium based on AI |
| CN117934177A (en) * | 2024-03-22 | 2024-04-26 | 湖南多层次商保科技有限公司 | Method and system for constructing insurance intelligent responsibility determination model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10679169B2 (en) | Cross-domain multi-attribute hashed and weighted dynamic process prioritization | |
| CN112035541A (en) | Client image drawing method and device, computer readable storage medium and terminal equipment | |
| CN111429006A (en) | Financial risk index prediction model construction method and device and risk situation prediction method and device | |
| CN106097043A (en) | The processing method of a kind of credit data and server | |
| CN112801430B (en) | Task issuing method and device, electronic equipment and readable storage medium | |
| CN116523661A (en) | Claim settlement method, device, equipment and storage medium based on artificial intelligence | |
| CN114679341B (en) | Network intrusion attack analysis method, equipment and medium combined with ERP system | |
| CN114266664A (en) | Transaction information prediction method and device, computer equipment and storage medium | |
| EP3764310A1 (en) | Prediction task assistance device and prediction task assistance method | |
| US11199832B2 (en) | Managing activities on industrial products according to compliance with reference policies | |
| CN116797345A (en) | Task processing method, device, computer equipment and storage medium | |
| CN113779116B (en) | Object ordering method, related equipment and medium | |
| US20220138552A1 (en) | Adapting ai models from one domain to another | |
| CN116542728A (en) | Service processing method, device, program product, computer equipment and medium | |
| CN111798299A (en) | Account settlement method and device for financial statement | |
| CN114331595B (en) | Product recommendation method, device, equipment and storage medium based on artificial intelligence | |
| CA3097644C (en) | Covariate processing with neural network execution blocks | |
| AU2021318113B2 (en) | Combining rules-based knowledge engineering with machine learning prediction | |
| EP4075348A1 (en) | Quality control of a machine learning model | |
| CN116561647A (en) | Policy classification method, device, equipment and storage medium based on artificial intelligence | |
| CN118096229A (en) | Client resource determining method, model training method, electronic device and storage medium | |
| CN116542762A (en) | User risk decision method, device, computer equipment and medium | |
| CN116227706A (en) | Prediction method, prediction apparatus, prediction device, storage medium, and program product | |
| CN117195848A (en) | Report generation method, device and equipment for business data and storage medium | |
| Jasinevicius et al. | Dynamic fuzzy expert maps: idea and implementation |
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 |