CN109685673A - A kind of insurance coupled customer service system and method based on artificial intelligence - Google Patents
A kind of insurance coupled customer service system and method based on artificial intelligence Download PDFInfo
- Publication number
- CN109685673A CN109685673A CN201811561551.XA CN201811561551A CN109685673A CN 109685673 A CN109685673 A CN 109685673A CN 201811561551 A CN201811561551 A CN 201811561551A CN 109685673 A CN109685673 A CN 109685673A
- Authority
- CN
- China
- Prior art keywords
- customer service
- answer
- artificial
- machine
- artificial customer
- 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
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000010168 coupling process Methods 0.000 claims abstract description 30
- 230000008878 coupling Effects 0.000 claims abstract description 29
- 238000005859 coupling reaction Methods 0.000 claims abstract description 29
- 238000004519 manufacturing process Methods 0.000 claims abstract description 18
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 10
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 14
- 230000004048 modification Effects 0.000 claims description 8
- 238000012986 modification Methods 0.000 claims description 8
- 230000008859 change Effects 0.000 claims description 7
- 230000008901 benefit Effects 0.000 abstract description 6
- 230000003993 interaction Effects 0.000 abstract description 5
- 238000005516 engineering process Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009466 transformation 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/225—Feedback of the input speech
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Multimedia (AREA)
- Acoustics & Sound (AREA)
- Human Computer Interaction (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Marketing (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Technology Law (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a kind of insurance coupled customer service system and method based on artificial intelligence, which includes artificial customer service selecting unit, voice recognition unit, machine artificial intelligence coupling unit, speech production unit;The artificial customer service selecting unit is directly realized by the chat of client Yu artificial customer service;The acoustic information of client is converted text information output by the voice recognition unit;The machine artificial intelligence coupling unit receives the information of text information and artificial customer service selecting unit simultaneously, realizes that the chat of client and artificial customer service or intelligence generate nature answer text, and carry out the intensified learning of machine;The natural answer text of input is sent to speech synthesis engine by the speech production unit, is generated acoustic information and is fed back to client.The present invention can be provided for client it is pre-sales, seek advice from after sale, the abilities such as Products Show and common knowledge problem.The advantages that high-accuracy, quickly answer, high reliability and interaction nature can be reached simultaneously.
Description
Technical field
The present invention relates to insurance customer service system, in particular to a kind of insurance coupled customer service systems based on artificial intelligence
And method.
Background technique
Insurance customer service is one of the service closest to client in insurance industry, it includes the publicity for insuring knowledge, protects
The expansion of dangerous business, the pre-sales consulting after sale of client etc..Insurance customer service is also wanted in having the function of in insurance industry, is connection
The tie of insurance company and client.The customer service of insurance industry at present is still artificial customer service, although artificial customer service has interaction
It is natural, answer the advantages that smooth, but be also inevitably present that main inertia is strong, the horizontal height of service quality is uneven, be unable to whole day
The disadvantages of waiting service.
With the arrival of internet and artificial intelligence epoch, insurance service is encountered by new opportunities and challenges.Current people
Work customer service reaches a bottleneck stage, and there are a certain distance for the interconnection networking and the intelligentized general trend of events with social development.Though
The disadvantages of right artificial intelligence technology has obtained significant progress, but intelligent answer customer service system is unstable there is also this at present.Such as
The advantages of how being based on artificial intelligence technology, and capable of being compatible with artificial customer service is that current insurance customer service field needs once to change.
Summary of the invention
In view of the problems of the existing technology, the present invention provides a kind of insurance coupled customer service system based on artificial intelligence
System and method.
To achieve the above object, the specific technical solution of the present invention is as follows:
A kind of insurance coupled customer service system based on artificial intelligence, comprising: artificial customer service selecting unit, speech recognition
Unit, machine artificial intelligence coupling unit, speech production unit;The artificial customer service selecting unit directly manually enters for one
Mouthful, this artificial entrance is clicked, then can be directly realized by the chat of client Yu artificial customer service;The voice recognition unit is by the sound of client
Message breath is converted into text information output;The machine artificial intelligence coupling unit receives text information and artificial customer service choosing simultaneously
The information of unit is selected, realizes that the chat of client and artificial customer service or intelligence generate nature answer text;The speech production unit
The natural answer text of input is sent to speech synthesis engine, acoustic information is generated and feeds back to client.
Preferably, the machine artificial intelligence coupling unit receives the letter of text information and artificial customer service selecting unit simultaneously
The method of breath, the chat or intelligence generation nature answer text of realizing client and artificial customer service is as follows:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service record manual feedback is answered
Case carries out intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent out
Artificial customer service feedback model is given, intensified learning, the final generation for carrying out nature answer are carried out.
Preferably, the machine artificial intelligence coupling unit includes that artificial customer service question and answer learn subelement, machine customer service is asked
Answer study subelement;The artificial customer service question and answer study subelement includes artificial customer service opening module, instant messaging module, customer service
Feedback module, the first intensified learning module;The machine customer service question and answer study subelement includes nature semantic understanding module, dialogue
State tracking module, dialog strategy study module, intelligent answer generation module, the second intensified learning module;Described first strengthens
Study module, the second intensified learning module are shared intensified learning module;
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and fed back by the intelligence answer generation module after intelligent answer generation
Module;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It is selected in artificial customer service
When signal Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
Preferably, the concrete operations of the intensified learning of the intensified learning module are as follows:
Enable the cost function of intensified learning are as follows:
V π (s)=Ε π (Rt+1+γRt+2+γ2Rt+2+Λ|St=s, at=a)
Wherein π is the strategic function of current machine intensified learning, and s is the state of current machine customer service, and a is current machine visitor
The movement of clothes, γ are the attenuation coefficient of reward or punishment, Rt+1 is the reward or punishment at t+1 moment, wherein in the first extensive chemical
Practise module, Rt+1Directly it is set as rewarding, while artificial customer service is selected into movement a of the answer as machine customer service, and give and encourage
It encourages;In second intensified learning module, Rt+1It is to be rewarded or punished that (such as artificial customer service is repaired according to the feedback of artificial customer service
Change dynamics or final feedback);The objective function of intensified learning is
argmaxπvπ(s)
Optimize the strategic function of machine intensified learning, so that the state of current machine customer service obtains maximum value.
Preferably, in the artificial customer service question and answer study subelement, the intensified learning module is mentioned using artificial customer service
The answer of the answer of confession or artificial customer service modification, using nitrification enhancement, the dynamics that artificial customer service is modified is as environment
The rewards and punishments factor, guidance machine customer service are learnt.
Preferably, in machine customer service question and answer study subelement, the intensified learning module receives artificial customer service
The feedback of reward and punishment, the study of machine question and answer will enter the intensified learning stage, be drawn using the direct rewards and punishments feedback of artificial customer service
Machine question and answer learning algorithm is led, and the answer that machine generates is directly output to speech production module.
The present invention also provides a kind of insurance coupled client service method based on artificial intelligence, includes the following steps:
It configures artificial customer service selecting unit, as a direct artificial entrance, clicks this artificial entrance, then it can be directly real
The chat of existing client and artificial customer service;
Voice recognition unit is set, converts text information output for the acoustic information of client;
Using machine artificial intelligence coupling unit, the information of text information and artificial customer service selecting unit is received simultaneously,
The chat or intelligence for realizing client and artificial customer service generate nature answer text;
The natural answer text of input is sent to speech synthesis engine using speech production unit, it is anti-to generate acoustic information
Feed client.
Preferably, the machine artificial intelligence coupling unit receives the letter of text information and artificial customer service selecting unit simultaneously
The method of breath, the chat or intelligence generation nature answer text of realizing client and artificial customer service is as follows:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service record manual feedback is answered
Case carries out intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent out
Artificial customer service feedback model is given, intensified learning, the final generation for carrying out nature answer are carried out.
Preferably, the machine artificial intelligence coupling unit includes that artificial customer service question and answer learn subelement, machine customer service is asked
Answer study subelement;The artificial customer service question and answer study subelement includes artificial customer service opening module, instant messaging module, customer service
Feedback module, the first intensified learning module;The machine customer service question and answer study subelement includes nature semantic understanding module, dialogue
State tracking module, dialog strategy study module, intelligent answer generation module, the second intensified learning module;Described first strengthens
Study module, the second intensified learning module are shared intensified learning module;
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and fed back by the intelligence answer generation module after intelligent answer generation
Module;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It is selected in artificial customer service
When signal Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
Preferably, in the artificial customer service question and answer study subelement, the intensified learning module is mentioned using artificial customer service
The answer of the answer of confession or artificial customer service modification, using nitrification enhancement, the dynamics that artificial customer service is modified is as environment
The rewards and punishments factor, guidance machine customer service are learnt.
Preferably, in machine customer service question and answer study subelement, the intensified learning module receives artificial customer service
The feedback of reward and punishment, the study of machine question and answer will enter the intensified learning stage, be drawn using the direct rewards and punishments feedback of artificial customer service
Machine question and answer learning algorithm is led, and the answer that machine generates is directly output to speech production module.
Present invention employs intensified learning technology, the insurance intelligent customer service system based on artificial intelligence technology can be realized,
Artificial customer service coupling technique is introduced simultaneously, strengthens the stability of practical insurance customer service system, while being also insurance intelligence visitor
Clothes study provides the feedback of external environment.The invention comprehensively utilizes the rapidities of machine intelligence customer service system, objectivity, low
The advantages that stability, naturality of cost-effectivenes, round-the-clock property and artificial customer service system, therefore technical method of the invention is with comprehensive
The technical effect for closing machine intelligence customer service system and artificial customer service system advantage is easy to be used by industry and client receives.
The present invention can be provided for client it is pre-sales, seek advice from after sale, the abilities such as Products Show and common knowledge problem.It can
The advantages that reaching high-accuracy, quickly answer, high reliability and interaction nature simultaneously.
Detailed description of the invention
Fig. 1 is present system composition block diagram;
Fig. 2 is the schematic diagram of machine artificial intelligence coupling unit in the present invention.
Specific embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is further described.
Referring to figs. 1 to Fig. 2, a kind of insurance coupled customer service system based on artificial intelligence of the present invention, comprising: artificial visitor
Take selecting unit, voice recognition unit, machine artificial intelligence coupling unit, speech production unit;The artificial customer service selection is single
Member is a direct artificial entrance, clicks this artificial entrance, then can be directly realized by the chat of client Yu artificial customer service;The voice
The acoustic information of client is converted text information output by recognition unit;The machine artificial intelligence coupling unit receives text simultaneously
The information of this information and artificial customer service selecting unit realizes that the chat of client and artificial customer service or intelligence generate nature answer text
This;The natural answer text of input is sent to speech synthesis engine by the speech production unit, is generated acoustic information and is fed back to
Client.
Wherein, the machine artificial intelligence coupling unit receives the letter of text information and artificial customer service selecting unit simultaneously
The method of breath, the chat or intelligence generation nature answer text of realizing client and artificial customer service is as follows:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service record manual feedback is answered
Case carries out intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent out
Artificial customer service feedback model is given, intensified learning, the final generation for carrying out nature answer are carried out.
As a preferred embodiment of the invention, the machine artificial intelligence coupling unit includes artificial customer service question and answer study
Subelement, machine customer service question and answer learn subelement;The artificial customer service question and answer study subelement include artificial customer service opening module,
Instant messaging module, customer service feedback module, the first intensified learning module;The machine customer service question and answer study subelement includes nature
Semantic understanding module, dialogue state tracking module, dialog strategy study module, intelligent answer generation module, the second intensified learning
Module;The first intensified learning module, the second intensified learning module are shared intensified learning module, intensified learning it is specific
It operates as follows:
Enable the cost function of intensified learning are as follows:
vπ(s)=Επ(Rt+1+γRt+2+γ2Rt+2+Λ|St=s, at=a)
Wherein π is the strategic function of current machine intensified learning, and s is the state of current machine customer service, and a is current machine visitor
The movement of clothes, γ are the attenuation coefficient of reward or punishment, Rt+1 is the reward or punishment at t+1 moment, wherein in the first extensive chemical
Practise module, Rt+1Directly it is set as rewarding, while artificial customer service is selected into movement a of the answer as machine customer service, and give and encourage
It encourages;In second intensified learning module, Rt+1It is to be rewarded or punished that (such as artificial customer service is repaired according to the feedback of artificial customer service
Change dynamics or final feedback);The objective function of intensified learning is
arg maxπvπ(s)
Optimize the strategic function of machine intensified learning, so that the state of current machine customer service obtains maximum value.
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and fed back by the intelligence answer generation module after intelligent answer generation
Module;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It is selected in artificial customer service
When signal Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
As a preferred embodiment of the invention, in the artificial customer service question and answer study subelement, the intensified learning
The answer of answer or artificial customer service modification that module is provided using artificial customer service is repaired artificial customer service using nitrification enhancement
The rewards and punishments factor of the dynamics changed as environment, guidance machine customer service are learnt.
As a preferred embodiment of the invention, in machine customer service question and answer study subelement, the intensified learning
Module receives the reward of artificial customer service and the feedback of punishment, and the study of machine question and answer will enter the intensified learning stage, and utilize artificial visitor
The direct rewards and punishments feedback guidance machine question and answer learning algorithm of clothes, and the answer that machine generates is directly output to speech production mould
Block.
The present invention also provides a kind of insurance coupled client service method based on artificial intelligence, includes the following steps:
It configures artificial customer service selecting unit, as a direct artificial entrance, clicks this artificial entrance, then it can be directly real
The chat of existing client and artificial customer service;
Voice recognition unit is set, converts text information output for the acoustic information of client;
Using machine artificial intelligence coupling unit, the information of text information and artificial customer service selecting unit is received simultaneously,
The chat or intelligence for realizing client and artificial customer service generate nature answer text;
The natural answer text of input is sent to speech synthesis engine using speech production unit, it is anti-to generate acoustic information
Feed client.
As a preferred embodiment of the invention, the machine artificial intelligence coupling unit receives text information and people simultaneously
The information of work customer service selecting unit realizes that the chat of client and artificial customer service or intelligence generate the method for nature answer text such as
Under:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service record manual feedback is answered
Case carries out intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent out
Artificial customer service feedback model is given, intensified learning, the final generation for carrying out nature answer are carried out.
As a preferred embodiment of the invention, the machine artificial intelligence coupling unit includes artificial customer service question and answer study
Subelement, machine customer service question and answer learn subelement;The artificial customer service question and answer study subelement include artificial customer service opening module,
Instant messaging module, customer service feedback module, the first intensified learning module;The machine customer service question and answer study subelement includes nature
Semantic understanding module, dialogue state tracking module, dialog strategy study module, intelligent answer generation module, the second intensified learning
Module;The first intensified learning module, the second intensified learning module are shared intensified learning module;
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and fed back by the intelligence answer generation module after intelligent answer generation
Module;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It is selected in artificial customer service
When signal Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
As a preferred embodiment of the invention, in the artificial customer service question and answer study subelement, the intensified learning
The answer of answer or artificial customer service modification that module is provided using artificial customer service is repaired artificial customer service using nitrification enhancement
The rewards and punishments factor of the dynamics changed as environment, guidance machine customer service are learnt.
As a preferred embodiment of the invention, in machine customer service question and answer study subelement, the intensified learning
Module receives the reward of artificial customer service and the feedback of punishment, and the study of machine question and answer will enter the intensified learning stage, and utilize artificial visitor
The direct rewards and punishments feedback guidance machine question and answer learning algorithm of clothes, and the answer that machine generates is directly output to speech production mould
Block.
The working principle of the invention:
As shown in figs. 1 and 2, the insurance coupled customer service system proposed by the present invention based on artificial intelligence includes four big single
Element module, including artificial customer service selecting unit, voice recognition unit, speech synthesis unit and the coupling of machine artificial intelligence are single
Member.Wherein machine artificial intelligence coupling module, based on artificial intelligence latest developments deeply study and natural language at
Reason technology carries out coupled, is the nucleus module of system.Insurance coupled customer service system step based on artificial intelligence is such as
Under:
Step 0: artificial customer service selecting unit, artificial customer service selecting unit mainly provide one directly for intelligent customer service system
The artificial entrance connect.This artificial entrance is clicked, then can be directly entered the chat that step 2 realizes client and artificial customer service.
Step 1: the acoustic information of client is converted text information by voice recognition unit.In client interaction, voice
Interaction is one of most natural interactive mode, we utilize third-party speech recognition engine, and the voice of client is inputted conversion
For text output.
Step 2: machine artificial intelligence coupling unit.Simultaneously speak Chinese text information and artificial customer service selection be input to this list
Element module.Artificial customer service selection signal if it exists then directly executes instant messaging, realizes the chat of client and artificial customer service, together
When machine customer service record manual feedback answer, carry out intensified learning.Artificial customer service selection signal if it does not exist, then execute machine intelligence
Energy customer service question and answer information, and the answer of generation is sent to artificial customer service feedback model, intensified learning is carried out, nature is finally carried out
The generation of answer.
Step 3: speech synthesis unit.Herein by the natural answer of input, speech synthesis engine is inputted, sound letter is generated
Breath, enables the customer to understand answer.
Step 4: step 0,1,2 and 3 are constantly repeated, until client terminates customer service system.
More specifically, the machine artificial intelligence coupling module in step 2, the behaviour of answer is generated to the problem of client's input
Make the processing that process will carry out Liang Ge branch according to the presence or absence of artificial customer service selection signal: artificial customer service question and answer study and machine
The study of device customer service question and answer.
The operation of artificial customer service question and answer study is as follows:
Operation 1: artificial customer service opening module.When receiving artificial customer service selection instruction, artificial customer service opening module works, by
Machine intelligence mode enters artificial customer service mode, opens artificial customer service interface;Client can carry out defeated at artificial customer service interface
Enter the operations such as text, input voice, uploading pictures.Artificial customer service mode will carry out always, refer to until receiving artificial customer service closing
It enables.
Operation 2: instant messaging module.After starting artificial customer service module, instant messaging module interface realize client with
The chat of artificial customer service.Chat system supports offline message and history message to obtain, and customer service receives client and carries out row by redis
Team's distribution, customer service multiterminal log in realization and force down line principle, and customer service platform access robot interface (directly returns by robot answer
Multiple client), support customer evaluation function.
Operation 3: customer service feedback model.Customer service was connected to after the problem of client, was fed back to customer issue.In artificial visitor
Mode is taken, the answer of client will be there are two types of mode: having artificial customer service to complete completely;Machine customer service provides candidate answers, by artificial
Customer service modification is completed.Later by answer feedback to intensified learning module.
Operation 4: intensified learning module.Intensified learning module modifies the answer provided using artificial customer service or artificial customer service
Answer, using nitrification enhancement, the dynamics that artificial customer service is modified as the rewards and punishments factor of environment, guide machine customer service into
Row study.
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;When inartificial customer service letter
Number when, customer issue enters machine problem, and interactive interface is maintained at machine question and answer interface.The operation of machine customer service question and answer study is such as
Under:
Operation 1: natural semantic understanding.Word segmentation processing is carried out to the text of input, the text after word segmentation processing is subjected to language
Justice analysis, that is, analyze the classification and customer demand of customer issue.These information inputs to dialogue are loaded into tracking operation.
Operation 2: dialogue state tracking.Based on current dialogue state, i.e. the information to be filled out wanted of required by task, and be input to
The customer issue of module is semantic, carries out state classification using classifier, judges the state of current task.
Operation 3: dialog strategy study.Based on current task status, times that mostly wheel dialogue needs to obtain in next step is determined
It is engaged in information to be filled out;And the classification of client's current problem and demand are input to next operation.
Operation 4: intelligent answer generates.Based on the classification of client's current problem and demand, provided in knowledge base and corpus
The reply of customer issue;Simultaneously based on current task information to be filled out, the sentence of inquiry client is generated.And it is raw by natural language
It is post-processed at algorithm answer case.
Operation 5.1: it artificial customer service feedback model: after intelligent answer generation, needs for answer to be input to artificial customer service
Feedback model.When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;In artificial customer service
When selection signal Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
Operation 6: intensified learning model.It is input to the reward and punishment of the only artificial customer service of intensified learning model at this time
Feedback.The study of machine question and answer will enter the intensified learning stage, utilize the direct rewards and punishments feedback guidance machine question and answer of artificial customer service
Practise algorithm.And the answer that machine generates is directly output to speech production module.
The present invention has following innovative point:
1, the machine intelligence customer service based on deep understanding.Semantic understanding, question and answer shape are carried out using natural language processing technique
State tracking and decision, accuracy is high, and speed is fast.
2, the organic coupling of artificial customer service and machine intelligence customer service.The excellent of artificial customer service and machine customer service can be comprehensively utilized
Point promotes the quality of service, reduces the cost of service.
3, intensified learning technology has been used, based on the feedback that artificial customer service provides, has guided machine intelligent customer service system
It practises.
The above description is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all at this
Under the inventive concept of invention, using equivalent structure transformation made by description of the invention and accompanying drawing content, or directly/use indirectly
It is included in other related technical areas in scope of patent protection of the invention.
Claims (10)
1. a kind of insurance coupled customer service system based on artificial intelligence characterized by comprising artificial customer service selection is single
Member, voice recognition unit, machine artificial intelligence coupling unit, speech production unit;
The artificial customer service selecting unit be a direct artificial entrance, click this artificial entrance, then can be directly realized by client and
The chat of artificial customer service;
The acoustic information of client is converted text information output by the voice recognition unit;
The machine artificial intelligence coupling unit receives the information of text information and artificial customer service selecting unit simultaneously, realizes client
Chat or intelligence with artificial customer service generate nature answer text;
The natural answer text of input is sent to speech synthesis engine by the speech production unit, is generated acoustic information and is fed back to
Client.
2. insurance coupled customer service system according to claim 1, which is characterized in that the machine artificial intelligence coupling
Unit receives the information of text information and artificial customer service selecting unit simultaneously, realizes that the chat of client and artificial customer service or intelligence are raw
Method at natural answer text is as follows:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service records manual feedback answer, into
Row intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent to
Artificial customer service feedback model carries out intensified learning, the final generation for carrying out nature answer.
3. insurance coupled customer service system according to claim 1, which is characterized in that the machine artificial intelligence coupling
Unit includes artificial customer service question and answer study subelement, machine customer service question and answer study subelement;Artificial customer service question and answer study
Unit includes artificial customer service opening module, instant messaging module, customer service feedback module, the first intensified learning module;The machine
It includes nature semantic understanding module, dialogue state tracking module, dialog strategy study module, intelligence that customer service question and answer, which learn subelement,
Answer generation module, the second intensified learning module;The first intensified learning module, the second intensified learning module be share it is strong
Change study module;
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and feeds back mould by the intelligence answer generation module after intelligent answer generation
Block;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It selects to believe in artificial customer service
When number Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
4. insurance coupled customer service system according to claim 3, which is characterized in that the intensified learning module it is strong
The concrete operations that chemistry is practised are as follows:
Enable the cost function of intensified learning are as follows:
vπ(s)=Επ(Rt+1+γRt+2+γ2Rt+2+Λ|St=s, at=a)
Wherein π is the strategic function of current machine intensified learning, and s is the state of current machine customer service, and a is current machine customer service
Movement, γ are the attenuation coefficient of reward or punishment, Rt+1It is the reward or punishment at t+1 moment, wherein in the first intensified learning mould
Block, Rt+1Directly it is set as rewarding, while artificial customer service is selected into movement a of the answer as machine customer service, and award;
In second intensified learning module, Rt+1It is to be rewarded or punished (the modification power of such as artificial customer service according to the feedback of artificial customer service
Degree or final feedback);The objective function of intensified learning is
argmaxπvπ(s)
Optimize the strategic function of machine intensified learning, so that the state of current machine customer service obtains maximum value.
5. insurance coupled customer service system according to claim 3, which is characterized in that in the artificial customer service question and answer
It practises in subelement, the answer of answer or artificial customer service modification that the intensified learning module is provided using artificial customer service, using strong
Change learning algorithm, the dynamics that artificial customer service is modified guides machine customer service to be learnt as the rewards and punishments factor of environment;
In machine customer service question and answer study subelement, the intensified learning module receives the reward and punishment of artificial customer service
Feedback, the study of machine question and answer will enter the intensified learning stage, utilize the direct rewards and punishments feedback guidance machine question and answer of artificial customer service
Algorithm is practised, and the answer that machine generates is directly output to speech production module.
6. a kind of insurance coupled client service method based on artificial intelligence, which comprises the steps of:
Artificial customer service selecting unit is configured, as a direct artificial entrance, this artificial entrance is clicked, then can be directly realized by visitor
The chat at family and artificial customer service;
Voice recognition unit is set, converts text information output for the acoustic information of client;
Using machine artificial intelligence coupling unit, the information of text information and artificial customer service selecting unit is received simultaneously, is realized
The chat or intelligence of client and artificial customer service generate nature answer text;
The natural answer text of input is sent to speech synthesis engine using speech production unit, acoustic information is generated and feeds back to
Client.
7. insurance coupled client service method according to claim 6, which is characterized in that the machine artificial intelligence coupling
Unit receives the information of text information and artificial customer service selecting unit simultaneously, realizes that the chat of client and artificial customer service or intelligence are raw
Method at natural answer text is as follows:
Artificial customer service selection signal if it exists then directly executes instant messaging, while machine customer service records manual feedback answer, into
Row intensified learning;
Artificial customer service selection signal if it does not exist then executes machine intelligence customer service question and answer information, and the answer of generation is sent to
Artificial customer service feedback model carries out intensified learning, the final generation for carrying out nature answer.
8. insurance coupled client service method according to claim 6, which is characterized in that the machine artificial intelligence coupling
Unit includes artificial customer service question and answer study subelement, machine customer service question and answer study subelement;Artificial customer service question and answer study
Unit includes artificial customer service opening module, instant messaging module, customer service feedback module, the first intensified learning module;The machine
It includes nature semantic understanding module, dialogue state tracking module, dialog strategy study module, intelligence that customer service question and answer, which learn subelement,
Answer generation module, the second intensified learning module;The first intensified learning module, the second intensified learning module be share it is strong
Change study module;
When artificial customer service selection signal comes into force, machine customer service question and answer learn in running background;
The answer of generation is input to artificial customer service and feeds back mould by the intelligence answer generation module after intelligent answer generation
Block;When artificial customer service selection signal comes into force, reference that the answer of generation is replied as artificial customer service;It selects to believe in artificial customer service
When number Pending The Entry Into Force, the answer of generation will only obtain the feedback of the reward and punishment of artificial customer service.
9. insurance coupled client service method according to claim 8, which is characterized in that in the artificial customer service question and answer
It practises in subelement, the answer of answer or artificial customer service modification that the intensified learning module is provided using artificial customer service, using strong
Change learning algorithm, the dynamics that artificial customer service is modified guides machine customer service to be learnt as the rewards and punishments factor of environment.
10. insurance coupled client service method according to claim 8, which is characterized in that in the machine customer service question and answer
Learn in subelement, the intensified learning module receives the reward of artificial customer service and the feedback of punishment, and the study of machine question and answer will be into
Enter the intensified learning stage, using the direct rewards and punishments feedback guidance machine question and answer learning algorithm of artificial customer service, and machine is generated
Answer is directly output to speech production module.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811561551.XA CN109685673A (en) | 2018-12-19 | 2018-12-19 | A kind of insurance coupled customer service system and method based on artificial intelligence |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811561551.XA CN109685673A (en) | 2018-12-19 | 2018-12-19 | A kind of insurance coupled customer service system and method based on artificial intelligence |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109685673A true CN109685673A (en) | 2019-04-26 |
Family
ID=66186401
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811561551.XA Pending CN109685673A (en) | 2018-12-19 | 2018-12-19 | A kind of insurance coupled customer service system and method based on artificial intelligence |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109685673A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110166643A (en) * | 2019-06-18 | 2019-08-23 | 深圳市一号互联科技有限公司 | Control method of attending a banquet, system and the speech robot people of coupled |
CN110276639A (en) * | 2019-06-05 | 2019-09-24 | 温州职业技术学院 | A kind of network marketing system and method |
CN110909147A (en) * | 2019-12-02 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Method and system for training sorting result selection model output standard question method |
CN110910694A (en) * | 2019-11-28 | 2020-03-24 | 大唐融合通信股份有限公司 | Intelligent customer service training system |
CN111651554A (en) * | 2020-04-17 | 2020-09-11 | 世纪保众(北京)网络科技有限公司 | Insurance question-answer method and device based on natural language understanding and processing |
CN112580371A (en) * | 2020-12-25 | 2021-03-30 | 江苏鑫盛通讯科技有限公司 | Man-machine coupling customer service system and method based on artificial intelligence |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105162688A (en) * | 2015-08-13 | 2015-12-16 | 北京智齿博创科技有限公司 | Method for chatting with online customer service representative |
CN105207890A (en) * | 2015-08-24 | 2015-12-30 | 北京智齿博创科技有限公司 | Online customer service method |
CN106357942A (en) * | 2016-10-26 | 2017-01-25 | 广州佰聆数据股份有限公司 | Intelligent response method and system based on context dialogue semantic recognition |
WO2017041372A1 (en) * | 2015-09-07 | 2017-03-16 | 百度在线网络技术(北京)有限公司 | Man-machine interaction method and system based on artificial intelligence |
CN106550156A (en) * | 2017-01-23 | 2017-03-29 | 苏州咖啦魔哆信息技术有限公司 | A kind of artificial intelligence's customer service system and its implementation based on speech recognition |
CN107135247A (en) * | 2017-02-16 | 2017-09-05 | 江苏南大电子信息技术股份有限公司 | A kind of service system and method for the intelligent coordinated work of person to person's work |
CN107239538A (en) * | 2017-06-01 | 2017-10-10 | 深圳追科技有限公司 | Parallel customer service robot system and its self-learning method with self-learning function |
CN108170835A (en) * | 2018-01-12 | 2018-06-15 | 深圳市富途网络科技有限公司 | A kind of intelligent customer service system of combinatorial artificial and AI |
CN108280521A (en) * | 2017-01-06 | 2018-07-13 | 南通使爱智能科技有限公司 | Intelligent robot service system |
-
2018
- 2018-12-19 CN CN201811561551.XA patent/CN109685673A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105162688A (en) * | 2015-08-13 | 2015-12-16 | 北京智齿博创科技有限公司 | Method for chatting with online customer service representative |
CN105207890A (en) * | 2015-08-24 | 2015-12-30 | 北京智齿博创科技有限公司 | Online customer service method |
WO2017041372A1 (en) * | 2015-09-07 | 2017-03-16 | 百度在线网络技术(北京)有限公司 | Man-machine interaction method and system based on artificial intelligence |
CN106357942A (en) * | 2016-10-26 | 2017-01-25 | 广州佰聆数据股份有限公司 | Intelligent response method and system based on context dialogue semantic recognition |
CN108280521A (en) * | 2017-01-06 | 2018-07-13 | 南通使爱智能科技有限公司 | Intelligent robot service system |
CN106550156A (en) * | 2017-01-23 | 2017-03-29 | 苏州咖啦魔哆信息技术有限公司 | A kind of artificial intelligence's customer service system and its implementation based on speech recognition |
CN107135247A (en) * | 2017-02-16 | 2017-09-05 | 江苏南大电子信息技术股份有限公司 | A kind of service system and method for the intelligent coordinated work of person to person's work |
CN107239538A (en) * | 2017-06-01 | 2017-10-10 | 深圳追科技有限公司 | Parallel customer service robot system and its self-learning method with self-learning function |
CN108170835A (en) * | 2018-01-12 | 2018-06-15 | 深圳市富途网络科技有限公司 | A kind of intelligent customer service system of combinatorial artificial and AI |
Non-Patent Citations (4)
Title |
---|
NAN JING: "Responding to Subjective Changes of Customer Requirements in Dynamic Service Execution Environment", IEEE * |
刘晓强: "基于领域本体的客服问答系统的设计与实现", 中国优秀硕士学位论文全文数据库, no. 2 * |
耿倩;: "面向校园咨询场景的自动问答系统设计", 自动化技术与应用, no. 06 * |
蔡志文;林建宗;: "基于购买意向的移动电子商务智能客服系统", 科技管理研究, no. 18 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110276639A (en) * | 2019-06-05 | 2019-09-24 | 温州职业技术学院 | A kind of network marketing system and method |
CN110166643A (en) * | 2019-06-18 | 2019-08-23 | 深圳市一号互联科技有限公司 | Control method of attending a banquet, system and the speech robot people of coupled |
CN110910694A (en) * | 2019-11-28 | 2020-03-24 | 大唐融合通信股份有限公司 | Intelligent customer service training system |
CN110909147A (en) * | 2019-12-02 | 2020-03-24 | 支付宝(杭州)信息技术有限公司 | Method and system for training sorting result selection model output standard question method |
CN111651554A (en) * | 2020-04-17 | 2020-09-11 | 世纪保众(北京)网络科技有限公司 | Insurance question-answer method and device based on natural language understanding and processing |
CN112580371A (en) * | 2020-12-25 | 2021-03-30 | 江苏鑫盛通讯科技有限公司 | Man-machine coupling customer service system and method based on artificial intelligence |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109685673A (en) | A kind of insurance coupled customer service system and method based on artificial intelligence | |
US20220006761A1 (en) | Systems and processes for operating and training a text-based chatbot | |
Moore et al. | Using a functional linguistics metalanguage to support academic language development in the English Language Arts | |
Mercer | The analysis of classroom talk: Methods and methodologies | |
East et al. | What constitutes effective feedback to postgraduate research students? The students’ perspective | |
CN106777257B (en) | Intelligent dialogue model construction system and method based on dialect | |
JP2017504883A (en) | Model-driven candidate sorting based on audio cues | |
Koulouri et al. | Do (and say) as I say: Linguistic adaptation in human–computer dialogs | |
CN112084318B (en) | Dialogue assistance method, system and device | |
Rosé | Discourse analytics | |
Graesser et al. | Building intelligent conversational tutors and mentors for team collaborative problem solving: Guidance from the 2015 Program for International Student Assessment | |
Poser et al. | Design and evaluation of a conversational agent for facilitating idea generation in organizational innovation processes | |
Pandey et al. | Disruptive artificial intelligence and sustainable human resource management: Impacts and innovations-The future of HR | |
CN113505606B (en) | Training information acquisition method and device, electronic equipment and storage medium | |
Spee et al. | Design thinking and the hype cycle in management education and in engineering education | |
Ockey et al. | Assessing interactional competence: ICE versus a human partner | |
CN113326367A (en) | Task type dialogue method and system based on end-to-end text generation | |
US11715463B1 (en) | Omni-channel orchestrated conversation system and virtual conversation agent for realtime contextual and orchestrated omni-channel conversation with a human and an omni-channel orchestrated conversation process for conducting realtime contextual and fluid conversation with the human by the virtual conversation agent | |
Negi et al. | Automatically extracting dialog models from conversation transcripts | |
Nair | HR based Chatbot using deep neural network | |
Nicoleta | The impact of intercultural communication on cultural intelligence | |
CN113918687A (en) | Dialogistic training assistance system, method, computer device, and storage medium | |
Xia | Addressing the Tasks and Opportunities of Agency Using AI-based Chatbots | |
Lockwood | Theorising communications coaching in Asian contact centres: what is the approach? | |
Aksyonov et al. | Development of a Chatbot for the Subject Area Train Schedule for Telegram |
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 |