CN116578690B - Insurance customer service method and system based on artificial intelligence - Google Patents
Insurance customer service method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN116578690B CN116578690B CN202310857578.8A CN202310857578A CN116578690B CN 116578690 B CN116578690 B CN 116578690B CN 202310857578 A CN202310857578 A CN 202310857578A CN 116578690 B CN116578690 B CN 116578690B
- Authority
- CN
- China
- Prior art keywords
- customer service
- dialogue
- insurance customer
- word
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 33
- 239000013598 vector Substances 0.000 claims abstract description 269
- 238000005457 optimization Methods 0.000 claims abstract description 50
- 230000011218 segmentation Effects 0.000 claims description 48
- 238000004364 calculation method Methods 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 17
- 238000010606 normalization Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 36
- 238000004891 communication Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- 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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- 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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3346—Query execution using probabilistic model
-
- 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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3347—Query execution using vector based model
-
- 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
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06N3/047—Probabilistic or stochastic networks
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Accounting & Taxation (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Probability & Statistics with Applications (AREA)
- Finance (AREA)
- Biomedical Technology (AREA)
- Technology Law (AREA)
- Human Computer Interaction (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention relates to the technical field of customer service, and discloses an artificial intelligence-based insurance customer service method and system, wherein the method comprises the following steps: sequentially carrying out vectorization processing, coding processing based on a text coding model and self-attention weighting processing on the user question-answering data to obtain a weighted coding vector sequence of the user question-answering data; and acquiring a weighted coding vector sequence of the insurance customer service reference service dialogue data, constructing an insurance customer service dialogue model and carrying out model optimization solving based on an optimization objective function. According to the invention, word vector representation is carried out based on word position information, so that the position information capture of word vectors is realized, the self-attention weight of the positions in sentences is improved, the word vector sequences are encoded and represented and the self-attention weighting processing is carried out by combining the effective information of each sentence of word vector sequences, the effective capture of user question and answer information by an insurance customer service dialogue model is realized, and customer service is carried out based on the occurrence probability of different insurance customer service reference answer data.
Description
Technical Field
The invention relates to the technical field of customer service, in particular to an artificial intelligence-based insurance customer service method and system.
Background
With the rapid development of natural language technology, automatic dialogue robots based on "encoding-decoding" structures are becoming mature. Because of the rigor of insurance industry terms, the existing conversation robot cannot distinguish word expression modes which are expressed in a similar way but have different meanings, and further tends to generate conversation content with low information content, so that poor user experience is caused. Aiming at the problem, the invention provides an artificial intelligence-based insurance customer service method and system, which meet the semantic relevance and knowledge point matching with the reference standard through the text which is automatically generated by the text generation task under the guidance of the reference standard, thereby achieving the insurance customer service effect with better user experience.
Disclosure of Invention
In view of the above, the present invention provides an artificial intelligence based insurance customer service method, which aims to: 1) The method comprises the steps that word vector representation is carried out on the basis of the position of each word segmentation result in sentence in insurance customer service reference dialogue data, wherein the position of the word segmentation result in the sentence is word vector representation weight, the weight of word vector representation of the position in the sentence is larger, the position information capture of the word vector is realized, the self-attention weight of the position in the sentence is improved, the insurance customer service dialogue model can capture the position information in the sentence more effectively, the above effective information of each sentence word vector sequence is combined, the word vector sequence is subjected to coding representation and self-attention weighting processing, and the coded vector sequence containing the above effective information and main information attention weighted is obtained, so that the effective capture of the insurance customer service dialogue model on the user question-answer information is realized; 2) The method comprises the steps of solving model parameters which enable probability distribution of different insurance customer service reference answer data output by an insurance customer service dialogue model to be similar to probability distribution of insurance customer service reference answer data in collected insurance customer service reference service dialogue data by taking a minimized and optimized objective function as a target, constructing an insurance customer service dialogue model according to the model parameters, automatically selecting an insurance customer service reference service dialogue with highest occurrence probability corresponding to main dialogue information in the insurance customer service reference service dialogue data according to each sentence of question answer data of a user, and outputting the insurance customer service reference dialogue, so that automatic insurance customer service is realized.
In order to achieve the above purpose, the invention provides an artificial intelligence-based insurance customer service method, which comprises the following steps:
s1: collecting insurance customer service reference service dialogue data, and carrying out vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence;
s2: inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector, and forming a coding vector sequence;
s3: performing self-attention weighting on the coded vector sequence to obtain a weighted coded vector sequence;
s4: calculating the probability of each code vector in the weighted code vector sequence;
s5: constructing an insurance customer service dialogue model, and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model is a neural network model;
s6: and carrying out optimization solving on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the insurance customer service dialogue model obtained by solving.
As a further improvement of the present invention:
optionally, in the step S1, insurance customer service reference service dialogue data is collected, and vectorization processing is performed on each word in the reference service dialogue data, including:
Collecting N groups of insurance customer service reference service dialogue data, wherein the form of the insurance customer service reference service dialogue data is as follows:
；
wherein:
representing the n-th group insurance customer service reference service dialogue data,>user question data in the n-th group insurance customer service reference service dialogue data>Representation pair->Insurance customer service reference answer data; in the embodiment of the invention, the dialogue data are text data;
each word in the reference service dialogue data is subjected to vectorization processing, wherein the vectorization processing flow of the reference service dialogue data is as follows:
s11: pre-constructing an insurance frequent word segmentation vocabulary, for any sentence text in reference service dialogue data, intercepting the first e text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, if the comparison is unsuccessful, intercepting the first e-1 text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, and so on until the comparison is successful or the text characters to be compared only remain one text character, deleting the text characters from the sentence text, if the comparison is successful, recording the text characters for comparison, deleting the text characters for comparison from the sentence text, and repeating the current steps until all the sentence text characters are split into word forms, and taking all the recorded text characters as word segmentation results; in the embodiment of the invention, the comma, the semicolon and the period are used for dividing the text sentence number of the reference service dialogue data;
Wherein the method comprises the steps ofThe word segmentation result of (2) is as follows:
；
；
wherein:
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
s12: constructing a status register with the length number of E, wherein E is the total number of common words in the insurance common word segmentation vocabulary, and a corresponding relation is established between each register position and the common words in the insurance common word segmentation vocabulary, and the initial value of each position of the status register is 0;
s13: matching each word segmentation result with a common word corresponding to each register bit in the state register, and adjusting the register bit successfully matched to 1 to obtain a coding result of the word segmentation result, wherein the word segmentation result is obtainedThe result of the encoding is->；
S14: vectorizing the coding result of the word segmentation result, wherein the coding resultIs a vectorized representation formula of (c)The method comprises the following steps:
；
wherein:
representation->Is a vectorized representation of the results;
s15: the vectorization representation results of all word segmentation results in each sentence of text of the reference service dialogue data form a group of word vector sequences to obtain the dialogue word vector sequence of each group of reference service dialogue data, wherein The sequence of dialogue word vectors of (a) is:
；
；
wherein:
representation->Is a sequence of dialogue word vectors,/>Representation->Is a sequence of dialogue word vectors;
representation->Middle->Word vector sequences of sentence text.
Optionally, in the step S2, the sequence of dialogue word vectors is input to a text coding model to obtain a coding vector corresponding to each dialogue word vector, which includes:
inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector and form a coding vector sequence, wherein the dialogue word vector based on the text coding modelThe coding flow of (a) is as follows:
s21: text encoding model receives dialogue word vectors；
S22: encoding arbitrary word vector sequences in dialogue word vectors, whereinThe encoding processing formula of (2) is:
；
；
wherein:
weight parameters representing the coding process, +.>Representing bias parameters for the encoding process;
representing an activation function->The method comprises the steps of carrying out a first treatment on the surface of the In the embodiment of the invention, the following steps are performedSet to->A function;
representing +.>Is a coded vector of (a);
representing from->The useful characteristic information extracted from the data is used for the data processing,；
s23: constructing a dialogue word vectorIs a sequence of encoded vectors of (a):
；
wherein:
representing dialogue word vector +. >Is described.
Optionally, the self-attention weighting processing is performed on the coded vector sequence in the step S3, including:
generating self-attention weight vectors for arbitrary coded vectors, wherein the coded vector sequencesThe self-attention weight vector of (2) is:
；
wherein:
representing the coded vector sequence->Is a self-attention weight vector of (2);
based on self-attention weight vectorCoding vector sequence->Performing self-attention weighting processing, wherein the self-attention weighting processing formula is as follows:
；
wherein:
representation->Is a self-attention weighted result of (2);
the weighted coded vector sequence is:
；
wherein:
representing the coded vector sequence->Is a self-attention weighted result of (c).
Optionally, the calculating in step S4 the probability of occurrence of each weighted coded vector sequence includes:
calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of user problem data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector sequence if the cosine similarity is higher than a preset similarity threshold;
calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of insurance customer service reference answer data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector if the cosine similarity is higher than a preset similarity threshold;
Calculating to obtain an arbitrary weighted coded vector sequenceAnd the weighted coded vector sequence +.>And probability of occurrence of the same weighted coded vector sequence:wherein->Representing an arbitrarily weighted coded vector sequence +.>And +.>Is a set of identical weighted coded vector sequences, is a set of weighted coded vector sequences>Representing a weighted coded vector sequence->And +.>Is used to encode the set of vector sequences.
Optionally, the step S5 constructs an insurance customer service dialogue model, and determines an optimization objective function based on the occurrence probability of the above coding vector, including:
constructing an insurance customer service dialogue model and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model comprises an input layer, a probability calculation layer and an output layer, the input layer is used for receiving weighted coding vector sequences of user question and answer data, the probability calculation layer comprises a convolution layer and a normalization layer and is used for calculating the occurrence probability of the weighted coding vector sequences corresponding to different insurance customer service reference answer data, and the output layer selects the insurance customer service reference answer data with the highest occurrence probability to output;
The optimization objective function of the insurance customer service dialogue model is as follows:
；
；
wherein:
an optimized objective function representing an insurance customer service dialogue model, < >>Convolution calculation parameters representing probability calculation layers in insurance customer service dialogue model, namely weight parameters in convolution layers, +.>Representing a convolution layer;
the representation will->Input to base +.>In the insurance customer service dialogue model, the +.f. calculated by the probability calculation layer in the insurance customer service dialogue model>Sum of occurrence probabilities of->Representing normalization processing;
representing +.>The sum of the occurrence probabilities of the medium weighted coded vector sequences;
and solving to obtain model parameters which enable probability distribution of different insurance customer service reference answer data output by the insurance customer service dialogue model to be similar to probability distribution of the insurance customer service reference answer data in the acquired insurance customer service reference service dialogue data by taking the minimum optimization objective function as a target.
Optionally, in the step S6, the optimizing and solving the insurance customer service dialogue model based on the optimizing objective function includes:
and carrying out optimization solution on the insurance customer service dialogue model based on an optimization objective function, wherein the optimization solution flow is as follows:
s61: setting parameters to be optimized of insurance customer service dialogue model The current optimization times are t, the initial value of t is 1, and the parameter to be optimized is randomly generated>And calculates the gradient of the current optimized objective function as +.>The parameter to be optimized obtained by the t-th iteration is +.>The corresponding gradient of the optimized objective function is +.>；
S62: if it isOutput +.>As parameters obtained by optimization and based on the parameters +.>Building an insurance customer service dialogue model, namely->Is a smaller positive number, wherein +.>
S63: calculating to obtain iteration step length：
；
S64: if it isSatisfying the following constraint, then remain +.>Otherwise let->：
；
S65: for a pair ofAnd (3) performing iterative optimization:
；
order theThe process returns to step S62.
Optionally, in the step S6, an insurance customer service session is performed by using the obtained insurance customer service session model, including:
and carrying out insurance customer service dialogue by utilizing the obtained insurance customer service dialogue model, wherein the dialogue flow based on the insurance customer service dialogue model is as follows:
acquiring user question-answer data, and sequentially carrying out vectorization processing, coding processing based on a text coding model and self-attention weighting processing on the user question-answer data to obtain a weighted coded vector sequence of the user question-answer data;
transmitting the weighted coded vector sequence of the user question-answering data to an insurance customer service dialogue model;
The probability calculation layer of the insurance customer service dialogue model calculates the occurrence probability of the coded vector sequence after being weighted corresponding to different insurance customer service reference answer data, and the output layer selects the corresponding insurance customer service reference answer data with the highest occurrence probability for output.
In order to solve the above problems, the present invention provides an artificial intelligence based insurance customer service system, the system comprising:
the text coding device is used for collecting the insurance customer service reference service dialogue data, carrying out vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence, inputting the dialogue word vector sequence into the text coding model to obtain a coding vector corresponding to each dialogue word vector, and forming a coding vector sequence;
the coding vector weighting module is used for carrying out self-attention weighting on the coding vector sequence to obtain a weighted coding vector sequence, and calculating the probability of each coding vector in the weighted coding vector sequence;
and the insurance customer service dialogue module is used for carrying out optimization solution on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the insurance customer service dialogue model obtained by the solution.
In order to solve the above-mentioned problems, the present invention also provides an electronic apparatus including:
a memory storing at least one instruction;
the communication interface is used for realizing the communication of the electronic equipment; and the processor executes the instructions stored in the memory to realize the artificial intelligence-based insurance customer service method.
In order to solve the above-mentioned problems, the present invention further provides a computer readable storage medium, in which at least one instruction is stored, the at least one instruction being executed by a processor in an electronic device to implement the above-mentioned artificial intelligence-based insurance customer service method.
Compared with the prior art, the invention provides an artificial intelligence-based insurance customer service method, which has the following advantages:
firstly, the scheme provides a text data coding mode, which carries out vectorization representation on a coding result of a segmentation result, wherein the coding resultThe vectorized expression formula of (2) is:
；
wherein:representation->Is a vectorized representation of the results; the vectorization representation results of all word segmentation results in each sentence of text of the reference service dialogue data form a group of word vector sequences, and the dialogue word vector sequences of each group of reference service dialogue data are obtained, wherein +. >The sequence of dialogue word vectors of (a) is:
；
；
wherein:representation->Is a sequence of dialogue word vectors,/>Representation->Is a sequence of dialogue word vectors; />Representation->Middle->Word vector sequences of sentence text. Inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector and forming a coding vector sequence, wherein dialogue word vectors based on the text coding model are->The coding flow of (a) is as follows: text coding model receives the dialogue word vector +.>The method comprises the steps of carrying out a first treatment on the surface of the Coding any word vector sequence in the dialogue word vector, wherein +.>The encoding processing formula of (2) is:
；
；
wherein:weight parameters representing the coding process, +.>Representing bias parameters for the encoding process; />Representing an activation function->；/>Representing dialogue word vectorsIs a coded vector of (a); />Representing from->The useful characteristic information extracted from the data is used for the data processing,. The method and the system are used for carrying out word vector representation based on the position of each word segmentation result in the sentence in the insurance customer service reference dialogue data, wherein the position of the word segmentation result in the sentence is word vector representation weight, the weight of the word vector representation of the position in the sentence is larger, the position information capture of the word vector is realized, the self-attention weight of the position in the sentence is improved, the insurance customer service dialogue model can more effectively capture the position information in the sentence, the above effective information of each sentence word vector sequence is combined, the word vector sequence is coded and the self-attention weighting processing is carried out, so that the coded vector sequence containing the above effective information and the main information after the attention weighting is carried out is obtained, and the effective capture of the insurance customer service dialogue model on the user question and answer information is realized.
Meanwhile, the scheme provides an insurance customer service dialogue model, and an optimization objective function is determined based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model comprises an input layer, a probability calculation layer and an output layer, the input layer is used for receiving weighted coding vector sequences of user question and answer data, the probability calculation layer comprises a convolution layer and a normalization layer and is used for calculating the occurrence probability of the weighted coding vector sequences corresponding to different insurance customer service reference answer data, and the output layer selects the insurance customer service reference answer data with the highest occurrence probability to output; the optimization objective function of the insurance customer service dialogue model is as follows:
；
；
wherein:an optimized objective function representing an insurance customer service dialogue model, < >>Convolution calculation parameters representing probability calculation layers in insurance customer service dialogue model, namely weight parameters in convolution layers, +.>Representing a convolution layer; />The representation will->Input to base +.>In the insurance customer service dialogue model, the +.f. calculated by the probability calculation layer in the insurance customer service dialogue model>Sum of occurrence probabilities of->Representing normalization processing; />Representing +.>The sum of the occurrence probabilities of the weighted coded vector sequences. The proposal solves and obtains model parameters which lead the probability distribution of different insurance customer service reference answer data output by an insurance customer service dialogue model to be similar to the probability distribution of the insurance customer service reference answer data in the collected insurance customer service reference service dialogue data by taking the minimized and optimized objective function as the target, and according to the model parameters The construction is carried out to obtain an insurance customer service dialogue model, main dialogue information is automatically selected according to each sentence of question and answer data of a user, and the insurance customer service reference service dialogue with the highest occurrence probability corresponding to the insurance customer service reference service dialogue data is output, so that automatic insurance customer service is realized.
Drawings
FIG. 1 is a schematic flow chart of an artificial intelligence-based insurance customer service method according to an embodiment of the application;
FIG. 2 is a functional block diagram of an artificial intelligence based insurance customer service system according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of an electronic device for implementing an artificial intelligence-based insurance customer service method according to an embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The embodiment of the application provides an artificial intelligence-based insurance customer service method. The execution subject of the artificial intelligence-based insurance customer service method includes, but is not limited to, at least one of a server, a terminal and the like capable of being configured to execute the method provided by the embodiment of the application. In other words, the artificial intelligence-based insurance customer service method may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
Example 1
S1: and collecting insurance customer service reference service dialogue data, and carrying out vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence.
In the step S1, insurance customer service reference service dialogue data are collected, and vectorization processing is carried out on each word in the reference service dialogue data, including:
collecting N groups of insurance customer service reference service dialogue data, wherein the form of the insurance customer service reference service dialogue data is as follows:
；
wherein:
representing the n-th group insurance customer service reference service dialogue data,>user question data in the n-th group insurance customer service reference service dialogue data>Representation pair->Insurance customer service reference answer data;
each word in the reference service dialogue data is subjected to vectorization processing, wherein the vectorization processing flow of the reference service dialogue data is as follows:
s11: pre-constructing an insurance frequent word segmentation vocabulary, for any sentence text in reference service dialogue data, intercepting the first e text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, if the comparison is unsuccessful, intercepting the first e-1 text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, and so on until the comparison is successful or the text characters to be compared only remain one text character, deleting the text characters from the sentence text, if the comparison is successful, recording the text characters for comparison, deleting the text characters for comparison from the sentence text, and repeating the current steps until all the sentence text characters are split into word forms, and taking all the recorded text characters as word segmentation results; in the embodiment of the invention, the comma, the semicolon and the period are used for dividing the text sentence number of the reference service dialogue data;
Wherein the method comprises the steps ofThe word segmentation result of (2) is as follows:
；
；
wherein:
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
s12: constructing a status register with the length number of E, wherein E is the total number of common words in the insurance common word segmentation vocabulary, and a corresponding relation is established between each register position and the common words in the insurance common word segmentation vocabulary, and the initial value of each position of the status register is 0;
s13: matching each word segmentation result with a common word corresponding to each register bit in the state register, and adjusting the register bit successfully matched to 1 to obtain a coding result of the word segmentation result, wherein the word segmentation result is obtainedThe result of the encoding is->；
S14: vectorizing the coding result of the word segmentation result, wherein the coding resultVector of (3)The formulation formula is:
；
wherein:
representation->Is a vectorized representation of the results;
s15: the vectorization representation results of all word segmentation results in each sentence of text of the reference service dialogue data form a group of word vector sequences to obtain the dialogue word vector sequence of each group of reference service dialogue data, wherein The sequence of dialogue word vectors of (a) is:
；
；
wherein:
representation->Is a sequence of dialogue word vectors,/>Representation->Is a sequence of dialogue word vectors;
representation->Middle->Word vector sequences of sentence text.
S2: and inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector, and forming a coding vector sequence.
And S2, inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector, wherein the S2 comprises the following steps:
inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector and form a coding vector sequence, wherein the dialogue word vector based on the text coding modelThe coding flow of (a) is as follows:
s21: text encoding model receives dialogue word vectors；
S22: encoding arbitrary word vector sequences in dialogue word vectors, whereinThe encoding processing formula of (2) is:
；
；
wherein:
weight parameters representing the coding process, +.>Representing bias parameters for the encoding process;
representing an activation function->The method comprises the steps of carrying out a first treatment on the surface of the In the embodiment of the invention, the following steps are performedSet as ReLU function
Representing +.>Is a coded vector of (a);
representing from- >The useful characteristic information extracted from the data is used for the data processing,；
s23: constructing a dialogue word vectorIs a sequence of encoded vectors of (a): />
；
Wherein:
representing dialogue word vector +.>Is described.
S3: and carrying out self-attention weighting processing on the coded vector sequence to obtain a weighted coded vector sequence.
And in the step S3, performing self-attention weighting processing on the coded vector sequence, wherein the self-attention weighting processing comprises the following steps:
generating self-attention weight vectors for arbitrary coded vectors, wherein the coded vector sequencesThe self-attention weight vector of (2) is:
；
wherein:
representing the coded vector sequence->Is a self-attention weight vector of (2);
based on self-attention weight vectorCoding vector sequence->Performing self-attention weighting processing, wherein the self-attention weighting processing formula is as follows:
；
wherein:
representation->Is weighted by self-attention of (2)Results;
the weighted coded vector sequence is:
；
wherein:
representing the coded vector sequence->Is a self-attention weighted result of (c).
S4: the probability of each coded vector in the weighted sequence of coded vectors occurring is calculated.
The step S4 of calculating the probability of occurrence of each weighted coded vector sequence includes:
calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of user problem data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector sequence if the cosine similarity is higher than a preset similarity threshold;
Calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of insurance customer service reference answer data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector if the cosine similarity is higher than a preset similarity threshold;
calculating to obtain an arbitrary weighted coded vector sequenceAnd the weighted coded vector sequence +.>And probability of occurrence of the same weighted coded vector sequence:wherein->Representing an arbitrarily weighted coded vector sequence +.>And +.>Is a set of identical weighted coded vector sequences, is a set of weighted coded vector sequences>Representing a weighted coded vector sequence->And +.>Is used to encode the set of vector sequences.
S5: and constructing an insurance customer service dialogue model, and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model is a neural network model.
And S5, constructing an insurance customer service dialogue model, and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the method comprises the following steps:
constructing an insurance customer service dialogue model and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model comprises an input layer, a probability calculation layer and an output layer, the input layer is used for receiving weighted coding vector sequences of user question and answer data, the probability calculation layer comprises a convolution layer and a normalization layer and is used for calculating the occurrence probability of the weighted coding vector sequences corresponding to different insurance customer service reference answer data, and the output layer selects the insurance customer service reference answer data with the highest occurrence probability to output;
The optimization objective function of the insurance customer service dialogue model is as follows:
；
；
wherein:
an optimized objective function representing an insurance customer service dialogue model, < >>Convolution calculation parameters representing probability calculation layers in insurance customer service dialogue model, namely weight parameters in convolution layers, +.>Representing a convolution layer;
the representation will->Input to base +.>In the insurance customer service dialogue model, the +.f. calculated by the probability calculation layer in the insurance customer service dialogue model>Sum of occurrence probabilities of->Representing normalization processing;
representing +.>The sum of the occurrence probabilities of the weighted coded vector sequences.
And solving to obtain model parameters which enable probability distribution of different insurance customer service reference answer data output by the insurance customer service dialogue model to be similar to probability distribution of the insurance customer service reference answer data in the acquired insurance customer service reference service dialogue data by taking the minimum optimization objective function as a target.
S6: and carrying out optimization solving on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the insurance customer service dialogue model obtained by solving.
In the step S6, the optimization solution is performed on the insurance customer service dialogue model based on the optimization objective function, and the method comprises the following steps:
And carrying out optimization solution on the insurance customer service dialogue model based on an optimization objective function, wherein the optimization solution flow is as follows:
s61: setting parameters to be optimized of insurance customer service dialogue modelThe current optimization times are t, the initial value of t is 1, and the parameter to be optimized is randomly generated>And calculates the gradient of the current optimized objective function as +.>The parameter to be optimized obtained by the t-th iteration is +.>The corresponding gradient of the optimized objective function is +.>；
S62: if it isOutput +.>As parameters obtained by optimization and based on the parameters +.>Building an insurance customer service dialogue model, namely->Is a smaller positive number, wherein +.>；/>
S63: calculating to obtain iteration step length：
；
S64: if it isSatisfying the following constraint, then remain +.>Otherwise let->：
S65: pairing pairsAnd (3) performing iterative optimization:
；
order theThe process returns to step S62.
And S6, carrying out insurance customer service dialogue by using the obtained insurance customer service dialogue model, wherein the insurance customer service dialogue comprises the following steps:
and carrying out insurance customer service dialogue by utilizing the obtained insurance customer service dialogue model, wherein the dialogue flow based on the insurance customer service dialogue model is as follows:
acquiring user question-answer data, and sequentially carrying out vectorization processing, coding processing based on a text coding model and self-attention weighting processing on the user question-answer data to obtain a weighted coded vector sequence of the user question-answer data;
Transmitting the weighted coded vector sequence of the user question-answering data to an insurance customer service dialogue model;
the probability calculation layer of the insurance customer service dialogue model calculates the occurrence probability of the coded vector sequence after being weighted corresponding to different insurance customer service reference answer data, and the output layer selects the corresponding insurance customer service reference answer data with the highest occurrence probability for output.
Example 2
Fig. 2 is a functional block diagram of an artificial intelligence-based insurance customer service system according to an embodiment of the present invention, which can implement the artificial intelligence-based insurance customer service method in embodiment 1.
The artificial intelligence based insurance customer service system 100 of the present invention can be installed in an electronic device. According to the implemented functions, the artificial intelligence-based insurance customer service system may include a text encoding device 101, an encoding vector weighting module 102 and an insurance customer service dialogue module 103. The module of the invention, which may also be referred to as a unit, refers to a series of computer program segments, which are stored in the memory of the electronic device, capable of being executed by the processor of the electronic device and of performing a fixed function.
The text encoding device 101 is configured to collect insurance customer service reference service dialogue data, perform vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence, input the dialogue word vector sequence into the text encoding model, obtain an encoding vector corresponding to each dialogue word vector, and form an encoding vector sequence;
The code vector weighting module 102 is configured to perform self-attention weighting on the code vector sequence to obtain a weighted code vector sequence, and calculate a probability of occurrence of each code vector in the weighted code vector sequence;
and the insurance customer service dialogue module 103 is used for carrying out optimization solution on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the solved insurance customer service dialogue model.
In detail, the modules in the artificial intelligence-based insurance customer service system 100 in the embodiment of the present invention use the same technical means as the artificial intelligence-based insurance customer service method described in fig. 1, and can produce the same technical effects, which are not described herein.
Example 3
Fig. 3 is a schematic structural diagram of an electronic device for implementing an artificial intelligence-based insurance customer service method according to an embodiment of the present invention.
The electronic device 1 may comprise a processor 10, a memory 11, a communication interface 13 and a bus, and may further comprise a computer program, such as program 12, stored in the memory 11 and executable on the processor 10.
The memory 11 includes at least one type of readable storage medium, including flash memory, a mobile hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device 1, such as a removable hard disk of the electronic device 1. The memory 11 may in other embodiments also be an external storage device of the electronic device 1, such as a plug-in mobile 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 electronic device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device 1. The memory 11 may be used not only for storing application software installed in the electronic device 1 and various types of data, such as codes of the program 12, but also for temporarily storing data that has been output or is to be output.
The processor 10 may be comprised of integrated circuits in some embodiments, for example, a single packaged integrated circuit, or may be comprised of multiple integrated circuits packaged with the same or different functions, including one or more central processing units (Central Processing unit, CPU), microprocessors, digital processing chips, graphics processors, combinations of various control chips, and the like. The processor 10 is a Control Unit (Control Unit) of the electronic device, connects respective components of the entire electronic device using various interfaces and lines, executes or executes programs or modules (a program 12 for implementing an artificial intelligence-based insurance customer service, etc.) stored in the memory 11, and invokes data stored in the memory 11 to perform various functions of the electronic device 1 and process data.
The communication interface 13 may comprise a wired interface and/or a wireless interface (e.g. WI-FI interface, bluetooth interface, etc.), typically used to establish a communication connection between the electronic device 1 and other electronic devices and to enable connection communication between internal components of the electronic device.
The bus may be a peripheral component interconnect standard (peripheral component interconnect, PCI) bus or an extended industry standard architecture (extended industry standard architecture, EISA) bus, among others. The bus may be classified as an address bus, a data bus, a control bus, etc. The bus is arranged to enable a connection communication between the memory 11 and at least one processor 10 etc.
Fig. 3 shows only an electronic device with components, it being understood by a person skilled in the art that the structure shown in fig. 3 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or may combine certain components, or may be arranged in different components.
For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for supplying power to each component, and preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that functions of charge management, discharge management, power consumption management, and the like are implemented through the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device 1 may further include various sensors, bluetooth modules, wi-Fi modules, etc., which will not be described herein.
The electronic device 1 may optionally further comprise a user interface, which may be a Display, an input unit, such as a Keyboard (Keyboard), or a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, or the like. The display may also be referred to as a display screen or display unit, as appropriate, for displaying information processed in the electronic device 1 and for displaying a visual user interface.
It should be understood that the embodiments described are for illustrative purposes only and are not limited to this configuration in the scope of the patent application.
It should be noted that, the foregoing reference numerals of the embodiments of the present invention are merely for describing the embodiments, and do not represent the advantages and disadvantages of the embodiments. And the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method 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, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, apparatus, article or method that comprises the element.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (7)
1. An artificial intelligence-based insurance customer service method, which is characterized by comprising the following steps:
s1: collecting insurance customer service reference service dialogue data, and carrying out vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence;
s2: inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector, and forming a coding vector sequence;
s3: performing self-attention weighting on the coded vector sequence to obtain a weighted coded vector sequence;
s4: calculating the probability of each code vector in the weighted code vector sequence;
calculating the probability of occurrence of each weighted coded vector sequence, comprising:
calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of user problem data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector sequence if the cosine similarity is higher than a preset similarity threshold;
Calculating cosine similarity of the weighted code vector sequences corresponding to any two groups of insurance customer service reference answer data in the N groups of insurance customer service reference service dialogue data, and taking the two groups of weighted code vector sequences as the same weighted code vector if the cosine similarity is higher than a preset similarity threshold;
calculating to obtain an arbitrary weighted coded vector sequenceAnd the weighted coded vector sequence +.>And probability of occurrence of the same weighted coded vector sequence:wherein: />Representing the coded vector sequence->Is a self-attention weighted result of (2); />Representing the coded vector sequence->Is a self-attention weighted result of (2); />Representing an arbitrarily weighted coded vector sequence +.>Andis a set of identical weighted coded vector sequences, is a set of weighted coded vector sequences>Representing a weighted coded vector sequence->And +.>Is a set of the same weighted coded vector sequences;
s5: constructing an insurance customer service dialogue model, and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model is a neural network model;
s6: carrying out optimization solution on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the insurance customer service dialogue model obtained by the solution;
And S5, constructing an insurance customer service dialogue model, and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the method comprises the following steps:
constructing an insurance customer service dialogue model and determining an optimization objective function based on the occurrence probability of the coding vector, wherein the insurance customer service dialogue model comprises an input layer, a probability calculation layer and an output layer, the input layer is used for receiving weighted coding vector sequences of user question and answer data, the probability calculation layer comprises a convolution layer and a normalization layer and is used for calculating the occurrence probability of the weighted coding vector sequences corresponding to different insurance customer service reference answer data, and the output layer selects the insurance customer service reference answer data with the highest occurrence probability to output;
the optimization objective function of the insurance customer service dialogue model is as follows:
；
；
wherein:
an optimized objective function representing an insurance customer service dialogue model, < >>Convolution calculation parameters representing probability calculation layers in insurance customer service dialogue model, namely weight parameters in convolution layers, +.>Representing a convolution layer;
the representation will->Input to base +.>In the insurance customer service dialogue model, the +.f. calculated by the probability calculation layer in the insurance customer service dialogue model>Sum of occurrence probabilities of->Representing normalization processing;
Representing +.>The sum of the occurrence probabilities of the weighted coded vector sequences.
2. The method for providing security services based on artificial intelligence as claimed in claim 1, wherein the step S1 of collecting security service reference service dialogue data and vectorizing each word in the reference service dialogue data comprises the steps of:
collecting N groups of insurance customer service reference service dialogue data, wherein the form of the insurance customer service reference service dialogue data is as follows:
；
wherein:
representing the n-th group insurance customer service reference service dialogue data,>user question data in the n-th group insurance customer service reference service dialogue data>Representation pair->Insurance customer service reference answer data;
each word in the reference service dialogue data is subjected to vectorization processing, wherein the vectorization processing flow of the reference service dialogue data is as follows:
s11: pre-constructing an insurance frequent word segmentation vocabulary, for any sentence text in reference service dialogue data, intercepting the first e text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, if the comparison is unsuccessful, intercepting the first e-1 text characters to be compared with the frequent words in the insurance frequent word segmentation vocabulary, and so on until the comparison is successful or the text characters to be compared only remain one text character, deleting the text characters from the sentence text, if the comparison is successful, recording the text characters for comparison, deleting the text characters for comparison from the sentence text, and repeating the current steps until all the sentence text characters are split into word forms, and taking all the recorded text characters as word segmentation results;
Wherein the method comprises the steps ofThe word segmentation result of (2) is as follows:
；
；
wherein:
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
representing +.>Middle->First->Personal word (s)/(s)>Representation->Text sentence number->Representation->Middle->Word segmentation word number of sentence text;
s12: constructing a status register with the length number of E, wherein E is the total number of common words in the insurance common word segmentation vocabulary, and a corresponding relation is established between each register position and the common words in the insurance common word segmentation vocabulary, and the initial value of each position of the status register is 0;
s13: matching each word segmentation result with a common word corresponding to each register bit in the state register, and adjusting the register bit successfully matched to 1 to obtain a coding result of the word segmentation result, wherein the word segmentation result is obtainedThe result of the encoding is->；
S14: vectorizing the coding result of the word segmentation result, wherein the coding resultThe vectorized expression formula of (2) is:
；
wherein:
representation->Is a vectorized representation of the results;
s15: the vectorization representation results of all word segmentation results in each sentence of text of the reference service dialogue data form a group of word vector sequences to obtain the dialogue word vector sequence of each group of reference service dialogue data, wherein The sequence of dialogue word vectors of (a) is:
；
；
wherein:
representation->Is a sequence of dialogue word vectors,/>Representation->Is a sequence of dialogue word vectors;
representation->Middle->Word vector sequences of sentence text.
3. The method for providing an artificial intelligence-based insurance customer service according to claim 2, wherein the step S2 of inputting the sequence of dialogue word vectors into a text coding model to obtain a coding vector corresponding to each dialogue word vector comprises:
inputting the dialogue word vector sequence into a text coding model to obtain a coding vector corresponding to each dialogue word vector and form a coding vector sequence, wherein the dialogue word vector based on the text coding modelThe coding flow of (a) is as follows:
s21: text encoding model receives dialogue word vectors；
S22: encoding arbitrary word vector sequences in dialogue word vectors, whereinThe encoding processing formula of (2) is:
；
；
wherein:
weight parameters representing the coding process, +.>Representing bias parameters for the encoding process;
representing an activation function->；
Representing +.>Is a coded vector of (a);
representing from->Useful characteristic information extracted from the information;
s23: constructing a dialogue word vectorIs a sequence of encoded vectors of (a):
；
Wherein:
representing dialogue word vector +.>Is described.
4. An artificial intelligence based insurance customer service method as defined in claim 3, wherein the self-attention weighting process is performed on the coded vector sequence in step S3, including:
generating self-attention weight vectors for arbitrary coded vectors, wherein the coded vector sequencesThe self-attention weight vector of (2) is:
；
wherein:
representing the coded vector sequence->Is a self-attention weight vector of (2);
based on self-attention weight vectorCoding vector sequence->Performing self-attention weighting processing, wherein the self-attention weighting processing formula is as follows:
；
wherein:
representation->Is a self-attention weighted result of (2);
the weighted coded vector sequence is:
；
wherein:
representing the coded vector sequence->Is a self-attention weighted result of (c).
5. The artificial intelligence-based insurance customer service method as set forth in claim 1, wherein the optimizing the solution of the insurance customer service dialogue model based on the optimizing objective function in step S6 includes:
and carrying out optimization solution on the insurance customer service dialogue model based on an optimization objective function, wherein the optimization solution flow is as follows:
s61: setting parameters to be optimized of insurance customer service dialogue model The current optimization times are t, the initial value of t is 1, and the parameter to be optimized is randomly generated>And calculates the gradient of the current optimized objective function as +.>The parameter to be optimized obtained by the t-th iteration is +.>The corresponding gradient of the optimized objective function is +.>；
S62: if it isOutput +.>As parameters obtained by optimization and based on the parameters +.>Building an insurance customer service dialogue model, namely->Is a smaller positive number, wherein +.>；
S63: calculating to obtain iteration step length：
S64: if it isSatisfying the following constraint, then remain +.>Otherwise let->：
；
S65: for a pair ofAnd (3) performing iterative optimization:
；
order theThe process returns to step S62.
6. The artificial intelligence-based insurance customer service method as set forth in claim 5, wherein the step S6 of performing insurance customer service session using the solved insurance customer service session model includes:
and carrying out insurance customer service dialogue by utilizing the obtained insurance customer service dialogue model, wherein the dialogue flow based on the insurance customer service dialogue model is as follows:
acquiring user question-answer data, and sequentially carrying out vectorization processing, coding processing based on a text coding model and self-attention weighting processing on the user question-answer data to obtain a weighted coded vector sequence of the user question-answer data;
Transmitting the weighted coded vector sequence of the user question-answering data to an insurance customer service dialogue model;
the probability calculation layer of the insurance customer service dialogue model calculates the occurrence probability of the coded vector sequence after being weighted corresponding to different insurance customer service reference answer data, and the output layer selects the corresponding insurance customer service reference answer data with the highest occurrence probability for output.
7. An artificial intelligence based insurance customer service system, the system comprising:
the text coding device is used for collecting the insurance customer service reference service dialogue data, carrying out vectorization processing on each word in the reference service dialogue data to form a dialogue word vector sequence, inputting the dialogue word vector sequence into the text coding model to obtain a coding vector corresponding to each dialogue word vector, and forming a coding vector sequence;
the coding vector weighting module is used for carrying out self-attention weighting on the coding vector sequence to obtain a weighted coding vector sequence, and calculating the probability of each coding vector in the weighted coding vector sequence;
the insurance customer service dialogue module is used for carrying out optimization solving on the insurance customer service dialogue model based on the optimization objective function, and carrying out insurance customer service dialogue by utilizing the insurance customer service dialogue model obtained by solving so as to realize the insurance customer service method based on artificial intelligence as claimed in any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310857578.8A CN116578690B (en) | 2023-07-13 | 2023-07-13 | Insurance customer service method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310857578.8A CN116578690B (en) | 2023-07-13 | 2023-07-13 | Insurance customer service method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116578690A CN116578690A (en) | 2023-08-11 |
| CN116578690B true CN116578690B (en) | 2023-09-22 |
Family
ID=87539999
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310857578.8A Active CN116578690B (en) | 2023-07-13 | 2023-07-13 | Insurance customer service method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116578690B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674271A (en) * | 2019-08-27 | 2020-01-10 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device |
| CN112749262A (en) * | 2020-07-24 | 2021-05-04 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device based on artificial intelligence, electronic equipment and storage medium |
| US11132988B1 (en) * | 2020-10-22 | 2021-09-28 | PolyAI Limited | Dialogue system, a dialogue method, and a method of training |
| WO2021218029A1 (en) * | 2020-04-26 | 2021-11-04 | 平安科技（深圳）有限公司 | Artificial intelligence-based interview method and apparatus, computer device, and storage medium |
| CN113704424A (en) * | 2021-08-27 | 2021-11-26 | 广东电力信息科技有限公司 | Natural language task generation method based on pointer network |
| WO2022036616A1 (en) * | 2020-08-20 | 2022-02-24 | 中山大学 | Method and apparatus for generating inferential question on basis of low labeled resource |
| CN114386426A (en) * | 2021-11-25 | 2022-04-22 | 北京容联易通信息技术有限公司 | Gold medal speaking skill recommendation method and device based on multivariate semantic fusion |
| CN114936274A (en) * | 2022-05-11 | 2022-08-23 | 平安科技（深圳）有限公司 | Model training method, dialogue generating device, dialogue training equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109002515A (en) * | 2018-07-04 | 2018-12-14 | 网宿科技股份有限公司 | A kind of method and apparatus of intelligent response |
| US11288324B2 (en) * | 2020-01-22 | 2022-03-29 | Adobe Inc. | Chart question answering |
-
2023
- 2023-07-13 CN CN202310857578.8A patent/CN116578690B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674271A (en) * | 2019-08-27 | 2020-01-10 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device |
| WO2021218029A1 (en) * | 2020-04-26 | 2021-11-04 | 平安科技（深圳）有限公司 | Artificial intelligence-based interview method and apparatus, computer device, and storage medium |
| CN112749262A (en) * | 2020-07-24 | 2021-05-04 | 腾讯科技（深圳）有限公司 | Question and answer processing method and device based on artificial intelligence, electronic equipment and storage medium |
| WO2022036616A1 (en) * | 2020-08-20 | 2022-02-24 | 中山大学 | Method and apparatus for generating inferential question on basis of low labeled resource |
| US11132988B1 (en) * | 2020-10-22 | 2021-09-28 | PolyAI Limited | Dialogue system, a dialogue method, and a method of training |
| CN113704424A (en) * | 2021-08-27 | 2021-11-26 | 广东电力信息科技有限公司 | Natural language task generation method based on pointer network |
| CN114386426A (en) * | 2021-11-25 | 2022-04-22 | 北京容联易通信息技术有限公司 | Gold medal speaking skill recommendation method and device based on multivariate semantic fusion |
| CN114936274A (en) * | 2022-05-11 | 2022-08-23 | 平安科技（深圳）有限公司 | Model training method, dialogue generating device, dialogue training equipment and storage medium |
Non-Patent Citations (5)
| Title |
|---|
| A diversity-promoting objective function for neural conversation models;Li Jiwei等;arXiv preprint arXiv:1510.03055;全文 * |
| T-Reader:一种基于自注意力机制的多任务深度阅读理解模型;郑玉昆;李丹;范臻;刘奕群;张敏;马少平;;中文信息学报(11);全文 * |
| 一种基于多粒度循环神经网络与词注意力的多轮对话回答选择方法;小型微型计算机系统;第42卷(第12期);第2553-2560页 * |
| 基于深度学习的回复类型预测聊天机器人;徐畅;周志平;赵卫东;;计算机应用研究(S1);全文 * |
| 智能对话系统研究综述;贾熹滨;李让;胡长建;陈军成;;北京工业大学学报(09);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116578690A (en) | 2023-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113378970B (en) | Sentence similarity detection method and device, electronic equipment and storage medium | |
| CN112988963B (en) | User intention prediction method, device, equipment and medium based on multi-flow nodes | |
| WO2020073533A1 (en) | Automatic question answering method and device | |
| CN113821622B (en) | Answer retrieval method and device based on artificial intelligence, electronic equipment and medium | |
| JP2022169743A (en) | Information extraction method and device, electronic equipment, and storage medium | |
| CN116821373A (en) | Map-based prompt recommendation method, device, equipment and medium | |
| CN115221276A (en) | Chinese image-text retrieval model training method, device, equipment and medium based on CLIP | |
| CN116701574A (en) | Text semantic similarity calculation method, device, equipment and storage medium | |
| CN114358023A (en) | Intelligent question-answer recall method and device, computer equipment and storage medium | |
| CN113918704A (en) | Question-answering method and device based on machine learning, electronic equipment and medium | |
| CN116341646A (en) | Pretraining method and device of Bert model, electronic equipment and storage medium | |
| CN114548114B (en) | Text emotion recognition method, device, equipment and storage medium | |
| CN116578690B (en) | Insurance customer service method and system based on artificial intelligence | |
| CN116450829A (en) | Medical text classification method, device, equipment and medium | |
| CN116340516A (en) | Entity relation cluster extraction method, device, equipment and storage medium | |
| CN116383766A (en) | Auxiliary diagnosis method, device, equipment and storage medium based on multi-mode data | |
| CN113469237B (en) | User intention recognition method, device, electronic equipment and storage medium | |
| CN114676307A (en) | Ranking model training method, device, equipment and medium based on user retrieval | |
| CN115840867A (en) | Generation method and device of mathematical problem solving model, electronic equipment and storage medium | |
| CN115238683B (en) | Method, device, equipment and medium for recognizing stop words of circulating self-attention | |
| CN114462411B (en) | Named entity recognition method, device, equipment and storage medium | |
| CN113656703B (en) | Intelligent recommendation method, device, equipment and storage medium based on new online courses | |
| CN117689117B (en) | Intelligent chemical industry planning consultation method and system | |
| CN111680513B (en) | Feature information identification method and device and computer readable storage medium | |
| CN116501838A (en) | Text recall method, device, electronic equipment and computer readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |