CN112699233A - Service processing method and device and electronic equipment - Google Patents
Service processing method and device and electronic equipment Download PDFInfo
- Publication number
- CN112699233A CN112699233A CN201910990265.3A CN201910990265A CN112699233A CN 112699233 A CN112699233 A CN 112699233A CN 201910990265 A CN201910990265 A CN 201910990265A CN 112699233 A CN112699233 A CN 112699233A
- Authority
- CN
- China
- Prior art keywords
- text
- recognition result
- information
- sample data
- consultation
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 103
- 230000004044 response Effects 0.000 claims abstract description 65
- 238000012545 processing Methods 0.000 claims abstract description 45
- 238000000034 method Methods 0.000 claims abstract description 38
- 238000003062 neural network model Methods 0.000 claims abstract description 31
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 63
- 230000011218 segmentation Effects 0.000 claims description 22
- 238000011176 pooling Methods 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 19
- 238000004891 communication Methods 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000003860 storage Methods 0.000 claims description 5
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000004927 fusion Effects 0.000 description 3
- 210000004205 output neuron Anatomy 0.000 description 3
- 230000002411 adverse Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000009223 counseling Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011022 operating instruction Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
Images
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
- G06F16/35—Clustering; Classification
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Machine Translation (AREA)
Abstract
The embodiment of the invention relates to the technical field of artificial intelligence, and discloses a service processing method, a service processing device and electronic equipment. The method comprises the following steps: training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order. Through the mode, the embodiment of the invention realizes the automatic analysis and response of the consultation contents, and greatly improves the question-answering precision of the customer service system and the overall efficiency of the customer service flow.
Description
Technical Field
The embodiment of the invention relates to the technical field of artificial intelligence, in particular to a service processing method and device and electronic equipment.
Background
In a large-scale business system such as a communication and e-commerce platform, a customer service (hereinafter, referred to as "customer service") provided for responding to a user's inquiry or question has a very significant influence and significance on the user experience.
Conventional customer service is typically implemented by establishing a call center (CallCenter). The call center needs to be equipped with special software and hardware systems and manual customer service personnel, and has high cost and poor expansibility.
However, with the development and popularization of the internet in recent years, a new generation of customer service system established by relying on the internet technology is applied in a plurality of fields, and has the advantages of low cost, all-weather and the like. However, the existing customer service system based on the internet technology has weak environmental adaptability, and in practical application, many limitations exist, so that the customer service quality cannot be effectively improved.
Disclosure of Invention
In view of the foregoing problems, embodiments of the present invention provide a service processing method, an apparatus, and an electronic device, which overcome the foregoing problems or at least partially solve the foregoing problems.
According to an aspect of an embodiment of the present invention, a method for processing a service is provided, where the method includes:
training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
In an optional mode, the response information comprises knowledge information matched with the text recognition result and notice information;
the step of outputting response information matching the text recognition result when there is response information matching the text recognition result, further comprising:
the step of outputting response information matching the text recognition result when there is response information matching the text recognition result, further comprising:
if the knowledge information matched with the text recognition result exists, outputting the knowledge information;
and if the announcement information matched with the text recognition result exists, outputting the announcement information.
In an optional manner, the step of obtaining the advisory information further includes:
acquiring the consultation content of the user in real time; and performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information.
In an optional manner, the step of outputting the knowledge information further includes: and displaying the knowledge information through a site established by a fifth generation hypertext markup language.
In an optional manner, the step of performing training of a neural network model through sample data to obtain a text classifier for automatically recognizing an input text further includes:
generating training sample data; extracting the feature vectors of the training sample data by the convolutional layer and outputting a plurality of feature graphs; extracting the maximum value in each feature map through a pooling layer to serve as an output feature vector; splicing each output feature vector to obtain a text feature vector corresponding to the training sample data; determining a classification label of the training sample data according to the text feature vector; and carrying out reverse training on the classification label of the training sample data, and optimizing the parameters of the convolutional layer and the pooling layer.
In an optional manner, the step of generating training sample data further includes: collecting the existing consultation content; performing word segmentation processing on the consulting content through a preset word segmentation dictionary to form a set consisting of a plurality of words; and coding the consultation data to obtain corresponding coded data serving as the training sample data.
In an optional manner, the extracting, by the convolutional layer, the feature vectors of the training sample data to form a plurality of feature maps, further includes:
converting each word in the training sample data into a word vector with the same length, and generating a matrix corresponding to the consultation content; accessing the matrix into a convolution layer, calculating the convolution of n word vectors each time, and outputting a plurality of characteristic graphs; each of the feature maps contains a plurality of feature vectors.
According to another aspect of the embodiments of the present invention, there is provided a service processing apparatus, including:
the off-line training module is used for training the neural network model through the sample data to obtain a text classifier for automatically identifying the input text;
the information acquisition module is used for acquiring the consultation information;
the text analysis module is used for determining a text recognition result corresponding to the consultation information through the text classifier;
the feedback module is used for outputting response information matched with the text recognition result when the response information matched with the text recognition result exists; and displaying a guidance page for creating the customer service order when there is no response information matching the text recognition result.
According to another aspect of the embodiments of the present invention, there is provided an electronic device for service processing, including: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is configured to store at least one executable instruction that causes the processor to:
training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
According to yet another aspect of the embodiments of the present invention, there is provided a computer storage medium having at least one executable instruction stored therein, the executable instruction causing the processor to:
training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
The embodiment of the invention analyzes and processes the consultation contents from the client through the neural network model, and automatically feeds back the corresponding response contents according to the analysis result, thereby greatly improving the question-answering precision of the customer service system and the overall efficiency of the customer service process. Moreover, a fusion interaction mode is used during feedback, so that passive question answering can be supported, flexible and active guidance can be performed, consultation activities of users can be effectively shunted, the submitting processing amount of customer service work orders is reduced, and the service processing efficiency is improved.
The foregoing description is only an overview of the technical solutions of the embodiments of the present invention, and the embodiments of the present invention can be implemented according to the content of the description in order to make the technical means of the embodiments of the present invention more clearly understood, and the detailed description of the present invention is provided below in order to make the foregoing and other objects, features, and advantages of the embodiments of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 shows a flow chart of a service processing method provided by an embodiment of the present invention;
FIG. 2 is a flow chart of an offline training method provided by an embodiment of the present invention;
FIG. 3 illustrates a flow chart of outputting reply messages provided by an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a neural mesh model provided by an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a service processing apparatus provided in an embodiment of the present invention;
FIG. 6 is a schematic structural diagram of an offline training module according to an embodiment of the present invention;
fig. 7 shows a schematic structural diagram of an electronic device for business processing according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
Currently, a customer service system established by means of internet technology may include the following:
one is client-based customer service software: the customer service client is installed on the user terminal, interaction between the user and the customer service staff is achieved, and the user can directly report the consultation content to the customer service staff through the client. The customer service staff can receive the user consultation in real time and inform and feed back the result of the user consultation through the client.
Secondly, customer service processing system based on work order: the user submits the consultation work order in the work order system to realize the consultation submission, acceptance and feedback of the user. The user needs to fill in necessary field information in the work order for customer service personnel to analyze and process.
Thirdly, the customer service system based on the chat group in the IM: by establishing forms of QQ, WeChat, Fei letter chat groups and the like, users and customer service personnel are integrated into the same chat group. The user can directly provide consultation for the customer service, and the customer service staff can directly feed back and respond in the chat group. At the same time, other users in the chat group can also share the information sent to the chat group.
For the first scheme, an additional client needs to be installed, a certain requirement is provided for the computer configuration and installation compatibility of the customer service side, and the core service plug-in is easily affected.
With the second scheme, in the case that all consultation is in asynchronous mode, part of common, simple and regular consultation problems also need to be queued for solution. Not only the waiting time of the user is wasted, but also the processing time and the human resources of the customer service personnel are wasted.
For the third scenario, this is done within the chat group. Therefore, when the user's speech volume increases, the customer service person may have a wrong answer or missed a question, which is difficult to handle when sensitive information (such as personal privacy information) is involved. Moreover, the content structure of the chat information is dispersed, and obstacles exist to subsequent analysis, which are not beneficial to improving the customer service efficiency. In order to solve the above problem, embodiments of the present invention provide a service processing method and apparatus, and an electronic device.
Fig. 1 shows a flowchart of an embodiment of a neural network model (CNN) -based service processing method of the present invention, which is applied to an electronic computer device of a customer service system. The customer service system is an integrated system used for collecting feedback of users on service use and improving use experience of the users according to the feedback of the users. The system can be built on any suitable type of electronic hardware equipment (such as a server, a workstation and the like) with enough logical operation capacity, and a series of processes of data acquisition, data processing, data output and the like are realized. As shown in fig. 1, the method comprises the steps of:
step 110: and training a neural network model through the sample data to obtain a text classifier for automatically identifying the input text.
Before the neural network model (such as the convolutional neural network model CNN) is used, the existing sample data needs to be trained to obtain a final usable analysis model, which may also be referred to as an "offline training process".
In the present embodiment, the term "text classifier" is used to describe the neural network model that can be used after completion. The text classifier can analyze and process the input text content, output corresponding category labels and determine the category to which the input text content belongs.
The specific training process can be determined according to the needs of the actual situation (such as the specific model structure used). Different training processes can be used, and only a neural network model which meets the target function needs to be obtained through training.
Fig. 2 is a flowchart of an off-line training method according to an embodiment of the present invention. As shown in fig. 2, the method comprises the following steps:
step 111: and generating training sample data.
The training sample data may be from the existing consultation items in the customer service system, or may be some template data preset by a technician.
Specifically, the step of generating training sample data may include:
first, the existing counseling contents are collected. Then, the consulting content is subjected to word segmentation processing through a preset word segmentation dictionary to form a set consisting of a plurality of words.
The word segmentation processing means that the existing knowledge dictionary is used for splitting the text of the consultation content into a set consisting of a plurality of words.
The "dictionary" is created based on expert knowledge or the like, and records a data set of words constituting text contents. Because the specialty of the consulting content of the customer service system is strong, in order to improve the accuracy of the word segmentation result, the dictionary can be optimized through various different optimization modes.
And finally, coding the consultation data to obtain corresponding coded data serving as the training sample data. And the data is encoded, so that the neural network model can be conveniently trained.
Step 112: and extracting the feature vectors of the training sample data by the convolutional layer and outputting a plurality of feature graphs.
The convolution layer is a processing layer for local perception, and a convolution kernel (filter) with a proper size is selected to extract a feature vector in training sample data to form a corresponding feature map.
The feature map is a feature set composed of a plurality of feature vectors, and includes the plurality of feature vectors and relationships between the feature vectors.
Step 113: and extracting the maximum value in each feature map through the pooling layer to serve as an output feature vector.
The pooling layer (Pooling) is a functional unit used to perform downsampling, which can reduce the dimensionality of the feature map to avoid over-focusing on local details. In this embodiment, the maximum value may be output using a max pooling (maxporoling) approach.
Step 114: and splicing each output feature vector to obtain a text feature vector corresponding to the training sample data.
And integrating a plurality of output feature vectors into one text feature vector by means of simple longitudinal splicing and the like, wherein the text feature vector is used for representing the training sample data.
Step 115: and determining the classification label of the training sample data according to the text feature vector.
And determining the classification label of the training sample data (namely the classification result of the text) according to the text feature vector by combining a set classification system. The specific conversion mode from the text feature vector to the classification label can be set according to the needs of actual situations.
For example, the Softmax function may be selected to convert the text feature vector into a probability of belonging to each class, and thereby determine the class label of the training sample data.
Step 116: and carrying out reverse training on the classification label of the training sample data, and optimizing the parameters of the convolutional layer and the pooling layer.
In addition to the training process from step 111 to step 115, obtaining the classification label of the training sample data, it is further necessary to optimize and adjust the parameters of the model by means of reverse training. After repeating the training and optimizing process for many times, an ideal neural network model can be finally obtained.
Step 120: and acquiring the consultation information.
The consultation information can be data information corresponding to the consultation content of the user acquired by the customer service system. To facilitate text data processing and analysis. The originally acquired advisory content may be processed in one or more preprocessing manners to obtain advisory information having a target data structure.
In some embodiments, step 120 may be specifically implemented as follows: first, the counseling content of the user is collected in real time. And then, performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information. The preset dictionary can adopt a dictionary the same as that of the offline training process, and only the same word segmentation effect can be realized.
Step 130: and determining a text recognition result corresponding to the consultation information through the text classifier.
On the basis of the trained text classifier, the consulting content of the customer service system can be identified and analyzed on line, and a corresponding text identification result is output. In this embodiment, based on the text classifier obtained by the offline training, the finally output text recognition result may be embodied as a text label of the advisory information.
Step 140: and outputting response information matched with the text recognition result when the response information matched with the text recognition result exists.
"response message" refers to a formatted or templated message content corresponding to the text label, which is set in advance. The response information can be fed back to the user to solve the consultation of the user, and the effect of automatic response is realized.
In some embodiments, the response information may include: knowledge information and announcement information. The knowledge information refers to the explanation or description about a specific question, which is stored in the system in advance, for example, the tariff information of the current account, the package status of the current account, and the like.
The announcement information is information content with certain timeliness, and usually has a specific coverage area or range, for example, a line repair announcement of a specific area.
Correspondingly, as shown in fig. 3, the step 140 may employ the following multiple determination steps to determine what should be presented or fed back to the user:
step 141: and judging whether knowledge information matched with the text recognition result exists or not. If yes, go to step 142, otherwise go to step 143.
As described in the above embodiments, the text recognition result may be embodied as a text label. The text label can be searched in the existing database to determine whether the database has knowledge information corresponding to the label.
Step 142: and outputting the knowledge information.
There are also many ways to output the knowledge information, including but not limited to a client or a web page, etc., based on different implementation forms of the customer service system. In some embodiments, the knowledge information may be presented via a site created in a fifth generation hypertext markup language.
The fifth generation hypertext markup language (HTML5) is a markup writing language for forming a webpage, a site realized based on the HTML5 standard has good migration performance, can be conveniently integrated into the existing system, can also be conveniently integrated into mobile application, and improves the convenience and centralization in the use of a customer service system.
Step 143: and judging whether the announcement information matched with the text recognition result exists or not. If yes, go to step 144, otherwise go to step 150.
The announcement information may be stored in another database or another part of the database. And after the matched knowledge information is not searched, the search is continued in the database related to the notice information.
Step 144: and outputting the notice information.
The announcement information may also be presented to the user in any suitable way, and the specific presentation way may be various and set by the technician according to the needs of the actual situation.
Step 150: and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
The customer service work order refers to an additionally established customer service task needing to answer a response. The customer service work order can contain a series of information fields, so that the customer service staff can clearly determine the questions and requirements of the user, quickly respond, and process the customer service work order.
The guide page may be presented in any suitable form. A number of different fields that can be filled in or selected can be provided on the guidance page to guide or prompt the user to accurately describe his or her query or consulting content.
Step 150 may actually be understood as a step of switching to manual processing. That is, when the analysis result of the text classifier fails to find the matching answer information in the database, the manual answer may be switched to the customer service person.
The mode combines the existing production guarantee flow, integrates the steps of self-service question answering, information announcement, independence and simplicity and the like, and can realize seamless switching between automatic answering and manual service.
The business processing method of the embodiment of the invention realizes intelligent analysis and identification of the consultation contents based on the neural network model, can automatically feed back corresponding response information to most of the consultation contents, and effectively improves the efficiency of the customer service system.
Furthermore, two modes of passive response and active guidance are integrated, the problems of simple question and answer process and insufficient flexibility of a traditional automatic response customer service system are solved, the submission amount of the customer service work order can be reduced through the distribution of the consultation contents, and the adverse effect on the service processing effect of the customer service system can be avoided while the number of the customer service work order processing personnel is reduced.
In addition, the consultation process of the whole user is brought into the system for centralized management and control, data can be analyzed in a centralized manner, and the later-stage service level of the customer service system is favorably improved.
FIG. 4 shows a schematic diagram of an embodiment of the neural network model of the present invention. The neural network model can be applied to the business processing method of the embodiment, so that analysis and processing of the consultation information are realized, and the text label of the consultation information is obtained. As shown in FIG. 4, the neural network model may employ an algorithm called TextCNN, which includes an input layer 410, a convolutional layer 420, a pooling layer 430, and a fully-connected layer 440.
The input consulting information is a set formed by a plurality of words after word segmentation processing. The input layer 410 converts each word therein into a vector having the same length, thereby converting the text into a matrix form (hereinafter, simply referred to as a text matrix).
The text matrix is accessed into the convolutional layer 420, and the convolutional layer 420 realizes local perception of the text matrix through a convolution kernel (filter) of a specific size. The width of the convolution kernel used is equal to the length of the word vector, and the length n of the convolution kernel can be set as required.
Thus, when calculating the convolution of n word vectors, order information between the words is taken into account, similar to the n-gram model. In such a convolution mode, not only the self word sense of each word is considered, but also the word order and the context relationship among the words are included, and a more accurate analysis result is obtained.
The output of convolutional layer 420 is coupled to pooling layer 430. Convolutional layer 420 outputs a feature map comprising a plurality of feature vectors. The pooling layer 430 may extract the maximum value in the feature map as an output through a strategy of maximum pooling.
The output of pooling layer 430 continues to be connected to full connection layer 440. The fully-connected layer 440 is eventually provided with a plurality of output neurons, combined with the classification system (e.g., n output neurons are provided when there are n classifications in the classification system). The output of each output neuron can be converted into the probability of belonging to each classification through a Softmax function, so that the text label corresponding to the consultation information is determined.
By the neural network model obtained based on the customer service system training, the knowledge graph is created, the question and answer precision of the customer service system is greatly improved, and the overall efficiency of the customer service process is improved. Accordingly, the method can improve the user consultation efficiency, improve the consultation reply quality, reduce the workload of customer service staff and improve the overall level of supporting customer service of operation businesses by means of accurate information pushing, automatic user guiding, intelligent recommendation and the like.
Fig. 5 is a schematic structural diagram of an embodiment of the neural network model-based service processing device of the present invention. As shown in fig. 5, the apparatus 500 includes: an offline training module 510, an information collection module 520, a text analysis module 530, and a feedback module 540.
The offline training module 510 is configured to perform training of a neural network model through sample data to obtain a text classifier for automatically identifying an input text. The information collecting module 520 is used for obtaining the consulting information. The text analysis module 530 is configured to determine a text recognition result corresponding to the advisory information through the text classifier. The feedback module 540 is configured to output response information matching the text recognition result when there is response information matching the text recognition result; and displaying a guidance page for creating the customer service order when there is no response information matching the text recognition result.
In an alternative approach, as shown in fig. 6, the offline training module 510 may include: a training sample data generating unit 511, a model training unit 512, and a training optimizing unit 513.
The training sample data generating unit 511 is configured to generate training sample data. The model training unit 512 is configured to extract feature vectors of the training sample data from the convolutional layer, and output a plurality of feature maps; extracting the maximum value in each feature map through a pooling layer to serve as an output feature vector; and splicing each output feature vector to obtain a text feature vector corresponding to the training sample data, and determining the classification label of the training sample data according to the text feature vector. The training optimization unit 513 is configured to perform reverse training on the classification label of the training sample data, and optimize parameters of the convolutional layer and the pooling layer.
In an optional manner, the training sample data generating unit 511 is further configured to collect the existing consulting content; performing word segmentation processing on the consulting content through a preset word segmentation dictionary to form a set consisting of a plurality of words; and coding the consultation data to obtain corresponding coded data serving as the training sample data.
In an optional manner, the model training unit 512 is further configured to convert each word in the training sample data into a word vector with the same length, and generate a matrix corresponding to the consulting content; accessing the matrix into a convolution layer, calculating the convolution of n word vectors each time, and outputting a plurality of characteristic graphs; each of the feature maps contains a plurality of feature vectors.
In an alternative mode, the response information includes knowledge information and notice information that match the text recognition result. Correspondingly, the feedback module 540 is further configured to: judging whether knowledge information matched with the text recognition result exists or not; if yes, outputting the knowledge information; if not, judging whether notice information matched with the text recognition result exists or not; and if so, outputting the announcement information.
In an alternative manner, the feedback module 540 displays the response information such as knowledge information through a website established by a fifth generation hypertext markup language. Therefore, the service processing device can be conveniently integrated into the existing system and can also be conveniently integrated into mobile application, and the convenience and centralization in use of the customer service system are improved.
In an optional manner, the information collecting module 520 is further configured to collect the consulting content of the user in real time; and performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information.
The business processing device of the embodiment of the invention realizes intelligent analysis on the consultation content and automatically feeds back the corresponding response content through the neural network model. The question and answer precision of the customer service system can be greatly improved, and the overall efficiency of the customer service flow is improved. Moreover, a fusion interaction mode is used during feedback, so that passive question answering can be supported, flexible and active guidance can be performed, the consultation activities of the user can be effectively shunted, the work submission processing amount is reduced, and the customer service efficiency is improved.
The embodiment of the invention provides a nonvolatile computer storage medium, wherein at least one executable instruction is stored in the computer storage medium, and the computer executable instruction can execute the service processing method in any method embodiment.
The executable instructions may be specifically configured to cause the processor to: training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
In an alternative mode, the response information includes knowledge information and notice information that match the text recognition result. The step of outputting response information matching the text recognition result when there is response information matching the text recognition result, further comprising:
judging whether knowledge information matched with the text recognition result exists or not; if yes, outputting the knowledge information; if not, judging whether notice information matched with the text recognition result exists or not; and if so, outputting the announcement information.
In an optional manner, the step of obtaining the advisory information further includes: acquiring the consultation content of the user in real time; and performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information.
In an optional manner, the step of outputting the knowledge information further includes: and displaying the knowledge information through a site established by a fifth generation hypertext markup language.
In an optional manner, the step of performing training of a neural network model through sample data to obtain a text classifier for automatically recognizing an input text further includes:
generating training sample data; extracting the feature vectors of the training sample data by the convolutional layer and outputting a plurality of feature graphs; extracting the maximum value in each feature map through a pooling layer to serve as an output feature vector; splicing each output feature vector to obtain a text feature vector corresponding to the training sample data; determining a classification label of the training sample data according to the text feature vector; and carrying out reverse training on the classification label of the training sample data, and optimizing the parameters of the convolutional layer and the pooling layer.
In an optional manner, the step of generating training sample data further includes: collecting the existing consultation content; performing word segmentation processing on the consulting content through a preset word segmentation dictionary to form a set consisting of a plurality of words; and coding the consultation data to obtain corresponding coded data serving as the training sample data.
In an optional manner, the extracting, by the convolutional layer, the feature vectors of the training sample data to form a plurality of feature maps, further includes:
converting each word in the training sample data into a word vector with the same length, and generating a matrix corresponding to the consultation content; accessing the matrix into a convolution layer, calculating the convolution of n word vectors each time, and outputting a plurality of characteristic graphs; each of the feature maps contains a plurality of feature vectors.
When the nonvolatile computer storage medium provided by the embodiment of the invention is applied to a customer service system, the consulting content of the user is intelligently analyzed and the corresponding response content is automatically fed back through the neural network model. The question and answer precision of the customer service system and the overall efficiency of the customer service flow can be greatly improved.
Moreover, a fusion interaction mode is used during feedback, so that passive question answering can be supported, flexible and active guidance can be performed, consultation activities of users can be effectively shunted, the submitting processing amount of the customer service work order is reduced, and the customer service efficiency is improved.
Fig. 7 is a schematic structural diagram of an embodiment of an electronic device for service processing according to the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the electronic device.
As shown in fig. 7, the electronic device for business processing may include: a processor (processor)702, a Communications Interface 704, a memory 706, and a communication bus 708.
Wherein: the processor 702, communication interface 704, and memory 706 communicate with each other via a communication bus 708. A communication interface 704 for communicating with network elements of other devices, such as clients or other servers. The processor 702 is configured to execute the program 710, and may specifically execute relevant steps in the service processing method embodiment of the electronic device.
In particular, the program 710 may include program code that includes computer operating instructions.
The processor 702 may be a central processing unit CPU, or an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits configured to implement an embodiment of the present invention. The electronic device comprises one or more processors, which can be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.
The memory 706 stores a program 710. The memory 706 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
The program 710 may specifically be used to cause the processor 702 to perform the following operations:
training a neural network model through sample data to obtain a text classifier for automatically identifying an input text; acquiring consultation information; determining a text recognition result corresponding to the consultation information through the text classifier; when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result; and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
In an optional mode, the response information comprises knowledge information matched with the text recognition result and notice information;
the step of outputting response information matching the text recognition result when there is response information matching the text recognition result, further comprising:
judging whether knowledge information matched with the text recognition result exists or not; if yes, outputting the knowledge information; if not, judging whether notice information matched with the text recognition result exists or not; and if so, outputting the announcement information.
In an optional manner, the step of obtaining the advisory information further includes:
acquiring the consultation content of the user in real time; and performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information.
In an optional manner, the step of outputting the knowledge information further includes: and displaying the knowledge information through a site established by a fifth generation hypertext markup language.
In an optional manner, the step of performing training of a neural network model through sample data to obtain a text classifier for automatically recognizing an input text further includes:
generating training sample data; extracting the feature vectors of the training sample data by the convolutional layer and outputting a plurality of feature graphs; extracting the maximum value in each feature map through a pooling layer to serve as an output feature vector; splicing each output feature vector to obtain a text feature vector corresponding to the training sample data; determining a classification label of the training sample data according to the text feature vector; and carrying out reverse training on the classification label of the training sample data, and optimizing the parameters of the convolutional layer and the pooling layer.
In an optional manner, the step of generating training sample data further includes: collecting the existing consultation content; performing word segmentation processing on the consulting content through a preset word segmentation dictionary to form a set consisting of a plurality of words; and coding the consultation data to obtain corresponding coded data serving as the training sample data.
In an optional manner, the extracting, by the convolutional layer, the feature vectors of the training sample data to form a plurality of feature maps, further includes:
converting each word in the training sample data into a word vector with the same length, and generating a matrix corresponding to the consultation content; accessing the matrix into a convolution layer, calculating the convolution of n word vectors each time, and outputting a plurality of characteristic graphs; each of the feature maps contains a plurality of feature vectors.
The electronic equipment of the embodiment of the invention realizes intelligent analysis and identification of the consultation contents based on the neural network model, can realize automatic response of most of the consultation contents, and effectively improves the service processing efficiency of the customer service system.
Furthermore, two modes of passive response and active guidance are integrated, the problems of simple question and answer process and insufficient flexibility of a traditional automatic response customer service system are solved, the submission amount of the customer service work order can be reduced through the distribution of the consultation contents, and adverse effects on the response effect of the customer service system can be avoided while the number of the customer service work order processing personnel is reduced.
The algorithms or displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. In addition, embodiments of the present invention are not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the invention and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names. The steps in the above embodiments should not be construed as limiting the order of execution unless specified otherwise.
Claims (10)
1. A method for processing a service, the method comprising:
training a neural network model through sample data to obtain a text classifier for automatically identifying an input text;
acquiring consultation information;
determining a text recognition result corresponding to the consultation information through the text classifier;
when response information matched with the text recognition result exists, outputting the response information matched with the text recognition result;
and when the response information matched with the text recognition result does not exist, displaying a guide page for creating the customer service work order.
2. The service processing method according to claim 1, wherein the response message includes knowledge information and announcement information that match the text recognition result;
the step of outputting response information matching the text recognition result when there is response information matching the text recognition result, further comprising:
if the knowledge information matched with the text recognition result exists, outputting the knowledge information;
and if the announcement information matched with the text recognition result exists, outputting the announcement information.
3. The service processing method according to claim 1, wherein the step of obtaining the consulting information further comprises:
acquiring the consultation content of the user in real time;
and performing word segmentation processing on the consultation content according to a preset dictionary to obtain the consultation information.
4. The traffic processing method according to claim 2 or 3, wherein the step of outputting the knowledge information further comprises:
and displaying the knowledge information through a site established by a fifth generation hypertext markup language.
5. The business processing method of claim 1, wherein the step of training the neural network model through the sample data to obtain a text classifier for automatically recognizing the input text further comprises:
generating training sample data;
extracting the feature vectors of the training sample data by the convolutional layer and outputting a plurality of feature graphs;
extracting the maximum value in each feature map through a pooling layer to serve as an output feature vector;
splicing each output feature vector to obtain a text feature vector corresponding to the training sample data;
determining a classification label of the training sample data according to the text feature vector;
and carrying out reverse training on the classification label of the training sample data, and optimizing the parameters of the convolutional layer and the pooling layer.
6. The method of claim 5, wherein the step of generating training sample data further comprises:
collecting the existing consultation content;
performing word segmentation processing on the consulting content through a preset word segmentation dictionary to form a set consisting of a plurality of words;
and coding the consultation data to obtain corresponding coded data serving as the training sample data.
7. The method of claim 6, wherein the extracting feature vectors of the training sample data from the convolutional layer to form a number of feature maps, further comprises:
converting each word in the training sample data into a word vector with the same length, and generating a matrix corresponding to the consultation content;
accessing the matrix into a convolution layer, calculating the convolution of n word vectors each time, and outputting a plurality of characteristic graphs; each of the feature maps contains a plurality of feature vectors.
8. A traffic processing apparatus, characterized in that the apparatus comprises:
the off-line training module is used for training the neural network model through the sample data to obtain a text classifier for automatically identifying the input text;
the information acquisition module is used for acquiring the consultation information;
the text analysis module is used for determining a text recognition result corresponding to the consultation information through the text classifier;
the feedback module is used for outputting response information matched with the text recognition result when the response information matched with the text recognition result exists; and displaying a guidance page for creating the customer service order when there is no response information matching the text recognition result.
9. An electronic device for business processing, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is configured to store at least one executable instruction that causes the processor to perform the business process method of any one of claims 1-7.
10. A computer storage medium having stored therein at least one executable instruction for causing a processor to perform a method of business processing according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990265.3A CN112699233A (en) | 2019-10-17 | 2019-10-17 | Service processing method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990265.3A CN112699233A (en) | 2019-10-17 | 2019-10-17 | Service processing method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112699233A true CN112699233A (en) | 2021-04-23 |
Family
ID=75504631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910990265.3A Pending CN112699233A (en) | 2019-10-17 | 2019-10-17 | Service processing method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112699233A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108597519A (en) * | 2018-04-04 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | A kind of bill classification method, apparatus, server and storage medium |
| CN109461039A (en) * | 2018-08-28 | 2019-03-12 | 厦门快商通信息技术有限公司 | A kind of text handling method and intelligent customer service method |
| CN109615407A (en) * | 2018-10-19 | 2019-04-12 | 国家电网有限公司 | Online electric power wisdom client service method based on artificial intelligence |
| CN109753565A (en) * | 2018-12-27 | 2019-05-14 | 厦门智融合科技有限公司 | Intellectual Property intelligent service method and system |
| CN109919671A (en) * | 2019-02-27 | 2019-06-21 | 上海宝尊电子商务有限公司 | The high efficiency smart client service method and system of a kind of Electronic Commerce field customer orientation |
| CN110059173A (en) * | 2019-04-19 | 2019-07-26 | 辽宁工程技术大学 | A kind of intelligent kitchen question and answer assistant system of knowledge based map |
| CN110147445A (en) * | 2019-04-09 | 2019-08-20 | 平安科技（深圳）有限公司 | Intension recognizing method, device, equipment and storage medium based on text classification |
-
2019
- 2019-10-17 CN CN201910990265.3A patent/CN112699233A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108597519A (en) * | 2018-04-04 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | A kind of bill classification method, apparatus, server and storage medium |
| CN109461039A (en) * | 2018-08-28 | 2019-03-12 | 厦门快商通信息技术有限公司 | A kind of text handling method and intelligent customer service method |
| CN109615407A (en) * | 2018-10-19 | 2019-04-12 | 国家电网有限公司 | Online electric power wisdom client service method based on artificial intelligence |
| CN109753565A (en) * | 2018-12-27 | 2019-05-14 | 厦门智融合科技有限公司 | Intellectual Property intelligent service method and system |
| CN109919671A (en) * | 2019-02-27 | 2019-06-21 | 上海宝尊电子商务有限公司 | The high efficiency smart client service method and system of a kind of Electronic Commerce field customer orientation |
| CN110147445A (en) * | 2019-04-09 | 2019-08-20 | 平安科技（深圳）有限公司 | Intension recognizing method, device, equipment and storage medium based on text classification |
| CN110059173A (en) * | 2019-04-19 | 2019-07-26 | 辽宁工程技术大学 | A kind of intelligent kitchen question and answer assistant system of knowledge based map |
Non-Patent Citations (2)
| Title |
|---|
| 涂文博等: ""针对文本分类的神经网络模型"", 《计算机***应用》 * |
| 白林亭等: "基于深度学习的视觉问答技术研究", 《航空计算技术》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111026842B (en) | Natural language processing method, natural language processing device and intelligent question-answering system | |
| CN108597519B (en) | Call bill classification method, device, server and storage medium | |
| CN111932144B (en) | Customer service agent distribution method and device, server and storage medium | |
| CN112380870A (en) | User intention analysis method and device, electronic equipment and computer storage medium | |
| CN112559865B (en) | Information processing system, computer-readable storage medium, and electronic device | |
| US20220358292A1 (en) | Method and apparatus for recognizing entity, electronic device and storage medium | |
| US20220237567A1 (en) | Chatbot system and method for applying for opportunities | |
| CN110516057B (en) | Petition question answering method and device | |
| CN111917878A (en) | Message processing method, device, equipment and storage medium | |
| CN111970400B (en) | Crank call identification method and device | |
| CN114090755B (en) | Reply sentence determining method and device based on knowledge graph and electronic equipment | |
| CN112926308B (en) | Method, device, equipment, storage medium and program product for matching text | |
| CN111651600A (en) | Sentence multi-intention recognition method, system, electronic device and storage medium | |
| CN110689359A (en) | Method and device for dynamically updating model | |
| CN109493186A (en) | The method and apparatus for determining pushed information | |
| CN113435209A (en) | Data management method and system based on shared laboratory platform | |
| CN116071150A (en) | Data processing method, bank product popularization, wind control system, server and medium | |
| CN110990527A (en) | Automatic question answering method and device, storage medium and electronic equipment | |
| CN116757855A (en) | Intelligent insurance service method, device, equipment and storage medium | |
| CN114461665B (en) | Method, apparatus and computer program product for generating a statement transformation model | |
| CN111026972A (en) | Subscription data pushing method, device, equipment and storage medium in Internet of things | |
| CN112699233A (en) | Service processing method and device and electronic equipment | |
| CN113886547B (en) | Client real-time dialogue switching method and device based on artificial intelligence and electronic equipment | |
| CN115909357A (en) | Target identification method based on artificial intelligence, model training method and device | |
| CN114943590A (en) | Object recommendation method and device based on double-tower model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210423 |
|
| RJ01 | Rejection of invention patent application after publication |