CN111813910B

CN111813910B - Customer service problem updating method, customer service problem updating system, terminal equipment and computer storage medium

Info

Publication number: CN111813910B
Application number: CN202010595834.7A
Authority: CN
Inventors: 侯翠琴; 文彬; 李剑锋
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2020-06-24
Filing date: 2020-06-24
Publication date: 2024-05-31
Anticipated expiration: 2040-06-24
Also published as: CN111813910A; WO2021114834A1

Abstract

The invention relates to the technical field of customer service robots, and discloses a customer service problem updating method, a customer service problem updating system, terminal equipment and a computer storage medium, wherein the customer service problem updating method comprises the following steps: constructing a problem pair according to customer service problems in a preset customer service problem set, and combining the problem pair into training data; training a similarity model based on the training data, and calculating clustering parameters of unresponsive customer service problems by utilizing the training converged similarity model; performing clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem; and generating new customer service questions according to the clustering result to update the preset customer service question set. In addition, the invention also relates to a blockchain technology, and a training convergence similarity model can be stored in the blockchain. The invention can obtain more valuable high-quality questions to update the customer service question set during clustering, and improves the question answering efficiency of the intelligent customer service robot.

Description

Customer service problem updating method, customer service problem updating system, terminal equipment and computer storage medium

Technical Field

The present invention relates to the field of customer service robots, and in particular, to a customer service problem update method, a customer service problem update system, a customer service problem update terminal device, and a customer service problem update computer storage medium.

Background

Intelligent customer service robots are typically composed of Speech recognition (Automatic Speech Recognition, ASR), intent recognition, question and answer modules, knowledge base management, knowledge maps, dialog management, text generation, speech synthesis (TTS), etc. The FAQ (Frequently Asked Questions, solution of common problem items and corresponding problems) question and answer module based on the knowledge base provides satisfactory answers for users by inquiring standard problems matched with the user problems in the knowledge base, and is the most important module in the intelligent customer service robot. The knowledge base is an important component of the FAQ question-answering module and consists of questions and corresponding answer pairs which are frequently asked by users, and in order to adapt to different question methods of different users on the same question, each standard question can be generalized to a plurality of similar questions. With the development of business, the question set of the knowledge base needs to be updated continuously to improve the answer rate of the customer service robot.

At present, in order to reduce the workload of maintenance operators, the problems which are not responded by the nearest robot customer service robot are clustered through a clustering technology, and then the operators rely on the clustering result and the knowledge base to sort out new standard problems and corresponding similar problems and add the new standard problems and the corresponding similar problems into the knowledge base to update the problem set. However, as the business progresses, a large number of new business problems are continuously presented, and the problem set is updated only by clustering the problems with high quality based on the clustering method for the unresponsive problems.

Disclosure of Invention

The invention mainly aims to provide a customer service problem updating method, a customer service problem updating system, a customer service problem updating terminal device and a customer service problem updating computer storage medium, and aims to solve the technical problem that a high-quality problem can not be obtained by clustering only in a clustering mode aiming at unresponsive problems in the prior art, and update the problem set of an intelligent customer service robot.

In order to achieve the above object, an embodiment of the present invention provides a method for updating a customer service problem, where the method for updating a customer service problem includes:

constructing a problem pair according to customer service problems in a preset customer service problem set, and combining the problem pair into training data;

Training a similarity model based on the training data, and calculating clustering parameters of unresponsive customer service problems by utilizing the training converged similarity model;

performing clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem;

and generating new customer service questions according to the clustering result to update the customer service question set.

Preferably, the customer service questions include standard questions and similar questions, and the step of constructing question pairs according to customer service questions in a preset customer service question set and combining the question pairs into training data includes:

extracting each standard problem and each similar problem corresponding to each standard problem from a preset customer service problem set;

According to a first standard problem in the standard problems and each similar problem corresponding to the first standard problem, constructing and obtaining each similar problem pair;

constructing each dissimilar problem pair according to the first standard problem and each similar problem corresponding to a second standard problem in the standard problems, wherein the first standard problem is different from the second standard problem;

and correspondingly combining each similar problem pair with each dissimilar problem pair to form each training data.

Preferably, the step of training a similarity model based on the training data includes:

splitting the training data into a first data set and a second data set;

Training the similarity model by using the first data set, and verifying the trained similarity model by using the second data set;

The validated similarity models are labeled as training converged similarity models, and the training converged similarity models are stored in the blockchain.

Preferably, the clustering parameters include: the vector representation and similarity value of the question sentence,

The step of calculating the clustering parameters of the unresponsive customer service problems by using the training convergence similarity model comprises the following steps:

Acquiring a question sentence of the unresponsive customer service question and a duplicate sentence of the question sentence, and inputting the question sentence and the duplicate sentence into a training convergence similarity model;

obtaining a training convergence similarity model to calculate a vector average value of similarity values between the problem sentences and the duplicate sentences;

taking the vector average value as a vector representation of the problem sentence;

Or traversing the problem sentences to construct sentence pairs, and inputting the sentence pairs into a training convergence similarity model;

and obtaining a similarity value output after the similarity value calculation of the sentence pair by the training convergence similarity model, and taking the similarity value as the similarity value of the problem sentence.

Preferably, the step of performing a clustering operation based on the clustering parameter to obtain a clustering result of the unresponsive customer service problem includes:

invoking a preset distance-based clustering algorithm, and clustering the unresponsive customer service problems by using the similarity value to obtain a clustering result of the unresponsive customer service problems;

Or calling a preset clustering algorithm based on sentence representation, and clustering the unresponsive customer service questions by using the vector representation to obtain a clustering result of the unresponsive customer service questions.

Preferably, the step of generating a new customer service problem according to the clustering result to update the preset customer service problem set includes:

generating new standard questions and similar questions corresponding to the new standard questions according to the clustering result and the customer service question set;

configuring corresponding service answers for the new standard questions and similar questions corresponding to the new standard questions;

And adding the new standard questions, the similar questions corresponding to the new standard questions and the service answers to the preset customer service question set to update the preset customer service question set.

Preferably, the method for updating customer service problems further includes:

And acquiring the non-response customer service questions at intervals of a preset time period, and storing the acquired non-response customer service questions in a blockchain.

In addition, in order to achieve the above object, the present invention further provides a system for updating a customer service problem, where the system for updating a customer service problem includes: memory, processor, communication bus, and update program for customer service problems stored on the memory,

The communication bus is used for realizing communication connection between the processor and the memory;

the processor is configured to execute an update procedure of the customer service problem to implement the following steps:

And generating new customer service questions according to the clustering result to update the preset customer service question set.

splitting the training data into a first data set and a second data set;

Preferably, the method for updating customer service problems further includes:

In addition, to achieve the above object, the present invention also provides a computer storage medium storing one or more programs executable by one or more processors for:

According to the customer service problem updating method, the customer service problem updating system, the terminal equipment and the computer readable storage medium, problem pairs are constructed according to customer service problems in a preset customer service problem set, and the problem pairs are combined into training data; training a similarity model based on the training data, and calculating clustering parameters of unresponsive customer service problems by utilizing the training converged similarity model; performing clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem; and generating new customer service questions according to the clustering result to update the preset customer service question set.

According to the invention, the problem pairs are constructed based on the fact that the intelligent customer service robot extracts customer service problems from the preset customer service problem set, then the training data is obtained based on the corresponding combination of the problem pairs, the similarity model is trained by utilizing the training data, then the clustering parameters of the non-responded customer service problems of the customer service robot are calculated based on the similarity model converged by training, and the subsequent process of updating the customer service problem set is carried out after the non-responded customer service problems are clustered based on the clustering parameters, so that the self-adaptive learning is carried out on the clustering parameters of the non-responded customer service problems according to the change dynamics of the customer service problems of the knowledge base on the basis that the new problems are obtained by simple clustering problems, the more valuable high-quality problems are obtained during clustering, the customer service problem set is updated, and the problem answering efficiency of the intelligent customer service robot is improved.

Drawings

FIG. 1 is a schematic structural diagram of a hardware operating environment of a terminal device according to a method of an embodiment of the present invention;

FIG. 2 is a flowchart illustrating an embodiment of a method for updating customer service problems according to the present invention;

fig. 3 is a schematic diagram of functional modules of a customer service problem update system according to the present invention.

The achievement of the objects, functional features and advantages of the present invention 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 invention.

The main solutions of the embodiments of the present invention are: constructing a problem pair according to customer service problems in a preset customer service problem set, and combining the problem pair into training data; training a similarity model based on the training data, and calculating clustering parameters of unresponsive customer service problems by utilizing the training converged similarity model; performing clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem; and generating new customer service questions according to the clustering result to update the preset customer service question set.

As the intelligent customer service robot technology business develops, the intelligent customer service robot relies on the knowledge base, which is an important component of the FAQ question-answering module, the question set needs to be updated continuously to improve the question answering efficiency of the customer service robot. In order to reduce the workload of maintenance operators in the prior art, the problems which are not responded by the nearest customer service robot are clustered by a clustering technology, and then the operators rely on the clustering result and the knowledge base to sort out new standard problems and corresponding similar problems and add the new standard problems and the corresponding similar problems into the knowledge base to update the problem set. However, as the business progresses, a large number of new business problems are continuously presented, and the problem set is updated only by clustering the problems with high quality based on the clustering method for the unresponsive problems.

According to the solution provided by the invention, the problem pairs are constructed based on the customer service questions extracted from the preset customer service question set by the intelligent customer service robot, then the training data is obtained based on the corresponding combination of the problem pairs, the similarity model is trained by utilizing the training data, then the clustering parameters of the non-response customer service questions of the customer service robot are calculated based on the training converged similarity model, the subsequent process of updating the customer service question set is carried out after the non-response customer service questions are clustered based on the clustering parameters, so that the self-adaptive learning is carried out on the clustering parameters of the non-response customer service questions according to the change dynamics of the customer service questions of the knowledge base on the basis that the new questions are obtained by the simple clustering questions, the more valuable high-quality questions are obtained during clustering, and the question answering efficiency of the intelligent customer service robot is improved.

As shown in fig. 1, fig. 1 is a schematic device structure diagram of a hardware operating environment of a terminal device according to an embodiment of the present invention.

The terminal equipment of the embodiment of the invention can be an intelligent customer service robot, and also can be terminal equipment such as a PC, a smart phone, a tablet personal computer, a portable computer and the like.

As shown in fig. 1, the terminal device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.

Optionally, the terminal device may further include a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, and the like. Among other sensors, such as light sensors, motion sensors, and other sensors. In particular, the light sensor may comprise an ambient light sensor that may adjust the brightness of the display screen according to the brightness of ambient light, and a proximity sensor that may turn off the display screen and/or backlight when the device is moved to the ear. As one of the motion sensors, the gravity acceleration sensor can detect the acceleration in all directions (generally three axes), and can detect the gravity and the direction when the device is stationary, and the device can be used for identifying the gesture of the device (such as horizontal and vertical screen switching, related games, magnetometer gesture calibration), vibration identification related functions (such as pedometer and knocking), and the like; of course, the mobile terminal may also be configured with other sensors such as a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, and the like, which are not described herein.

It will be appreciated by those skilled in the art that the device structure shown in fig. 1 does not constitute a limitation of the terminal device, and in other embodiments, the terminal device may also include more or less components than illustrated, or may combine certain components, or may have a different arrangement of components.

As shown in fig. 1, an operating system, a network communication module, a user interface module, and an update program of a customer service problem may be included in a memory 1005 as one type of computer storage medium.

In the terminal device shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server, and performing data communication with the background server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to call an update program of the customer service problem stored in the memory 1005, and perform the following steps:

Further, the customer service questions include standard questions and similar questions, and the processor 1001 may be configured to call an update program of the customer service questions stored in the memory 1005, and further perform the following steps:

Further, the processor 1001 may be configured to call an update program of the customer service problem stored in the memory 1005, and further perform the following steps:

splitting the training data into a first data set and a second data set;

Further, the clustering parameters include: the processor 1001 may be configured to call an update program of the customer service problem stored in the memory 1005, and further perform the following steps:

The embodiment of the terminal device related to the updating method of the customer service problem in the present invention is basically the same as each embodiment of the updating method of the customer service problem described below, and will not be described herein.

The invention provides a customer service problem updating method.

Referring to fig. 2, fig. 2 is a flow chart of a first embodiment of a customer service problem update method of the present invention, where the first embodiment of the customer service problem update method is applied to an intelligent customer service robot, and in this embodiment, the customer service problem update method includes:

step S100, constructing a problem pair according to customer service problems in a preset customer service problem set, and combining the problem pair into training data;

The intelligent customer service robot extracts customer service problems from a preset customer service problem set, then constructs each problem pair based on the customer service problems, and finally combines the problem pairs in a one-to-one correspondence manner to form training data.

It should be noted that, in this embodiment, the FAQ question-answering module in the intelligent customer service robot has a knowledge base, and the knowledge base is an important component of the FAQ question-answering module. The preset customer service problem set is a customer service problem set stored in the knowledge base and consists of a plurality of standard problems and one or more similar problems corresponding to the standard problems, so that the intelligent customer service robot queries standard problems matched with customer service problems set by a user in the knowledge base through the FAQ question answering module, and a satisfactory answer is provided for the user. In addition, the training data formed by combining the problem pairs in a one-to-one correspondence by the intelligent customer service robot may be one data set composed of a plurality of pieces of training data.

Specifically, for example, the intelligent customer service robot periodically or randomly initiates an operation of updating a customer service problem set, accesses a knowledge base of the FAQ question-answering module, extracts standard problems and similar problems from the customer service problem set stored in the knowledge base, constructs problem pairs based on the standard problems and similar problems, and finally forms a plurality of training data by correspondingly combining the problem pairs according to a one-to-one rule to form a data set.

Further, in an embodiment, in step S100, it may include:

Step S101, extracting each standard problem and each similar problem corresponding to each standard problem from a preset customer service problem set;

step S102, constructing and obtaining each similar problem pair according to a first standard problem in the standard problems and each similar problem corresponding to the first standard problem;

Specifically, for example, after initiating an operation of updating a customer service question set, the intelligent customer service robot extracts a plurality of standard questions and similar questions corresponding to the standard questions from the customer service question set stored in a knowledge base based on accessing the knowledge base of the FAQ question-answering module, and then randomly combines any one of the standard questions with the similar questions of the standard questions, thereby obtaining a plurality of similar question pairs by the component.

Step S103, constructing and obtaining each dissimilar problem pair according to the first standard problem and each similar problem corresponding to the second standard problem in each standard problem;

in this embodiment, the first standard problem is different from the second standard problem.

Specifically, for example, after the intelligent customer service robot accesses the knowledge base of the FAQ question-answering module and extracts a plurality of standard questions and similar questions corresponding to the standard questions from the customer service questions set stored in the knowledge base, the intelligent customer service robot may perform (or may perform asynchronously of course) the operation of constructing the dissimilar question pair while constructing the similar question pair, that is, the intelligent customer service robot randomly combines any one standard question among the standard questions with a plurality of similar questions of any standard question other than the standard question among the standard questions, thereby obtaining a plurality of dissimilar question pairs.

Step S104, corresponding and combining each similar problem pair and each dissimilar problem pair to form each training data.

Specifically, for example, the intelligent customer service robot correspondingly combines the plurality of similar problem pairs and the plurality of dissimilar problem pairs formed by combination according to a rule of one similar problem pair and one dissimilar problem pair to form one piece of training data, and then composes a data set from the plurality of pieces of training data constructed according to the rule.

Further, in another embodiment, when the intelligent customer service robot forms training data based on the combination of the plurality of similar problem pairs and the plurality of dissimilar problem pairs, the intelligent customer service robot may also form one piece of training data by combining the plurality of dissimilar problem pairs and the plurality of similar problem pairs, so as to construct and obtain a plurality of pieces of training data to form a data set.

It should be noted that, in this embodiment, the similar problem pair and the dissimilar problem pair that compose the same piece of training data are the same as the respective standard problem, and it can be understood that, based on different design requirements of practical application, the similar problem pair and the dissimilar problem pair that compose the same piece of training data may also be different from the respective standard problem.

Step S200, training a similarity model based on the training data, and calculating clustering parameters of unresponsive customer service problems by using the training converged similarity model;

the intelligent customer service robot extracts a customer service problem set from a preset knowledge base, then builds training data based on the customer service problem set, trains a similarity model by using the training data until the similarity model converges, and then calculates and obtains subsequent clustering parameters for clustering the unresponsive customer service problem by using the similarity model which is trained to converge based on the training data.

In this embodiment, the similarity model may be any one of a bert model (Bidirectional Encoder Representations from Transformers, which is a general pre-training language representation model with very good effect proposed by Google recently), a xlnet (a general autoregressive pre-training method) model, a siamesecnn (twin convolution) network, and a siameselstm (twin) network. In addition, the unresponsive customer service problem is a customer service problem recorded by the intelligent customer service robot when the customer service problem presented by the user cannot be successfully matched with the standard problem in the whole operation process of the intelligent customer service robot, and thus an answer is not obtained for outputting to the user (i.e., the unresponsive customer service problem is a customer service problem which is not answered by the intelligent customer service robot).

Specifically, for example, the intelligent customer service robot uses bert model as similarity model, after the intelligent customer service robot builds multiple pieces of training data based on multiple similar problem pairs and multiple dissimilar problem pairs, part of training data in the multiple pieces of training data is used as training sample to be input into bert model to train the bert model, until the intelligent customer service robot verifies that the bert model converges, the bert model which has been trained and converged is utilized to calculate the clustering parameters of one or more pre-recorded unresponsive customer service problems, so as to perform clustering operation on the unresponsive customer service problems based on the clustering parameters.

Step S300, carrying out clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem;

The intelligent customer service robot calculates and outputs the clustering parameters of the unresponsive customer service problems based on the training convergence similarity model by calling the existing mature clustering algorithm, and then performs clustering operation on the unresponsive customer service problems by utilizing the distance parameters, so as to obtain the clustering result of the unresponsive customer service problems.

And step S400, generating new customer service questions according to the clustering result so as to update the customer service question set.

After the intelligent customer service robot obtains a clustering result of the unresponsive customer service problems through clustering operation, automatically sorting to obtain new customer service problems according to the clustering result and the customer service problem set stored in the knowledge base of the FAQ question-answering module, and then adding the new customer service problems into the customer service problem set in the knowledge base, thereby completing updating of the customer service problem set.

It should be noted that, in this embodiment, after the intelligent customer service robot obtains a clustering result of the unresponsive customer service questions through a clustering operation, the clustering result may be output to an operator of the intelligent customer service robot, so as to assist the operator to sort out new customer service questions according to the clustering result and the customer service question set stored in the knowledge base of the FAQ question-answering module, and then the intelligent customer service robot receives the new customer service questions input by the operator and adds the new customer service questions to the customer service question set in the knowledge base, thereby completing updating of the customer service question set.

In the embodiment, a customer service problem set is extracted from a preset knowledge base through an intelligent customer service robot, and training data is constructed based on the customer service problem set; extracting a customer service problem set from a preset knowledge base, constructing and obtaining training data based on the customer service problem set, training a similarity model by using the training data until the similarity model converges, and calculating non-response customer service problems which are not responded by using the training data to obtain clustering parameters for clustering the non-response customer service problems; after calculating and outputting the clustering parameters of the unresponsive customer service problems based on the training convergence similarity model, the intelligent customer service robot performs clustering operation on the unresponsive customer service problems by using the distance parameters, so as to obtain a clustering result of the unresponsive customer service problems; and finally, automatically sorting to obtain new customer service questions according to the clustering result and the customer service question set stored in the knowledge base of the FAQ question-answering module by the intelligent customer service robot, and adding the new customer service questions into the customer service question set in the knowledge base, thereby completing updating of the customer service question set.

The method comprises the steps of extracting customer service questions from a preset customer service question set based on an intelligent customer service robot to construct a question pair, then carrying out corresponding combination based on the question pair to obtain training data, training a similarity model by utilizing the training data, calculating clustering parameters of unresponsive customer service questions of the customer service robot based on the training converged similarity model, carrying out subsequent updating of the customer service question set after clustering the unresponsive customer service questions based on the clustering parameters, and carrying out self-adaptive learning on the clustering parameters of the unresponsive customer service questions according to the change dynamics of the customer service questions of a knowledge base on the basis of obtaining new questions by simple clustering, so that the customer service question set is updated by obtaining more valuable high-quality questions during clustering, and the question answering efficiency of the intelligent customer service robot is improved.

Further, a second embodiment of the method for updating a customer service problem is provided on the basis of the first embodiment of the method for updating a customer service problem according to the present invention, in this embodiment, in the step S200, the step of training a similarity model based on the training data may include:

Step S201, the training data is segmented into a first data set and a second data set;

the intelligent customer service robot extracts the standard questions and the similar questions corresponding to the standard questions from the customer service question set stored in the preset knowledge base based on the preset knowledge base, then combines the similar questions and the dissimilar questions to obtain a similar question pair, constructs the similar question pair and the dissimilar question pair to obtain a data set of training data, and then cuts the data set according to a preset proportion or randomly to obtain a first data set and a second data set.

It should be noted that, in this embodiment, the preset proportion may be a proportion set by an operator based on design requirements, and it should be understood that, based on different design requirements of practical applications, the preset proportion may be set to any numerical proportion, and the method for updating customer service problems in the present invention does not specifically limit the magnitude of the preset proportion.

Step S202, training the similarity model by using the first data set, and verifying the trained similarity model by using the second data set;

After the data set of the training data is segmented into the first data set and the second data set according to a preset proportion or randomly, the intelligent customer service robot inputs the training data in the first data set into the similarity model to train the similarity model, and immediately inputs the training data in the second data set into the similarity training model after training of each wheel of the similarity model by using the training data is finished, so that whether the similarity model is converged after training of the current wheel is finished is verified according to the output result of the similarity model.

Further, in another embodiment, the intelligent customer service robot may also input training data in the second data set to the similarity model to train the similarity model, and immediately input training data in the first data set to the similarity training model after training is completed for each round of the similarity model by using the training data, so as to verify whether the similarity model has converged after training of the current round according to an output result of the similarity model.

In step S203, the similarity model that passes the verification is marked as a training converged similarity model, and the training converged similarity model is stored in the blockchain.

If the intelligent customer service robot inputs the training data in the second data set into the trained similarity training model, verifies that the similarity model is trained and converged according to the output result of the similarity model, confirms that the verification of the similarity model by the second data set is passed, marks the similarity model as a training and converged similarity model in real time, and stores the training and converged similarity model in a node of a blockchain.

It should be noted that, in this embodiment, in order to ensure that the training convergence similarity model is not modified or removed by mistake, the training convergence similarity model may be stored in a node of a blockchain, so as to ensure the security of the index tag, further ensure the accuracy of performing subsequent cluster parameter calculation on the unresponsive customer service problem of the intelligent customer service robot based on the training convergence similarity model, and further improve the efficiency of self-adaptively learning the new service problem of the intelligent customer service robot.

In the embodiment, the intelligent customer service robot divides the data set of the training data obtained by construction into a first data set and a second data set; training the similarity model by using training data in the first data set, and verifying the trained similarity model by using training data in the second data set after the training of each round is finished; finally, the similarity model passing through verification is marked as a training convergence similarity model, and the training convergence similarity model is stored in a blockchain for later calling. According to the method, a data set consisting of similar problem pairs and dissimilar problem pairs is constructed according to a customer service problem set (standard problem + similar problem of standard problem) stored in a knowledge base of a current FAQ question-answering module, and then a similarity model is trained on the basis of the data set to learn clustering parameters of non-responded customer service problems in a follow-up self-adaptive mode, so that when the non-responded customer service problems are clustered, more valuable new problems can be obtained to update the customer service problem set, and the problem answering efficiency of the intelligent customer service robot is improved.

Further, on the basis of the first embodiment of the method for updating a customer service problem, a third embodiment of the method for updating a customer service problem is provided, and in this embodiment, the method for updating a customer service problem may further include:

and step A, acquiring the non-response customer service questions at intervals of preset time length, and storing the acquired non-response customer service questions in a blockchain.

It should be noted that, in this embodiment, the intelligent customer service robot encounters an unresponsive customer service problem that cannot output an answer to a user in the continuous operation process, so that the intelligent customer service robot may pre-establish a temporary storage space, and once the unresponsive customer service problem that cannot output an answer to a user is encountered, the unresponsive customer service problem is cached in the storage space. In addition, the preset time period may be specifically 24 hours, and it should be understood that, based on different design requirements of practical applications, the preset time period may of course be any other time period, and the method for updating the customer service problem according to the present invention is not specifically limited to the preset time period.

Specifically, for example, the intelligent customer service robot acquires an unresponsive customer service problem recorded by the intelligent customer service robot during operation and outputting an answer to a customer service problem submitted by a user within 24 hours before the 0:00 from a temporary storage space established in advance in the early morning of 0:00 a day, and then stores the unresponsive customer service problem in a node of a blockchain for later recall.

Further, in another embodiment, the intelligent customer service robot may directly perform persistent caching on the unresponsive customer service questions recorded by the intelligent customer service robot in the operation process and used for outputting answers to customer service questions proposed by users within 24 hours before 0:00 in a pre-established temporary storage space, so as to be directly invoked during subsequent calculation of clustering parameters of the unresponsive customer service questions and clustering operation.

Further, in the present embodiment, the intelligent customer service robot uses the cluster parameters of the unresponsive customer service problem calculated by the similarity model which is trained and converged, including but not limited to: the step of calculating the clustering parameters of the unresponsive customer service questions using the training converged similarity model in the step S200 may include:

Step S204, obtaining a question sentence of the unresponsive customer service question and a duplicate sentence of the question sentence, and inputting the question sentence and the duplicate sentence into a training convergence similarity model;

After the intelligent customer service robot obtains the unresponsive customer service problem, the problem sentence of the unresponsive customer service problem is read through the existing mature language processing technology, the problem sentence is copied to obtain the copied sentence of the problem sentence, and then the intelligent customer service robot inputs the problem sentence and the copied sentence of the problem sentence into a training convergence similarity model.

Specifically, for example, after the intelligent customer service robot acquires one or more unresponsive customer service questions from nodes of the blockchain, invoking an existing arbitrarily mature natural language processing technology, reading texts of the unresponsive customer service questions one by one as question sentences, continuing to copy the question sentences to obtain copied sentences corresponding to the question sentences, and then inputting the question sentences and the copied sentences corresponding to the question sentences one by one into a similarity model which is also acquired from the nodes of the blockchain and has been trained and converged.

Step S205, obtaining a training convergence similarity model to calculate a vector average value of similarity values between the problem sentences and the duplicate sentences;

Step S206, taking the vector average value as a vector representation of the problem sentence;

After inputting a problem sentence which does not respond to a customer service problem and a duplicate sentence corresponding to the problem sentence into a similarity model which is trained and converged, the intelligent customer service robot acquires the vector of the problem sentence and the vector of the duplicate sentence when the similarity model calculates the similarity value between the problem sentence and the duplicate sentence, calculates the vector average value of the vector of the problem sentence and the vector of the duplicate sentence, and takes the vector average value as the vector representation of the problem sentence which does not respond to the customer service problem.

Further, in another embodiment, in the step S200, the step of calculating the cluster parameters of the unresponsive customer service problem by using the training converged similarity model may further include:

step S207, traversing the problem sentences to construct sentence pairs, and inputting the sentence pairs into a training convergence similarity model;

Step S208, obtaining a similarity value output after the similarity value calculation of the sentence pair by the training convergence similarity model, and taking the similarity value as the similarity value of the problem sentence.

The intelligent customer service robot acquires an unresponsive customer service problem, reads a problem sentence of the unresponsive customer service problem through the existing mature language processing technology, traverses the problem sentence to construct a sentence pair according to the current problem sentence and other problem sentences similar to the current problem sentence, and inputs the sentence pair into a trained and converged similarity model for the similarity model to calculate a similarity value between the problem sentence and other problem sentences similar to the problem sentence aiming at the sentence pair. And finally, the intelligent customer service robot calculates the similarity value of the similarity model based on the input sentence pair and outputs the similarity value, and takes the similarity value as the similarity value of the question sentence of the unresponsive customer service question.

Further, in this embodiment, step S300, performing a clustering operation based on the clustering parameter to obtain a clustering result of the unresponsive customer service problem may include:

step S301, a preset distance-based clustering algorithm is called, and the un-responded customer service problems are clustered by using the similarity value to obtain a clustering result of the un-responded customer service problems;

Aiming at the similarity value of the problem sentences of the unresponsive customer service problems output by the trained and converged similarity model, the intelligent customer service robot invokes the existing mature preset distance-based clustering algorithm to perform clustering operation on the model of the unresponsive customer service problems by using the similarity value, so as to obtain the clustering result of the unresponsive customer service problems.

It should be noted that, in this embodiment, the preset distance-based clustering algorithm includes, but is not limited to: affinity propagation (an AP clustering algorithm, affinity propagation belongs to one of message-passing algorithms (message passing algorithm) in a broad sense), spectral clustering (spectral clustering algorithm).

Further, in this embodiment, the step S300 may further include:

Step S302, a preset clustering algorithm based on sentence representation is called, and the vector representation is utilized to cluster the unresponsive customer service problems to obtain a clustering result of the unresponsive customer service problems.

Aiming at the similarity model which is trained and converged, when the similarity model calculates the similarity value between the problem sentence and the duplicate sentence, the intelligent customer service robot acquires the vector of the problem sentence and the vector of the duplicate sentence, calculates the vector representation of the problem sentence of the unresponsive customer service problem, calls the existing mature clustering algorithm which is preset and based on the sentence representation, and utilizes the vector representation to perform clustering operation on the model of the unresponsive customer service problem, so that the clustering result of the unresponsive customer service problem is obtained.

It should be noted that, in the present embodiment, the preset clustering algorithm based on sentence representation includes, but is not limited to: k-means (k-means clustering algorithm: k-means clustering algorithm, which is an iteratively solved clustering algorithm), DBSCAN (DBSCAN (Density-Based Spatial Clustering of Applications with Noise, which is a more representative Density-based clustering algorithm), and hierarchical clustering algorithm.

In this embodiment, the intelligent customer service robot collects the non-response customer service questions at intervals of a preset duration, and stores the collected non-response customer service questions in a blockchain or directly performs persistent caching on the non-response customer service questions.

When the intelligent customer service robot calculates the clustering parameters (vector representation and similarity value of the problem sentences of the non-response customer service problems), acquiring the problem sentences of the non-response customer service problems and the duplicate sentences of the problem sentences, and inputting the problem sentences and the duplicate sentences into a training convergence similarity model; then obtaining a vector average value of similarity values between the problem sentence and the duplicate sentence of the training convergence similarity model; and representing the vector average value as a vector; or constructing sentence pairs by traversing the problem sentences, and inputting the sentence pairs into a training converged similarity model; and then obtaining a similarity value output after the similarity value calculation of the sentence pairs by the training convergence similarity model, and taking the similarity value as the similarity value of the problem sentence. In addition, in the case of the optical fiber,

When the intelligent customer service robot performs clustering operation on the unresponsive customer service problems, a preset distance-based clustering algorithm is called, and the unresponsive customer service problems are clustered by using the similarity value of the problem sentences to obtain a clustering result of the unresponsive customer service problems; or calling a preset clustering algorithm based on sentence representation, and clustering the unresponsive customer service questions by using the vector representation of the question sentences to obtain a clustering result of the unresponsive customer service questions.

On the basis of simply clustering the unresponsive customer service questions to obtain new questions, the method further carries out self-adaptive learning on the clustering parameters of the unresponsive customer service questions according to the change of the knowledge base customer service questions, so that the more valuable high-quality questions can be obtained to update the customer service questions when the unresponsive customer service questions are clustered, and the question answering efficiency of the intelligent customer service robot is improved.

Further, based on the first embodiment of the method for updating a customer service problem, a fourth embodiment of the method for updating a customer service problem is provided, in this embodiment, the step S400 generates a new customer service problem according to the clustering result to update the customer service problem set may include:

Step S401, generating new standard questions and similar questions corresponding to the new standard questions according to the clustering result and the customer service question set;

after clustering is carried out on the unresponsive customer service problems by the intelligent customer service robot to obtain a clustering result, the clustering result is combined with existing standard problems and similar problems in the customer service problem set stored in a knowledge base of the FAQ question-answering module, and a new standard problem and similar problems of the new standard problem are generated in a sorting mode.

Step S402, configuring corresponding service answers for the new standard questions and similar questions corresponding to the new standard questions;

Step S403, adding the new standard question, the similar questions corresponding to the new standard question, and the service answer to the customer service question set to update the customer service question set.

After the intelligent customer service robot collates and generates a new standard question and a similar question corresponding to the new standard question, the intelligent customer service robot configures service answers corresponding to the new standard question and the similar question under different scenes respectively, then the intelligent customer service robot establishes an association relation between the new standard question and the corresponding service answer, between the similar question of the new standard question and the corresponding service answer, and then adds the new standard question and the corresponding service answer, and the similar question of the new standard question and the corresponding service answer to a customer service question set stored in a knowledge base of the FAQ question-answering module so as to update the customer service question set.

Further, in an embodiment, after clustering is performed on unresponsive customer service questions to obtain a clustering result, the intelligent customer service robot may further directly output the clustering result to an operator, so as to assist the operator in sorting and generating new standard questions and similar questions of the new standard questions according to the clustering result and existing standard questions and similar questions in a customer service question set stored in a knowledge base of the FAQ question-answering module, and the operator performs a process of configuring service answers corresponding to the new standard questions and the similar questions of the new standard questions, then receives the new standard questions and corresponding service answers input by the operator, similar questions of the new standard questions and corresponding service answers, and adds the new standard questions and similar questions of the new standard questions to the service answers corresponding to the service answers, and similar questions of the new standard questions to the service answers stored in the knowledge base of the FAQ question-answering module, so as to update the customer service question set.

In this embodiment, according to the clustering parameters of the unresponsive customer service problems calculated by using the training convergence similarity model, clustering operation is performed on the unresponsive customer service problems to obtain a clustering result, and then a new standard problem and a similar problem corresponding to the new standard problem are generated by sorting based on the clustering result and the existing customer service problem set; then configuring corresponding service answers for the new standard questions and similar questions corresponding to the new standard questions; and finally, adding the new standard questions, the similar questions corresponding to the new standard questions and the corresponding service answers to the existing customer service question sets to update the customer service question sets. The method and the device realize the self-adaptive learning of the clustering parameters of the unresponsive customer service questions by using a machine learning model, so that the more valuable high-quality questions can be obtained to update the customer service questions when the unresponsive customer service questions are clustered, and the question answering efficiency of the intelligent customer service robot is improved.

In addition, the present invention also provides a customer service problem update system, please refer to fig. 3, fig. 3 is a functional module schematic diagram of the customer service problem update system of the present invention, the customer service problem update system includes:

A construction module 101, configured to construct a problem pair according to a customer service problem in a preset customer service problem set, and combine the problem pair into training data;

The learning module 102 is configured to train a similarity model based on the training data, and calculate a cluster parameter of the unresponsive customer service problem by using the training converged similarity model;

A clustering module 103, configured to perform a clustering operation based on the clustering parameter to obtain a clustering result of the unresponsive customer service problem;

and the updating module 104 is configured to generate a new customer service problem according to the clustering result so as to update the preset customer service problem set.

Preferably, the customer service problem includes a standard problem and a similar problem, and the constructing module 101 includes:

The extraction unit is used for extracting each standard problem and each similar problem corresponding to each standard problem from a preset customer service problem set;

The first construction unit is used for constructing each similar problem pair according to a first standard problem in the standard problems and each similar problem corresponding to the first standard problem;

A first construction unit, configured to construct each dissimilar problem pair according to the first standard problem and each similar problem corresponding to a second standard problem in the standard problems, where the first standard problem is different from the second standard problem;

and the data combination unit is used for correspondingly combining each similar problem pair with each dissimilar problem pair to form each training data.

Preferably, the learning module 102 includes:

the segmentation unit is used for segmenting the training data into a first data set and a second data set;

The model training unit is used for training the similarity model by using the first data set and verifying the trained similarity model by using the second data set;

And the storage unit is used for marking the similarity model passing verification as a training convergence similarity model and storing the training convergence similarity model in the blockchain.

Preferably, the clustering parameters include: the learning module 102 further includes:

the acquisition unit is used for acquiring the problem sentences which do not respond to the customer service problems and the duplicate sentences of the problem sentences, and inputting the problem sentences and the duplicate sentences into a training convergence similarity model;

A calculation unit for obtaining a training convergence similarity model to calculate a vector average value of similarity values between the question sentence and the duplicate sentence;

A first tagging unit configured to take the vector average value as a vector representation of the question sentence;

the input unit is used for traversing the problem sentences to construct sentence pairs, and inputting the sentence pairs into a training convergence similarity model;

And the second marking unit is used for acquiring the similarity value output after the similarity value calculation of the sentence pair by the training convergence similarity model, and taking the similarity value as the similarity value of the problem sentence.

Preferably, the clustering module 103 comprises:

the first clustering operation unit is used for calling a preset distance-based clustering algorithm, and clustering the unresponsive customer service problems by using the similarity value to obtain a clustering result of the unresponsive customer service problems;

And the second clustering operation unit is used for calling a preset clustering algorithm based on sentence representation, and clustering the unresponsive customer service problems by using the vector representation to obtain a clustering result of the unresponsive customer service problems.

Preferably, the updating module 104 includes:

the sorting unit is used for generating new standard questions and similar questions corresponding to the new standard questions according to the clustering result and the customer service question set;

the configuration unit is used for configuring corresponding service answers for the new standard questions and similar questions corresponding to the new standard questions;

and the updating unit is used for adding the new standard questions, the similar questions corresponding to the new standard questions and the service answers to the preset customer service question set so as to update the preset customer service question set.

Preferably, the customer service problem updating system of the present invention further comprises:

the acquisition module is used for acquiring the non-response customer service problems at intervals of preset time length and storing the acquired non-response customer service problems in a blockchain.

The specific implementation manner of the customer service problem updating system of the present invention is basically the same as each embodiment of the customer service problem updating method, and will not be described herein.

Furthermore, the present invention provides a computer storage medium storing one or more programs executable by one or more processors for:

In addition, the one or more programs may be further executable by the one or more processors to:

The specific implementation manner of the computer storage medium of the present invention is basically the same as each embodiment of the update method of the customer service problem, and will not be described herein.

It should be noted that, the blockchain referred to in the present invention is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm, etc. The blockchain (Blockchain), essentially a de-centralized database, is a string of data blocks that are generated in association using cryptographic methods, each of which contains information from a batch of network transactions for verifying the validity (anti-counterfeit) of its information and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.

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) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, 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

1. The method for updating the customer service problem is characterized by comprising the following steps:

training a similarity model based on the training data, and calculating a clustering parameter of the unresponsive customer service problem by utilizing the training converged similarity model, wherein the clustering parameter comprises a vector representation and a similarity value of a problem sentence of the unresponsive customer service problem;

generating new customer service questions according to the clustering result to update the preset customer service question set;

the customer service questions comprise standard questions and similar questions, and the step of constructing question pairs according to customer service questions in a preset customer service question set and combining the question pairs into training data comprises the following steps:

Constructing each dissimilar problem pair according to the first standard problem and each similar problem corresponding to a second standard problem in the standard problems, wherein the first standard problem is different from the second standard problem or the first standard problem is the same as the second standard problem;

correspondingly combining each similar problem pair with each dissimilar problem pair according to the rule of one similar problem pair and one dissimilar problem pair to form each training data;

or combining each similar problem pair with each dissimilar problem pair according to one similar problem pair and a plurality of dissimilar problem pairs to form each training data.

2. A method of updating a customer service problem according to claim 1, wherein the step of training a similarity model based on the training data comprises:

splitting the training data into a first data set and a second data set;

3. A method for updating customer service problems according to claim 1, wherein the step of calculating cluster parameters of unresponsive customer service problems using a training converged similarity model comprises:

4. A customer service problem updating method according to claim 3, wherein the step of performing a clustering operation based on the clustering parameters to obtain a clustering result of the non-responding customer service problem comprises:

5. The method for updating a customer service problem according to claim 1, wherein the step of generating a new customer service problem according to the clustering result to update the preset customer service problem set comprises:

6. The method for updating a customer service problem according to claim 1, wherein the method for updating a customer service problem further comprises:

7. The customer service problem updating system is characterized by comprising the following steps:

The construction module is used for constructing a problem pair according to customer service problems in a preset customer service problem set and combining the problem pair into training data;

The learning module is used for training a similarity model based on the training data and calculating clustering parameters of the unresponsive customer service questions by utilizing the training converged similarity model, wherein the clustering parameters comprise vector representations and similarity values of the question sentences of the unresponsive customer service questions;

The clustering module is used for carrying out clustering operation based on the clustering parameters to obtain a clustering result of the unresponsive customer service problem;

the updating module is used for generating new customer service problems according to the clustering result so as to update the preset customer service problem set;

The customer service problems include standard problems and similar problems, and the construction module comprises:

a first construction unit, configured to construct each dissimilar problem pair according to the first standard problem and each similar problem corresponding to a second standard problem in the standard problems, where the first standard problem is different from the second standard problem, or the first standard problem is the same as the second standard problem;

a data combination unit, configured to correspondingly combine each of the similar problem pairs with each of the dissimilar problem pairs according to a rule of one of the similar problem pairs and one of the dissimilar problem pairs to form each training data;

8. A terminal device, characterized in that the terminal device comprises: memory, processor, communication bus, and update program for customer service problems stored on the memory,

the processor is configured to execute an update procedure of an internet-based customer service problem to implement the steps of the method for updating a customer service problem according to any one of claims 1 to 6.

9. A computer storage medium, wherein a customer service problem update program is stored on the computer storage medium, and wherein the customer service problem update program, when executed by a processor, implements the steps of the customer service problem update method according to any one of claims 1 to 6.