CN117093685A - Intelligent customer service application method, device, equipment and storage medium - Google Patents
Intelligent customer service application method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN117093685A CN117093685A CN202310954778.5A CN202310954778A CN117093685A CN 117093685 A CN117093685 A CN 117093685A CN 202310954778 A CN202310954778 A CN 202310954778A CN 117093685 A CN117093685 A CN 117093685A
- Authority
- CN
- China
- Prior art keywords
- data
- response
- dialogue model
- dialogue
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000004044 response Effects 0.000 claims abstract description 229
- 238000012549 training Methods 0.000 claims abstract description 119
- 238000004458 analytical method Methods 0.000 claims abstract description 55
- 238000007405 data analysis Methods 0.000 claims abstract description 25
- 238000012544 monitoring process Methods 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims description 22
- 238000012795 verification Methods 0.000 claims description 19
- 230000003190 augmentative effect Effects 0.000 claims description 17
- 238000005192 partition Methods 0.000 claims description 17
- 238000011160 research Methods 0.000 claims description 16
- 238000013507 mapping Methods 0.000 claims description 14
- 238000013434 data augmentation Methods 0.000 claims description 10
- 238000011835 investigation Methods 0.000 claims description 8
- 238000004088 simulation Methods 0.000 claims description 8
- 238000013519 translation Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 230000003416 augmentation Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 5
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000003058 natural language processing Methods 0.000 description 23
- 238000005516 engineering process Methods 0.000 description 11
- 230000008451 emotion Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000010801 machine learning Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 238000011084 recovery Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000003032 molecular docking Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Databases & Information Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Game Theory and Decision Science (AREA)
- Human Computer Interaction (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to the technical field of information processing, in particular to an intelligent customer service application method, device, equipment and storage medium, which comprise the steps of obtaining service data of a target enterprise, and carrying out data analysis on the service data to obtain reply key analysis data; collecting common problem scheme information and forming basic response data; constructing a response knowledge base according to the basic response data and the response key analysis data; constructing a pre-dialogue model based on NLP, and performing iterative training on the pre-dialogue model through a response knowledge base to obtain a dialogue model; monitoring the dialogue model to acquire user feedback data, and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base; updating and training the dialogue model according to the response knowledge updating library to obtain an improved dialogue model; by monitoring the dialogue model, the dialogue model can be continuously and iteratively optimized during training, so that the dialogue model can be suitable for continuously changing business scenes.
Description
Technical Field
The present invention relates to the field of information processing technologies, and in particular, to an intelligent customer service application method, apparatus, device, and storage medium.
Background
At present, the logistics customer service generally adopts intelligent customer service to process a work order, but the existing intelligent customer service call recovery can not completely cover all problems initiated by customers, so that the problem that the intelligent customer service recovery is not covered in a closed loop exists, and the capability of solving the problem of the intelligent customer service and the experience of a user are greatly influenced.
It can be seen that there is a need for improvements and improvements in the art.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide an intelligent customer service application method, device, equipment and storage medium, which can analyze emotion and context of a user and continuously conduct iterative training on intelligent customer service in a conversation process so as to improve the response capability of the intelligent customer service.
The first aspect of the invention provides an intelligent customer service application method, which comprises the following steps: acquiring service data of a target enterprise, and performing data analysis on the service data to obtain reply key analysis data; collecting common problem scheme information and forming basic response data; constructing a response knowledge base according to the basic response data and the response key analysis data; constructing a pre-dialogue model based on NLP, and performing iterative training on the pre-dialogue model through a response knowledge base to obtain a dialogue model; monitoring the dialogue model to acquire user feedback data, and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base; and updating and training the dialogue model according to the response knowledge updating library to obtain an improved dialogue model.
Optionally, in a second implementation manner of the first aspect of the present invention, the acquiring service data of the target enterprise and performing data analysis on the service data to obtain reply key analysis data includes: pre-constructing a target enterprise investigation table, and acquiring service data of each target enterprise according to the target enterprise investigation table; carrying out data analysis on the service data to obtain customer feedback data, complaint record data and market research data; and carrying out scene simulation reply according to the customer feedback data, the complaint record data and the market research data to obtain reply key analysis data.
Optionally, in a third implementation manner of the first aspect of the present invention, the constructing a response knowledge base according to the basic response data and the reply emphasis analysis data includes: a basic knowledge base is constructed in advance, and the basic knowledge base is partitioned according to different response scenes; carrying out standardized processing on the basic response data and the response key analysis data to obtain question data and response data; according to the response scene, storing the problem data and the response data into corresponding partitions in the basic knowledge base; and establishing a mapping relation between the problem data in each partition and the corresponding answer data to obtain an answer knowledge base.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the constructing a pre-dialogue model based on NLP, and performing iterative training on the pre-dialogue model through a response knowledge base to obtain a dialogue model, includes: constructing a pre-dialogue model based on NLP; extracting question data and answer data from an answer knowledge base; generating a plurality of question flow charts according to the question data; addressing the answering data according to the mapping relation to obtain an answering flow table corresponding to the questioning flow table; generating a pre-training set and a verification set according to the questioning flow sheet and the answering flow sheet; performing data augmentation processing on the pre-training set to obtain a training set; performing iterative training on the pre-dialogue model according to the training set to obtain a dialogue training model; and performing verification and evaluation on the dialogue training model according to the verification set, and taking the dialogue training model with the highest score as the dialogue model.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the performing data augmentation processing on the pre-training set to obtain a training set includes: extracting the character features of questions of each question flow chart according to the pre-training set; carrying out semantic conversion on the character features of the problem according to the semantic converter so as to obtain first augmentation character data; performing back translation processing on the first augmented character data to obtain second augmented character data; word noise is added to the second augmented character data to obtain a training set.
Optionally, in a sixth implementation manner of the first aspect of the present invention, the monitoring the dialogue model to collect user feedback data, and updating the response knowledge base according to the user feedback data to obtain a response knowledge update base includes: tracking performance indexes, log records and user feedback data of the dialogue model in real time; acquiring a response flow, response time and response accuracy according to the performance index and the log record; data analysis is carried out on the user feedback data and the response flow so as to obtain improved response data; and storing the improved response data into the corresponding partition of the response knowledge base according to the response scene to form a response knowledge updating base.
Optionally, in a seventh implementation manner of the first aspect of the present invention, the updating training of the dialogue model according to the response knowledge updating base to obtain an improved dialogue model includes: extracting customer service operation data and basic response data according to the response knowledge updating base; clustering the client speech data and the basic response data to obtain one or more groups of clustered data; determining one or more dialogue strategies according to the clustering data, and taking the dialogue strategies as an updated training set; and updating and training the dialogue model through the updating training set so as to obtain an improved dialogue model.
The second aspect of the present invention provides an intelligent customer service application device, comprising: the analysis module is used for acquiring the business data of the target enterprise and carrying out data analysis on the business data to obtain reply key analysis data; the acquisition module is used for acquiring information of a common problem scheme and forming basic response data; the database building module is used for building a response knowledge base according to the basic response data and the response key analysis data; the modeling module is used for constructing a pre-dialogue model based on the NLP, and carrying out iterative training on the pre-dialogue model through the response knowledge base so as to obtain a dialogue model; the monitoring module is used for monitoring the dialogue model to acquire user feedback data and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base; and the optimization module is used for updating and training the dialogue model according to the response knowledge updating base so as to obtain an improved dialogue model.
Optionally, in a first implementation manner of the second aspect of the present invention, the analysis module includes: the system comprises a presetting unit, a target enterprise search table, a control unit and a control unit, wherein the presetting unit is used for presetting a target enterprise search table and acquiring service data of each target enterprise according to the target enterprise search table; the analysis unit is used for carrying out data analysis on the service data to obtain customer feedback data, complaint record data and market research data; and the simulation unit is used for carrying out scene simulation reply according to the customer feedback data, the complaint record data and the market research data so as to obtain reply key analysis data.
Optionally, in a second implementation manner of the second aspect of the present invention, the library creating module includes: the partition unit is used for pre-constructing a basic knowledge base and partitioning the basic knowledge base according to different response scenes; the standardized unit is used for carrying out standardized processing on the basic response data and the response key analysis data so as to obtain problem data and response data; the storage unit is used for storing the problem data and the answer data into corresponding partitions in the basic knowledge base according to the answer scene; and the mapping unit is used for establishing a mapping relation between the problem data in each partition and the corresponding answer data so as to obtain an answer knowledge base.
Optionally, in a third implementation manner of the second aspect of the present invention, the modeling module includes: the modeling unit is used for constructing a pre-dialogue model based on NLP; an extraction unit for extracting question data and answer data from the answer knowledge base; the first generation unit is used for generating a plurality of question flow charts according to the question data; the addressing unit is used for addressing the answering data according to the mapping relation to obtain an answering flow table corresponding to the questioning flow table; the second generation unit is used for generating a pre-training set and a verification set according to the questioning flow sheet and the answering flow sheet; the augmentation unit is used for carrying out data augmentation processing on the pre-training set so as to obtain a training set; the first training unit is used for carrying out iterative training on the pre-dialogue model according to the training set so as to obtain a dialogue training model; and the verification unit is used for verifying and evaluating the dialogue training model according to the verification set, and taking the dialogue training model with the highest score as the dialogue model.
Optionally, in a fourth implementation manner of the second aspect of the present invention, the amplifying unit includes: the extraction part is used for extracting the problem character characteristics of each question flow chart according to the pre-training set; a conversion part for performing semantic conversion on the problematic character features according to the semantic converter to obtain first augmented character data; a back translation part for performing back translation processing on the first augmented character data to obtain second augmented character data; and an adding part for adding word noise to the second augmented character data to obtain a training set.
Optionally, in a fifth implementation manner of the second aspect of the present invention, the monitoring module includes: the tracking unit is used for tracking performance indexes, log records and user feedback data of the dialogue model in real time; the acquisition unit is used for acquiring a response flow, response time and response accuracy according to the performance index and the log record; the feedback unit is used for carrying out data analysis on the user feedback data and the response flow so as to obtain improved response data; and the updating unit is used for storing the improved response data into the corresponding partition of the response knowledge base according to the response scene so as to form a response knowledge updating base.
Optionally, in a sixth implementation manner of the second aspect of the present invention, the optimizing module includes: the telephone operation unit is used for extracting customer service telephone operation data and basic response data according to the response knowledge updating base; the clustering unit is used for clustering the client voice operation data and the basic response data to obtain one or more groups of clustering data; the determining unit is used for determining one or more dialogue strategies according to the clustering data and taking the dialogue strategies as an updating training set; and the second training unit is used for updating and training the dialogue model through the updating training set so as to obtain an improved dialogue model.
A third aspect of the present invention provides an intelligent customer service application device, including: a memory and at least one processor, the memory having instructions stored therein; at least one of the processors invokes the instructions in the memory to cause the intelligent customer service application device to perform the steps of the intelligent customer service application method of any of the above.
A fourth aspect of the present invention provides a computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the steps of the intelligent customer service application method of any of the above.
According to the technical scheme, a response knowledge base is constructed according to the basic response data and the response key analysis data, a data base is provided for intelligent dialogue of intelligent customer service through the response knowledge base, so that the intelligent customer service and a user are more personified during dialogue, and good data support is provided for the intelligent customer service with multiple rounds of complex dialogue capability; constructing a pre-dialogue model which can use natural language processing and machine learning technology based on an NLP technology, and then carrying out iterative training on the pre-dialogue model through data in a response knowledge base so as to improve the response accuracy and response speed of the dialogue model, so that the dialogue model can meet the continuously changing user requirements and business scenes; the dialogue model is monitored and tracked to obtain continuously-changing user feedback data, the user feedback data is added into the response knowledge base, the response knowledge base is updated, the dialogue model can be continuously and iteratively optimized during training, continuous progress of the dialogue model is ensured to be suitable for continuously-changing business scenes, and the dialogue model can be gradually enabled to tend to return to a closed-loop coverage state.
Drawings
FIG. 1 is a first flowchart of an intelligent customer service application method according to an embodiment of the present invention;
FIG. 2 is a second flowchart of an intelligent customer service application method according to an embodiment of the present invention;
FIG. 3 is a third flowchart of an intelligent customer service application method according to an embodiment of the present invention;
FIG. 4 is a fourth flowchart of an intelligent customer service application method according to an embodiment of the present invention;
FIG. 5 is a fifth flowchart of an intelligent customer service application method according to an embodiment of the present invention;
FIG. 6 is a sixth flowchart of an intelligent customer service application method according to an embodiment of the present invention;
fig. 7 is a seventh flowchart of an intelligent customer service application method according to an embodiment of the present invention
Fig. 8 is a schematic structural diagram of an intelligent customer service application device according to an embodiment of the present invention;
fig. 9 is a schematic structural diagram of another intelligent customer service application device according to an embodiment of the present invention;
fig. 10 is a schematic structural diagram of an intelligent customer service application device according to an embodiment of the present invention.
Detailed Description
The invention provides an intelligent customer service application method, a device, equipment and a storage medium, wherein a response knowledge base is constructed according to basic response data and response key analysis data, a data base is provided for intelligent dialogue of intelligent customer service through the response knowledge base, so that the intelligent customer service and a user are more personified during dialogue, and good data support is provided for the intelligent customer service with multiple rounds of complex dialogue capability; constructing a pre-dialogue model which can use natural language processing and machine learning technology based on an NLP technology, and then carrying out iterative training on the pre-dialogue model through data in a response knowledge base so as to improve the response accuracy and response speed of the dialogue model, so that the dialogue model can meet the continuously changing user requirements and business scenes; the dialogue model is monitored and tracked to obtain continuously-changing user feedback data, the user feedback data is added into the response knowledge base, the response knowledge base is updated, the dialogue model can be continuously and iteratively optimized during training, continuous progress of the dialogue model is ensured to be suitable for continuously-changing business scenes, and the dialogue model can be gradually enabled to tend to return to a closed-loop coverage state.
The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, a specific flow of an embodiment of the present invention is described below with reference to fig. 1, where an embodiment of the method for intelligent customer service application in the embodiment of the present invention includes:
101. acquiring service data of a target enterprise, and performing data analysis on the service data to obtain reply key analysis data;
102. Collecting common problem scheme information and forming basic response data;
103. constructing a response knowledge base according to the basic response data and the response key analysis data;
in this embodiment, firstly, the business data of the main docking client (i.e. the target enterprise) needs to be acquired, and the needs of the enterprise are further known according to the business data of the target enterprise, so as to obtain the reply key analysis data for the specific enterprise. The reply key analysis data reflects the data such as common problems, pain point information and the like of the target enterprise; then common problem scheme information (namely basic problems encountered in the daily customer service flow) is collected from channels such as manual customer service and the like to form basic response data; and finally, constructing a response knowledge base according to the basic response data and the response key analysis data, providing a data base for intelligent conversations of the intelligent customer service through the response knowledge base, enabling the intelligent customer service and a user to be more personified during conversations, and providing good data support for the intelligent customer service with multiple rounds of complex conversational capability.
It should be noted that, analyzing the business data of the target enterprise refers to performing deep knowledge and evaluation on the business target, the service scope and the client requirement of the enterprise to determine the specific requirements that the intelligent customer service system needs to meet. The following are several common methods of interpreting the analysis of business requirements:
1. Cooperation with business segments: cooperation and communication with various business departments within the enterprise, including sales, customer service, operations, etc. Through communication with the relevant personnel, their views, experience and knowledge of the customer's needs are known. This may provide insight into the business aspects and feedback of specific problems.
2. Data analysis: information about common problems and pain points of customers is obtained through analysis of existing data of enterprises, including customer feedback, complaint records, market research data and the like. The data analysis may reveal trends, hotspots, and priorities of customer needs, helping to determine key functions of the intelligent customer service system and emphasis on the speech recovery coverage.
3. User feedback: feedback opinions, suggestions, and complaints of the user on the products and services of the enterprise are collected and analyzed. This may be done by way of online surveys, user satisfaction surveys, social media monitoring, and the like. User feedback is an important way to directly understand customer needs and common problems, and can provide valuable references and improvement directions for intelligent customer service systems.
Business demands of enterprises and common problems of clients can be known in depth through cooperation with business departments, data analysis, user feedback and the like. This helps determine the critical functionality of the intelligent customer service system, knowledge base content, and coverage of the session replies. Meanwhile, data are continuously collected and analyzed, and the intelligent customer service system is timely adjusted and optimized to adapt to continuously changing business requirements and customer expectations.
104. Constructing a pre-dialogue model based on NLP, and performing iterative training on the pre-dialogue model through a response knowledge base to obtain a dialogue model;
in this embodiment, a pre-dialogue model that can use natural language processing and machine learning techniques is pre-built based on the NLP technique, and then the pre-dialogue model is iteratively trained by data in the response knowledge base, so as to improve the response accuracy and response speed of the dialogue model, so that the dialogue model can meet the continuously changing user requirements and business scenarios.
Note that NLP is a technology related to understanding and processing human language. In developing a dialog model, NLP-based techniques are used to parse and analyze user inputs to identify key information, intent, and entities. This includes identifying the subject matter of the problem, the user's needs, and the products or services that may be involved.
And (5) intention recognition: the intention recognition is to recognize the purpose or intention of a user's question by analyzing the language expression of the user. It can help the intelligent customer service system determine the answer that the user really needs and map it to the appropriate solution.
Entity identification: entity identification refers to identifying specific things, keywords, or named entities from a user's question. Through entity recognition, the intelligent customer service system can better understand important information in user questions and provide relevant answers or suggestions.
Dialog flow management: dialog flow management is a process that ensures consistency and consistency of dialog. When developing a dialogue model, a proper dialogue flow needs to be designed, so that the system can process multiple rounds of dialogue and establish a connection among a plurality of questions, and the user is ensured to obtain accurate and consistent answers.
Emotion analysis: emotion analysis is the understanding of the emotional state of a user by analyzing emotion and emotion in the user's language. The dialogue model can identify the emotion tendency of the user by using emotion analysis technology, so that the emotion requirement of the user can be responded better, and corresponding answers and support are provided.
105. Monitoring the dialogue model to acquire user feedback data, and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base;
106. and updating and training the dialogue model according to the response knowledge updating library to obtain an improved dialogue model.
In this embodiment, the dialogue model is monitored and tracked to obtain continuously-changing user feedback data (i.e. service scene, user requirement, question-answer flow, etc.), and the user feedback data is added into the answer knowledge base to update the answer knowledge base, so that the dialogue model can be continuously and iteratively optimized during training, continuous progress of the dialogue model is ensured to be suitable for the continuously-changing service scene, and the dialogue model can be gradually made to tend to return to the closed-loop coverage state.
In the embodiment of the invention, a response knowledge base is constructed according to the basic response data and the response key analysis data, and a data basis is provided for intelligent dialogue of intelligent customer service through the response knowledge base, so that the intelligent customer service and a user are more personified during dialogue, and good data support is provided for the intelligent customer service with multiple rounds of complex dialogue capability; constructing a pre-dialogue model which can use natural language processing and machine learning technology based on an NLP technology, and then carrying out iterative training on the pre-dialogue model through data in a response knowledge base so as to improve the response accuracy and response speed of the dialogue model, so that the dialogue model can meet the continuously changing user requirements and business scenes; the dialogue model is monitored and tracked to obtain continuously-changing user feedback data, the user feedback data is added into the response knowledge base, the response knowledge base is updated, the dialogue model can be continuously and iteratively optimized during training, continuous progress of the dialogue model is ensured to be suitable for continuously-changing business scenes, and the dialogue model can be gradually enabled to tend to return to a closed-loop coverage state.
Referring to fig. 2, a second embodiment of the intelligent customer service application method according to the present invention includes:
201. Pre-constructing a target enterprise investigation table, and acquiring service data of each target enterprise according to the target enterprise investigation table;
202. carrying out data analysis on the service data to obtain customer feedback data, complaint record data and market research data;
203. and carrying out scene simulation reply according to the customer feedback data, the complaint record data and the market research data to obtain reply key analysis data.
In the embodiment, market research is performed in advance by service personnel to construct a target enterprise research table, and then all service data of each target enterprise are called according to a list of the target enterprise research table; carrying out data analysis on the service data to obtain client feedback data, complaint record data and market research data of related clients so as to determine key functions of intelligent customer service and key points of speech recovery coverage; and, the interactive scene simulation is performed according to the relevant information of the customer feedback data, the complaint record data and the market research data, so as to obtain the reply key analysis data (namely, the relevant simulated response flow of the target enterprise).
Referring to fig. 3, a third embodiment of the intelligent customer service application method according to the present invention includes:
301. A basic knowledge base is constructed in advance, and the basic knowledge base is partitioned according to different response scenes;
302. carrying out standardized processing on the basic response data and the response key analysis data to obtain question data and response data;
303. according to the response scene, storing the problem data and the response data into corresponding partitions in the basic knowledge base;
in this embodiment, when the knowledge response knowledge base is constructed, the basic knowledge base needs to be partitioned in advance according to different response scenes, so as to manage the data of the different response scenes; after receiving the basic response data and the response key analysis data, the basic response data and the response key analysis data are subjected to standardized processing, so that the basic response data and the response key analysis data are converted into character features which are clear in expression and easy to understand, the character features can be converted into a multilingual state so as to meet the use requirements of different users, and finally, the response scenes of different question data and response data are classified and stored so as to be stored in corresponding partitions in a basic knowledge base, so that a developer can manage the response knowledge base conveniently.
304. And establishing a mapping relation between the problem data in each partition and the corresponding answer data to obtain an answer knowledge base.
In this embodiment, a mapping relationship needs to be established for each question data and corresponding answer data, so that the corresponding answer data can be addressed quickly according to the question data, and the response speed of intelligent customer service is improved.
Referring to fig. 4, a fourth embodiment of the intelligent customer service application method in the embodiment of the present invention includes:
401. constructing a pre-dialogue model based on NLP;
402. extracting question data and answer data from an answer knowledge base;
403. generating a plurality of question flow charts according to the question data;
404. addressing the answering data according to the mapping relation to obtain an answering flow table corresponding to the questioning flow table;
405. generating a pre-training set and a verification set according to the questioning flow sheet and the answering flow sheet;
406. performing data augmentation processing on the pre-training set to obtain a training set;
407. performing iterative training on the pre-dialogue model according to the training set to obtain a dialogue training model;
408. and performing verification and evaluation on the dialogue training model according to the verification set, and taking the dialogue training model with the highest score as the dialogue model.
In this embodiment, the construction of the dialogue model is divided into four steps, as follows: firstly, constructing a pre-dialogue model according to NLP; secondly, constructing a pre-training set, wherein during construction, different question flow charts are generated according to random combination of question data to form a simulated query initiation flow of a user, then the questions in the question flow charts are addressed according to the mapping relation to obtain corresponding answer flow charts, and a pre-training set and a verification set are generated according to the question flow charts and the answer flow charts to perform iterative training and score verification on a pre-dialogue model; thirdly, data augmentation is carried out on the pre-training set so as to improve the robustness of the training set; training and verifying the pre-dialogue model, training the pre-dialogue model according to the training set after data augmentation, verifying and evaluating the trained pre-dialogue model according to the verification set, and taking the dialogue training model with the highest score as a dialogue model; the application range and the response speed of the dialogue model are effectively improved through continuous iterative training.
Referring to fig. 5, a fifth embodiment of the intelligent customer service application method according to the present invention includes:
501. extracting the character features of questions of each question flow chart according to the pre-training set;
502. carrying out semantic conversion on the character features of the problem according to the semantic converter so as to obtain first augmentation character data;
503. performing back translation processing on the first augmented character data to obtain second augmented character data;
504. word noise is added to the second augmented character data to obtain a training set.
In this embodiment, before the pre-training set is subjected to data augmentation, the problem character features of the question flow chart are extracted in advance, and then the problem character features are more diversified in the modes of semantic conversion, back translation processing and word noise addition, so that the overall robustness of the training set is improved, and the dialogue model after training can be more sensitive.
Referring to fig. 6, a sixth embodiment of the intelligent customer service application method according to the present invention includes:
601. tracking performance indexes, log records and user feedback data of the dialogue model in real time;
602. acquiring a response flow, response time and response accuracy according to the performance index and the log record;
603. Data analysis is carried out on the user feedback data and the response flow so as to obtain improved response data;
604. storing the improved response data into the corresponding subareas of the response knowledge base according to the response scene to form a response knowledge updating base;
in this embodiment, by tracking the whole flow of the dialogue model during operation in real time, performance indexes, diary records and user feedback data of the dialogue model are obtained, so that a developer can grasp the use state and use performance of the dialogue model, can quickly find pain points of the dialogue model for optimization, and perform data analysis according to the user feedback data and the response flow to optimize and improve the response flow, so as to obtain improved response data, and finally store the improved response data in a response knowledge base to form a response knowledge update base, and can perform optimization training on the dialogue model through the response knowledge update base, so as to improve accuracy of the dialogue model in answering questions.
Referring to fig. 7, a seventh embodiment of the intelligent customer service application method according to the present invention includes:
701. extracting customer service operation data and basic response data according to the response knowledge updating base;
702. clustering the client speech data and the basic response data to obtain one or more groups of clustered data;
703. Determining one or more dialogue strategies according to the clustering data, and taking the dialogue strategies as an updated training set;
704. updating and training the dialogue model through the updating training set to obtain an improved dialogue model;
in this embodiment, during the optimization training, the dialogue operation data and the basic response data are required to be randomly combined in a clustering manner, so as to form one or more dialogue strategies with certain randomness, improve the robustness of an update training set, update and train the dialogue model through the update training set, effectively improve the accuracy of the dialogue model during the reply, effectively improve the use experience of the user, discover and solve problems in time, optimize the performance and functions of the dialogue model, enable the dialogue model to better meet the requirements of the user, and establish good interaction and communication with the user. Iterative optimization is also an environment that ensures continuous progress of the dialog model and adaptation to changes.
The method for intelligent customer service application in the embodiment of the present invention is described above, and the device for intelligent customer service application in the embodiment of the present invention is described below, referring to fig. 8, where an embodiment of the device for intelligent customer service application in the embodiment of the present invention includes:
The analysis module 801 is configured to obtain service data of a target enterprise, and perform data analysis on the service data to obtain reply key analysis data;
the acquisition module 802 is used for acquiring information of a common problem scheme and forming basic response data;
the database building module 803 is configured to build a response knowledge base according to the basic response data and the response key analysis data;
the modeling module 804 is configured to construct a pre-dialogue model based on the NLP, and perform iterative training on the pre-dialogue model through the response knowledge base to obtain a dialogue model;
the monitoring module 805 is configured to monitor the dialogue model to collect user feedback data, and update the response knowledge base according to the user feedback data to obtain a response knowledge update base;
and an optimizing module 806, configured to update and train the dialogue model according to the response knowledge update base, so as to obtain an improved dialogue model.
In this embodiment, the analysis module 801 and the library building module 803 construct a response knowledge base according to the basic response data and the response key analysis data, and provide a data basis for the intelligent dialogue of the intelligent customer service through the response knowledge base, so that the intelligent customer service and the user are more personified during the dialogue, and good data support is provided for the intelligent customer service with multiple rounds of complex dialogue capability; the library building module 804 builds a pre-dialogue model based on NLP technology, which can use natural language processing and machine learning technology, and then carries out iterative training on the pre-dialogue model through data in the response knowledge base so as to improve the response accuracy and response speed of the dialogue model, so that the dialogue model can meet the continuously changing user requirements and business scenes; the monitoring module 805 and the optimizing module 806 monitor and track the dialogue model to obtain continuously changing user feedback data, and add the user feedback data into the response knowledge base, update the response knowledge base, so that the dialogue model can be continuously and iteratively optimized during training, and the dialogue model can be ensured to continuously progress to adapt to continuously changing business scenes, and gradually can tend to return to a closed-loop coverage state.
Referring to fig. 9, another embodiment of the intelligent customer service application apparatus according to the present invention includes:
the analysis module 801 is configured to obtain service data of a target enterprise, and perform data analysis on the service data to obtain reply key analysis data;
the acquisition module 802 is used for acquiring information of a common problem scheme and forming basic response data;
the database building module 803 is configured to build a response knowledge base according to the basic response data and the response key analysis data;
the modeling module 804 is configured to construct a pre-dialogue model based on the NLP, and perform iterative training on the pre-dialogue model through the response knowledge base to obtain a dialogue model;
the monitoring module 805 is configured to monitor the dialogue model to collect user feedback data, and update the response knowledge base according to the user feedback data to obtain a response knowledge update base;
and an optimizing module 806, configured to update and train the dialogue model according to the response knowledge update base, so as to obtain an improved dialogue model.
In this embodiment, the analysis module 801 includes: the presetting unit 8011 is configured to pre-construct a target enterprise investigation table, and obtain service data of each target enterprise according to the target enterprise investigation table; the analysis unit 8012 is used for carrying out data analysis on the service data to obtain customer feedback data, complaint record data and market research data; the simulation unit 8013 is configured to perform scene simulation reply according to the customer feedback data, the complaint recording data, and the market research data, so as to obtain reply key analysis data.
In this embodiment, the library creating module 803 includes: the partition unit 8031 is used for pre-constructing a basic knowledge base and partitioning the basic knowledge base according to different response scenes; a standardization unit 8032, configured to perform standardization processing on the basic response data and the reply key analysis data, so as to obtain question data and answer data; the storage unit 8033 is used for storing the question data and the answer data into corresponding partitions in the basic knowledge base according to the answer scene; and a mapping unit 8034, configured to establish a mapping relationship between the problem data in each partition and the corresponding answer data, so as to obtain an answer knowledge base.
In this embodiment, the modeling module 804 includes: a modeling unit 8041 for constructing a pre-dialogue model based on the NLP; an extracting unit 8042 for extracting question data and answer data from the answer knowledge base; a first generating unit 8043 for generating a plurality of question flowcharts according to the question data; addressing unit 8044, configured to address the answer data according to the mapping relationship, so as to obtain an answer flow table corresponding to the question flow table; a second generating unit 8045, configured to generate a pre-training set and a verification set according to the question flow table and the answer flow table; an augmentation unit 8046, configured to perform data augmentation processing on the pre-training set to obtain a training set; a first training unit 8047, configured to perform iterative training on the pre-dialogue model according to the training set, so as to obtain a dialogue training model; the verification unit 8048 is configured to perform verification and evaluation on the dialogue training model according to the verification set, and take the dialogue training model with the highest score as the dialogue model.
In this embodiment, the augmentation unit 8046 includes: an extracting unit 80461 for extracting the question character features of each question flow table according to the pre-training set; a conversion section 80462 for semantically converting the problematic character features according to the semantic converter to obtain first augmented character data; a back translation unit 80463 for performing back translation processing on the first augmented character data to obtain second augmented character data; an adding section 80464 for adding word noise to the second augmented character data to obtain a training set.
In this embodiment, the monitoring module 805 includes: the tracking unit 8051 is used for tracking performance indexes, log records and user feedback data of the dialogue model in real time; an obtaining unit 8052, configured to obtain a response flow, response time, and response accuracy according to the performance index and the log record; the feedback unit 8053 is configured to perform data analysis on the user feedback data and the response flow, so as to obtain improved response data; and the updating unit 8054 is used for storing the improved response data into the corresponding partition of the response knowledge base according to the response scene so as to form a response knowledge updating base.
In this embodiment, the optimizing module 806 includes: the telephone operation unit 8061 is used for extracting customer service telephone operation data and basic response data according to the response knowledge update base; the clustering unit 8062 is used for clustering the client speech data and the basic response data to obtain one or more groups of clustered data; a determining unit 8063, configured to determine one or more dialogue policies according to the cluster data, and use the dialogue policies as an update training set; a second training unit 8064 for updating and training the dialogue model by updating the training set to obtain an improved dialogue model
The intelligent customer service application device in the embodiment of the present invention is described in detail from the point of view of modularized functional entities in fig. 8 and fig. 9, and the intelligent customer service application device in the embodiment of the present invention is described in detail from the point of view of hardware processing.
Fig. 10 is a schematic structural diagram of an intelligent customer service application device provided in an embodiment of the present invention, where the intelligent customer service application device 900 may have a relatively large difference due to different configurations or performances, and may include one or more processors (central processing units, CPU) 910 (e.g., one or more processors) and a memory 920, and one or more storage media 930 (e.g., one or more mass storage devices) storing application programs 933 or data 932. Wherein the memory 920 and storage medium 930 may be transitory or persistent storage. The program stored on the storage medium 930 may include one or more modules (not shown), each of which may include a series of instruction operations for the intelligent customer service application device 900. Still further, the processor 910 may be configured to communicate with a storage medium 930, and execute a series of instruction operations in the storage medium 930 on the intelligent service application device 900 to implement the steps of the intelligent service application method provided in the above method embodiments.
The intelligent customer service application device 900 may also include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input/output interfaces 960, and/or one or more operating systems 931, such as Windows Serve, mac OS X, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the intelligent customer service application device architecture shown in fig. 10 is not limiting on the intelligent customer service based application device and may include more or fewer components than shown, or may combine certain components, or may be a different arrangement of components.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, and may also be a volatile computer readable storage medium, where instructions are stored that, when executed on a computer, cause the computer to perform the steps of the intelligent customer service application method.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the system or apparatus and unit described above may refer to the corresponding process in the foregoing method embodiment, which is not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Finally, it should be noted that: the foregoing is merely a preferred example of the present invention, and the present invention is not limited thereto, but it is to be understood that modifications and equivalents of some of the technical features described in the foregoing embodiments may be made by those skilled in the art, although the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. An intelligent customer service application method is characterized by comprising the following steps:
acquiring service data of a target enterprise, and performing data analysis on the service data to obtain reply key analysis data;
collecting common problem scheme information and forming basic response data;
constructing a response knowledge base according to the basic response data and the response key analysis data;
constructing a pre-dialogue model based on NLP, and performing iterative training on the pre-dialogue model through a response knowledge base to obtain a dialogue model;
monitoring the dialogue model to acquire user feedback data, and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base;
and updating and training the dialogue model according to the response knowledge updating library to obtain an improved dialogue model.
2. The intelligent customer service application method according to claim 1, wherein the obtaining the service data of the target enterprise and performing data analysis on the service data to obtain reply key analysis data comprises:
pre-constructing a target enterprise investigation table, and acquiring service data of each target enterprise according to the target enterprise investigation table;
carrying out data analysis on the service data to obtain customer feedback data, complaint record data and market research data;
And carrying out scene simulation reply according to the customer feedback data, the complaint record data and the market research data to obtain reply key analysis data.
3. The intelligent customer service application method according to claim 1, wherein the constructing a response knowledge base according to the basic response data and the response emphasis analysis data comprises:
a basic knowledge base is constructed in advance, and the basic knowledge base is partitioned according to different response scenes;
carrying out standardized processing on the basic response data and the response key analysis data to obtain question data and response data;
according to the response scene, storing the problem data and the response data into corresponding partitions in the basic knowledge base;
and establishing a mapping relation between the problem data in each partition and the corresponding answer data to obtain an answer knowledge base.
4. The intelligent customer service application method according to claim 3, wherein the constructing a pre-dialogue model based on NLP and performing iterative training on the pre-dialogue model through a response knowledge base to obtain the dialogue model comprises:
constructing a pre-dialogue model based on NLP;
extracting question data and answer data from an answer knowledge base;
generating a plurality of question flow charts according to the question data;
Addressing the answering data according to the mapping relation to obtain an answering flow table corresponding to the questioning flow table;
generating a pre-training set and a verification set according to the questioning flow sheet and the answering flow sheet;
performing data augmentation processing on the pre-training set to obtain a training set;
performing iterative training on the pre-dialogue model according to the training set to obtain a dialogue training model;
and performing verification and evaluation on the dialogue training model according to the verification set, and taking the dialogue training model with the highest score as the dialogue model.
5. The intelligent customer service application method according to claim 4, wherein the performing data augmentation processing on the pre-training set to obtain a training set comprises:
extracting the character features of questions of each question flow chart according to the pre-training set;
carrying out semantic conversion on the character features of the problem according to the semantic converter so as to obtain first augmentation character data;
performing back translation processing on the first augmented character data to obtain second augmented character data;
word noise is added to the second augmented character data to obtain a training set.
6. The intelligent customer service application method according to claim 1, wherein the monitoring the dialogue model to collect user feedback data and updating the response knowledge base according to the user feedback data to obtain the response knowledge update base comprises:
Tracking performance indexes, log records and user feedback data of the dialogue model in real time;
acquiring a response flow, response time and response accuracy according to the performance index and the log record;
data analysis is carried out on the user feedback data and the response flow so as to obtain improved response data;
and storing the improved response data into the corresponding partition of the response knowledge base according to the response scene to form a response knowledge updating base.
7. The intelligent customer service application method according to claim 1, wherein the updating training of the dialogue model according to the response knowledge updating base to obtain an improved dialogue model comprises:
extracting customer service operation data and basic response data according to the response knowledge updating base;
clustering the client speech data and the basic response data to obtain one or more groups of clustered data;
determining one or more dialogue strategies according to the clustering data, and taking the dialogue strategies as an updated training set;
and updating and training the dialogue model through the updating training set so as to obtain an improved dialogue model.
8. An intelligent customer service application device, comprising:
the analysis module is used for acquiring the business data of the target enterprise and carrying out data analysis on the business data to obtain reply key analysis data;
The acquisition module is used for acquiring information of a common problem scheme and forming basic response data;
the database building module is used for building a response knowledge base according to the basic response data and the response key analysis data;
the modeling module is used for constructing a pre-dialogue model based on the NLP, and carrying out iterative training on the pre-dialogue model through the response knowledge base so as to obtain a dialogue model;
the monitoring module is used for monitoring the dialogue model to acquire user feedback data and updating the response knowledge base according to the user feedback data to obtain a response knowledge updating base;
and the optimization module is used for updating and training the dialogue model according to the response knowledge updating base so as to obtain an improved dialogue model.
9. An intelligent customer service application device, characterized in that the intelligent customer service application device comprises: a memory and at least one processor, the memory having instructions stored therein;
at least one of the processors invokes the instructions in the memory to cause the smart service application device to perform the steps of the smart service application method of any of claims 1-7.
10. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the steps of the intelligent customer service application method of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954778.5A CN117093685A (en) | 2023-07-31 | 2023-07-31 | Intelligent customer service application method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310954778.5A CN117093685A (en) | 2023-07-31 | 2023-07-31 | Intelligent customer service application method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117093685A true CN117093685A (en) | 2023-11-21 |
Family
ID=88769058
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310954778.5A Pending CN117093685A (en) | 2023-07-31 | 2023-07-31 | Intelligent customer service application method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117093685A (en) |
-
2023
- 2023-07-31 CN CN202310954778.5A patent/CN117093685A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109213910B (en) | A kind of government intelligence customer service method, system, electronic equipment and storage medium | |
| US10923115B2 (en) | Dynamically generated dialog | |
| KR102033388B1 (en) | Apparatus and method for question answering | |
| CN104679910A (en) | Intelligent answering method and system | |
| Hill et al. | Identifying shared understanding in design using document analysis | |
| US11238132B2 (en) | Method and system for using existing models in connection with new model development | |
| CN109299865A (en) | Psychological assessment system and method, information data processing terminal based on semantic analysis | |
| CN117271767A (en) | Operation and maintenance knowledge base establishing method based on multiple intelligent agents | |
| KR20200145299A (en) | Intelligent recruitment support platform based on online interview video analysis and social media information analysis | |
| CN113157867A (en) | Question answering method and device, electronic equipment and storage medium | |
| US11907863B2 (en) | Natural language enrichment using action explanations | |
| CN112287082A (en) | Data processing method, device, equipment and storage medium combining RPA and AI | |
| CN117312521A (en) | Processing method for intelligent customer service dialogue and related products | |
| CN115438158A (en) | Intelligent dialogue method, device, equipment and storage medium | |
| CN116860938A (en) | Voice question-answering construction method, device and medium based on large language model | |
| CN110929085A (en) | Power customer service message generation model sample processing system and method based on meta-semantic decomposition | |
| CN117093685A (en) | Intelligent customer service application method, device, equipment and storage medium | |
| EP3992876A1 (en) | Integration navigator and intelligent monitoring for living systems | |
| CN114969375A (en) | Method and system for giving artificial intelligence learning to machine based on psychological knowledge | |
| CN114860883A (en) | Intelligent question and answer method and system | |
| CN113468306A (en) | Voice conversation method, device, electronic equipment and storage medium | |
| CN111859148A (en) | Theme extraction method, device and equipment and computer readable storage medium | |
| CN111309990A (en) | Statement response method and device | |
| Prajapati et al. | Semantic network based cognitive, NLP powered question answering system for teaching electrical motor concepts | |
| CN118505175B (en) | AI interview system, AI interview method, AI interview equipment and AI interview storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |