CN116303982B - Intelligent response and service processing method and system based on RPA and self-learning mechanism - Google Patents
Intelligent response and service processing method and system based on RPA and self-learning mechanism Download PDFInfo
- Publication number
- CN116303982B CN116303982B CN202310591238.5A CN202310591238A CN116303982B CN 116303982 B CN116303982 B CN 116303982B CN 202310591238 A CN202310591238 A CN 202310591238A CN 116303982 B CN116303982 B CN 116303982B
- Authority
- CN
- China
- Prior art keywords
- rpa
- flow
- intention
- customer
- service
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000004044 response Effects 0.000 title claims abstract description 32
- 230000007246 mechanism Effects 0.000 title claims abstract description 23
- 238000003672 processing method Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 claims abstract description 105
- 230000008569 process Effects 0.000 claims abstract description 90
- 238000003058 natural language processing Methods 0.000 claims abstract description 35
- 230000006399 behavior Effects 0.000 claims abstract description 18
- 238000010801 machine learning Methods 0.000 claims abstract description 18
- 238000005516 engineering process Methods 0.000 claims abstract description 10
- 230000000694 effects Effects 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 230000001364 causal effect Effects 0.000 claims description 5
- 238000007726 management method Methods 0.000 claims description 4
- 230000009471 action Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 3
- 230000008451 emotion Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 230000003068 static effect Effects 0.000 claims description 2
- 238000012795 verification Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000004801 process automation Methods 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 206010033799 Paralysis Diseases 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the technical field of RPA intelligent customer service, and particularly relates to an intelligent response and service processing method and system based on an RPA and a self-learning mechanism. The method comprises the following steps: s1, analyzing a problem raised by a client through natural language processing NLP to obtain a client intention; s2, if the client intention analyzed by the NLP is matched with the flow in the RPA flow library, obtaining a result according to the RPA flow operation and feeding back the client; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed; s3, after the intervention of the manual service, the service responds and operates correspondingly according to the customer questions and generates a large amount of behavior track and business flow data; s4, through a machine learning technology, the RPA system identifies an automatizable business process from a large number of behavior tracks and business process data generated in the step S3, and meanwhile generates an RPA automation process corresponding to the intention and adds the RPA automation process into an RPA process library.
Description
Technical Field
The invention belongs to the technical field of RPA intelligent customer service, and particularly relates to an intelligent response and service processing method and system based on an RPA and a self-learning mechanism.
Background
The RPA (Robotic Process Automation, robot process automation) technology is a process automation technology, and can design and configure a business operation process capable of being automatically executed through a process editor provided by RPA software, package the business operation process into a form of a software robot or a virtual robot, deploy the form to a production environment and a business system for execution, simulate a series of operations of a person on a computer, and have three core functions of process discovery, process design and process execution. First, an automation flow is discovered, then automation steps are designed according to the flow, and finally the RPA system executes the automation flows.
Customer service is very important for electronic commerce, and the service quality directly influences the commodity yield, conversion rate and polling rate.
According to the 2022 e-commerce customer service experience report, it appears that the customer's service experience can significantly impact their brand preferences and purchasing decisions. But the quality of service and the efficiency of service are often difficult to be two-way. Improving the quality of service means that customer service is able to answer each customer's question accurately. In this way, the time per session will be longer, which also means that the merchant needs to pay higher labor costs. With the aim of keeping the market development of electronic commerce still hot, the consumer consultation demand is increasing, and the processing capacity and efficiency of manual customer service are challenged. At the same time, better quality of service means more specialized customer service is sufficient, and the personnel in the customer service industry flow greatly, and the training cost is increased. The adoption of intelligent customer service to automatically reply to customer consultation becomes the choice of many merchants. The interactive robot replaces manual customer service, so that the customer service cost is greatly reduced. However, intelligent customer service also has relatively large limitations. The traditional intelligent customer service system can answer questions of users, but cannot solve the problems of the users, especially the problems with complex service attributes. Many times, users are presented with questions not just to obtain a simple reply, but to be able to obtain the resource support of the enterprise to solve the self-presented questions. However, intelligent customer service that is not flexible enough often not only does not alleviate the consultation burden, but also often irritates the customer because their mechanical return cannot help the customer, reducing satisfaction.
At present, the existing RPA intelligent customer service mode and related technology have the following disadvantages:
1. need to manually configure RPA flow in advance
The existing RPA customer service system has the advantages that the flow is manually configured in advance, the flow template basis is absent when a new scene appears, and the existing RPA customer service system is manually configured only by configuration personnel. For example, the RPA automation procedure is compiled by natural language description business procedures as described in patent application No. CN 202011010218.7. For example, the intelligent customer service response described in CN202010797890.9 is manually preset, and in an actual service scenario, a problem that a manual intervention is required due to an unpreserved flow often occurs.
2. The manual preset flow is too complex or does not conform to the business
The manually configured RPA flow is sometimes considered to be too complex, the flow nodes are numerous, the response efficiency is affected, and sometimes the configured processing flow deviates to cause that the service requirement is not met. When the RPA flow is manually configured, configuration personnel often subjectively configure various pre-verification logics, most of core verification is already verified during actual service operation, which leads to the reduction of flow efficiency caused by repeated verification, and the verification standard during the manual configuration verification is maintained by the configuration personnel, so that errors occur in the flow which is supposed to be normally executed if deviation occurs.
3. Failure to deal with burst mass questions
When customer service asks a large number of new questions in a sudden manner, the RPA flow is not configured manually or how to configure the flow is not determined, the intelligent customer service robot is in a semi-paralyzed state at the moment, the manual customer service is subjected to huge pressure, and the customer experience is also reduced linearly.
Therefore, it is very important to design an intelligent response and business processing method and system based on RPA and self-learning mechanism, which can make intelligent customer service more flexible and efficient to process customer questions, avoid untimely feedback of customer service processing caused by sudden high-frequency similar question questions, and further achieve the effect of reducing labor cost.
Disclosure of Invention
The invention provides an intelligent response and service processing method and system based on RPA and a self-learning mechanism, which can enable intelligent customer service to process customer questions more flexibly and efficiently, avoid untimely customer service processing feedback caused by sudden high-frequency similar question questions, and further achieve the effect of reducing labor cost.
In order to achieve the aim of the invention, the invention adopts the following technical scheme:
the intelligent response and service processing method based on the RPA and the self-learning mechanism comprises the following steps of;
s1, analyzing a problem raised by a client through natural language processing NLP to obtain a client intention;
s2, if the client intention analyzed by the NLP is matched with the flow in the RPA flow library, obtaining a result according to the RPA flow operation and feeding back the client; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed;
s3, after the intervention of the manual service, the service responds and operates correspondingly according to the customer questions and generates a large amount of behavior track and business flow data;
s4, through a machine learning technology, the RPA system identifies an automatizable business process from a large number of behavior tracks and business process data generated in the step S3, and meanwhile generates an RPA automation process corresponding to the intention and adds the RPA automation process into an RPA process library.
Preferably, the step S1 includes the steps of:
s11, extracting basic features of a problem presented by a customer, wherein the basic features comprise word segmentation, semantic tag extraction, emotion analysis and named entity recognition;
s12, carrying out intention understanding on the questions presented by the clients, wherein the intention understanding comprises question field classification, intention recognition and attribute extraction;
s13, entering a dialogue management stage, including state tracking and dialogue decision-making; the state tracking specifically comprises the steps of determining the field and intention of a current customer for giving out a problem according to the context state, analyzing the intention of the customer, and matching corresponding response through the intention; the dialogue decision is specifically to decide the follow-up action according to the intention of the current customer to question.
Preferably, in step S2, the processes stored in the RPA process library include a process manually preconfigured and a process generated by RPA acquisition and machine learning; each flow in the RPA flow library corresponds to a different intent.
Preferably, each process in the RPA process library is used as a process template for providing reference and quick configuration when a new process needs manual configuration.
Preferably, step S3 includes the steps of:
s31, deploying an RPA platform to the terminal equipment, and grabbing behavior track and business flow data of customer service by an RPA system in a silent state;
s32, after the intervention of the manual customer service, the customer service responds and operates correspondingly according to the customer questioning;
s33, the RPA system collects operation process generation operation record data on the customer service computer.
Preferably, step S4 includes the steps of:
s41, constructing a flow chart:
considering that the input of the flow generation is an operation log, the operation log needs to be converted into a flow chart, and the flow chart is constructed by a Petri Net model method and used for representing the concurrent flow;
the Petri Net model is a bipartite graph composed of a stay position place and a transition; the Petri Net model is a static network diagram.
S42, flow generation:
adopting Alpha algorithm to realize flow generation and setting L 1 For an example operation log document, A, B, C, D, E represents event names, respectively, the upper right of the event names represents the frequency of each flow, and the slave operation log document L is represented by petri_net 1 A process of the prior art.
Preferably, the Alpha algorithm defines an order relationship of 4 activities based on the oplogging log document, specifically including:
directional connection, X > Y, X followed by Y;
causality, X- > Y, if and only if X is directly followed by Y, but Y is not directly followed by X;
parallel relationship, X Y, if and only if X is directly followed by Y, Y is also directly followed by X;
irrespective of the relationship, X#Y, if and only if X is not directly followed by Y, Y is also not directly followed by X;
wherein X and Y refer to activity in the log document;
based on the order relation of the 4 activities, 5 control flows are constructed, and the control flows are respectively as follows: sequential mode, parallel forking mode, parallel merging mode, select forking mode, select merging mode.
Preferably, the flow of the Alpha algorithm is specifically as follows:
s421, recording the log document L according to the operation 1 Constructing all active sets { A, B, C, D, E };
s422, determining a start position start (TI);
s423, determining a termination position end (TO);
s424, finding out all pairs (A, B), wherein each element belonging to the set A has a causal relationship with each element belonging to the set B, namely A- > B, and the elements in all sets A are independent of each other, and the elements in the set B are independent of each other; the pairs refer to event set pairs;
s425, deleting non-maximal bands to obtain final bandsThe method comprises the steps of carrying out a first treatment on the surface of the maximal pairs refer to the largest set, i.e., the (A, B) set retains the most inclusive events, and deletes less inclusive events, e.g., A can go through p1 to B, E, there is one ({ A }, { B, E }) pair, then ({ A }, { B }) and ({ A }, { R }) are non-maximal pairs, and delete;
s426, determining all the rest positions including p1, p2, p3 and p4;
s427, willThe pairs are connected to form a flow chart.
The invention also provides an intelligent response and service processing system based on the RPA and the self-learning mechanism, which comprises the following steps of;
the NLP intention analysis module is used for analyzing the problem raised by the client through natural language processing NLP to obtain the client intention;
the RPA flow library matching module is used for acquiring a result according to the RPA flow operation and feeding back the client if the client intention analyzed by the NLP is matched with the flow in the RPA flow library; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed;
the manual customer service operation module is used for carrying out corresponding response and operation according to the customer questioning after the manual customer service intervention, and generating a large amount of behavior track and business flow data;
the RPA process generation module is used for identifying an automatically-operable business process from a large amount of behavior tracks and business process data by the RPA system through a machine learning technology, generating an RPA automatic process corresponding to the intention, and adding the RPA automatic process into the RPA process library.
Compared with the prior art, the invention has the beneficial effects that: (1) According to the invention, the automatic generation of the RPA flow is achieved by combining the NLP analysis intention with the RPA acquisition and the machine learning, so that the manual pre-configuration of the RPA flow is avoided, the manual intervention operation in the case of burst massive questioning is faced, and the labor cost is effectively reduced; (2) As the generated flows accumulate, the flows generated by machine learning become more accurate through more and more data training generation, and provide references and templates for manual configuration.
Drawings
FIG. 1 is a schematic flow chart of an intelligent response and service processing method based on RPA and a self-learning mechanism in the invention;
FIG. 2 is a schematic diagram of a network architecture of the Petri Net model of the present invention;
fig. 3 is a flowchart of a practical application of an intelligent response and service processing method based on RPA and a self-learning mechanism according to an embodiment of the present invention.
Detailed Description
In order to more clearly illustrate the embodiments of the present invention, specific embodiments of the present invention will be described below with reference to the accompanying drawings. It is evident that the drawings in the following description are only examples of the invention, from which other drawings and other embodiments can be obtained by a person skilled in the art without inventive effort.
The intelligent response and service processing method based on the RPA and the self-learning mechanism shown in FIG. 1 comprises the following steps of;
1. analyzing the problem raised by the client through natural language processing NLP to obtain the client intention;
first, basic features of a problem presented by a customer need to be extracted, for example: word segmentation, semantic tag extraction, emotion analysis, NER (named entity, named entity recognition) recognition, etc.; then enter the next layer-intention understanding layer, mainly have problem domain classification, intention discern and attribute extraction; the intention understanding then proceeds to this stage of dialog management, where the dialog management module mainly consists of two parts, state tracking (DST, dialog State Tracking), dialog decision. State tracking (DST, dialog State Tracking) specifies the current user's domain and intent based on the context state, and matches the corresponding response by intent after analyzing the user's intent. The dialogue decision module decides the follow-up action according to the current intention of the user.
If the user asking intention is not clear, the user description can be further given according to the questions presented by the user, for example, the user presents "me complaints", the field is not clear, clear fields such as logistics, after-sales, customer service and the like can be presented to ask customers to confirm that the customer presents the logistics too slowly, and the intention can be confirmed that the logistics progress matches the corresponding operation and response through the RPA flow library. If the problem presented by the customer is received as 'the refrigerator is not refrigerated', the first choice is to judge which field the problem belongs to; when the obtained problem belongs to the field of merchants, the intention is judged, the intention is clarified, the intention is known to be 'how to apply for after-sales', and then an RPA flow corresponding to the refrigerator after-sales intention is searched in an RPA flow library.
2. If the NLP analyzed intention is matched with the flow in the RPA flow library, the result is obtained according to the RPA flow operation and fed back to the client, and if the NLP is not analyzed intention or is not matched with the flow in the RPA flow library after the client asking for the problem description, the client intervenes manually.
The processes stored in the RPA process library can be manually preconfigured, processes generated through RPA acquisition and machine learning can also exist, each process in the process library corresponds to different intentions, and meanwhile, the process library can also be used as a process template to provide reference and quick configuration when new processes need manual configuration.
3. The RPA platform is deployed to terminals such as a computer and a mobile phone, and the system captures the behavior track and service data of the user in a silent state. After the intervention of the manual customer service, the customer service responds and operates correspondingly according to the customer questioning, for example, the customer questioning 'where the my order is sent', the customer service queries the logistics progress operation in the business background according to the customer information, and the RPA collects the operation process on the customer service computer to generate operation record data. The staff only need to log in the business system according to the daily working mode of the staff and carry out business operation. Business operations by numerous employees produce a large amount of behavior trace and business process data, and the business processes that can be automated are hidden in the data.
4. Through a machine learning technology, the RPA system identifies an automatically-operable business process from a large amount of behavior tracks and business process data, generates an RPA automatic process corresponding to the intention, and adds the RPA automatic process into an RPA process library to realize automation of process discovery and process design. Of course, the business process automatically designed by the system may be imperfect, and the business personnel can modify the business process based on the business process. The difficulty of making a partial modification to an existing process must be significantly lower than designing the entire process from scratch. In addition, the RPA system can learn the process of the process modification by the service personnel, so as to improve a machine learning algorithm model, and the RPA automatic process provided later is more reasonable, and fewer places need to be modified. The generated flow enters the RPA flow library, and when a corresponding intention is provided by a customer, the corresponding flow is matched so as to trigger the RPA to automatically operate and feed back the customer, so that the intervention of manual customer service is avoided, and the effect of reducing the labor cost is achieved.
After intervention of the manual service, the RPA records the operation flow of the manual service. These operational flows produce a large amount of behavior trace and business flow data in which the automatable business flow is hidden. Therefore, it is necessary to mine out the executable business process from a large amount of business process data and add it to the RPA process library. The input of the flow generation is an operation log, the operation log needs to be converted into a flow chart, and the Petri Net model method provides a construction method of the flow chart and can represent the concurrent flow very compactly. The Petri Net model method can embody the relationships of sequence, selection, concurrency, circulation and the like among various operations in the business process. The Petri Net model is bipartite (bipartite graph) consisting of place and transition.
Adopting Alpha algorithm to realize flow generation and setting L 1 For an example operation log document, A, B, C, D, E represents event names, respectively, the upper right of the event names represents the frequency of each flow, and the slave operation log document L is represented by petri_net 1 A process of the prior art.
The flow of Alpha algorithm is specifically as follows:
recording log document L according to operation 1 Constructing all active sets { A, B, C, D, E };
determining a start position start (TI);
determining a termination location end (TO);
finding out all pairs (A, B), wherein each element belonging to the set A has a causal relationship with each element belonging to the set B, namely A- > B, and the elements in all the sets A are independent of each other, and the elements in the sets B are independent of each other; the pairs refer to event set pairs;
deleting non-maximal bands to obtain final bandsThe method comprises the steps of carrying out a first treatment on the surface of the maximal pairs refer to the largest set, i.e., the (A, B) set retains the most inclusive events, and deletes less inclusive events, e.g., A can go through p1 to B, E, there is one ({ A }, { B, E }) pair, then ({ A }, { B }) and ({ A }, { R }) are non-maximal pairs, and delete;
determining all the rest positions including p1, p2, p3, p4;
will beThe middle pair is connected to formA flow chart.
Therefore, the process of generating the flow model is specifically;
1.1 definition of the order relationship
The Alpha algorithm defines the order relation of 4 activities based on the operation record log document, and specifically comprises the following steps:
directional connection, X > Y, X followed by Y;
causality, X- > Y, if and only if X is directly followed by Y, but Y is not directly followed by X;
parallel relationship, X Y, if and only if X is directly followed by Y, Y is also directly followed by X;
irrespective of the relationship, X#Y, if and only if X is not directly followed by Y, Y is also not directly followed by X;
wherein X, Y refer to activity in the log;
based on the order relation of the 4 activities, 5 control flows are constructed, and the control flows are respectively as follows: sequential mode, parallel forking mode, parallel merging mode, select forking mode, select merging mode.
1.2 footprint matrix
For example log L 1 ={<A,B,C,D>,<A,C,B,D>,<E>In which the immediate relationship (also called direct following relationship) is A>B, B>C, C>D，A>C, C>B, B>D, E; example Log L 1 The corresponding footprint matrix is as follows:
table 1 example log L 1 Corresponding footprint matrix example table
1.3 Discovery process model
Discovering a flow model according to the footprint matrix relationship, as shown in fig. 2, wherein the rectangular box in fig. 2 represents events, including an event A, B, C, D, E; circles represent order relationships. According to table 1, a flow model as shown in fig. 2 is constructed to represent the order relationship between events A, B, C, D, E, with the specific order being: event E is irrelevant to event A, B, C, D; event a is causal to event B, C; the event B and the event C are in parallel relation; event B, C is causal to event D.
The invention also provides an intelligent response and service processing system based on the RPA and the self-learning mechanism, which comprises the following steps of;
the NLP intention analysis module is used for analyzing the problem raised by the client through natural language processing NLP to obtain the client intention;
the RPA flow library matching module is used for acquiring a result according to the RPA flow operation and feeding back the client if the client intention analyzed by the NLP is matched with the flow in the RPA flow library; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed;
the manual customer service operation module is used for carrying out corresponding response and operation according to the customer questioning after the manual customer service intervention, and generating a large amount of behavior track and business flow data;
the RPA process generation module is used for identifying an automatically-operable business process from a large amount of behavior tracks and business process data by the RPA system through a machine learning technology, generating an RPA automatic process corresponding to the intention, and adding the RPA automatic process into the RPA process library.
Fig. 3 is a flow chart showing a practical application of the intelligent response and service processing method based on RPA and self-learning mechanism. As can be seen from fig. 3, when the problem posed by the client a does not match with the corresponding RPA flow, the RPA system acquires operation data from the client computer, and provides the operation data to the machine learning system to generate the corresponding RPA operation of the client a question, and when other clients pose similar problems, the corresponding flow is generated in the RPA flow library, so as to achieve automatic RPA reply, and the specific implementation flow is as follows:
a, a customer gives out a product failure problem and provides information such as orders and the like;
NLP analyzes the intention of the customer to be after-sales due to product faults;
3. after the product is failed and sold, the corresponding RPA flow is not configured in advance so as to perform manual customer service intervention treatment;
the RPA acquisition program generates data through the operation of acquiring customer service and provides the data to a machine learning system;
5. collecting an operation log of the manual customer service by adopting a Petri Net method, converting the operation log into an operation flow chart, and then generating a corresponding RPA flow by adopting an alpha algorithm to realize the flow and storing the RPA flow into an RPA flow library; for example, the flow of fig. 2 may be a refund service of a user, where an event denoted by a is a front audit of refund of the user, an event denoted by B is a total refund operation, an event denoted by C is a partial refund operation, an event denoted by D is a refund operation, an event denoted by E is a refund user mark, and the refund service event A, B, C, D, E of the user is substituted into the process of generating the flow model, so that a corresponding RPA flow of refund service of the user may be generated;
6. the after-sale problem of the product failure of the subsequent customer is matched with the corresponding flow in the RPA flow library, so that the reply is automatically processed.
Through machine learning and RPA acquisition, the corresponding process can be generated for the intention of the unconfigured process, so that the processing capacity of intelligent customer service is improved, and the effect of reducing the labor cost is achieved.
The invention effectively combines NLP, RPA and machine learning, can make intelligent customer service process customer questions more flexibly and efficiently, and avoid the condition of untimely customer service processing feedback caused by sudden high-frequency similar question questions, thereby achieving the effect of reducing labor cost.
The innovation points of the invention are as follows:
1. the invention creatively binds the user intention with the RPA flow, and directly matches the user intention with the corresponding RPA operation;
2. according to the invention, the manual operation is finally converted into the RPA flow by the Petri Net model method and the Alpha algorithm, and the subsequent generated flow is more accurate along with the accumulation of service data;
3. the invention creatively adopts the NLP+RPA+machine learning mode, can ensure that an enterprise can generate an automatic RPA processing flow according to manual operation after manual intervention is performed on questions which are not pre-configured when customer questions are processed by customer service;
4. the data collected and generated by the mode of the invention not only takes the current generation flow as the basis, but also provides a data basis for the generation of each subsequent business flow, and the accumulation of data volume can enable the generated template to be more similar to the business.
The foregoing is only illustrative of the preferred embodiments and principles of the present invention, and changes in specific embodiments will occur to those skilled in the art upon consideration of the teachings provided herein, and such changes are intended to be included within the scope of the invention as defined by the claims.
Claims (8)
1. The intelligent response and service processing method based on the RPA and the self-learning mechanism is characterized by comprising the following steps of;
s1, analyzing a problem raised by a client through natural language processing NLP to obtain a client intention;
s2, if the client intention analyzed by the NLP is matched with the flow in the RPA flow library, obtaining a result according to the RPA flow operation and feeding back the client; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed;
s3, after the intervention of the manual service, the service responds and operates correspondingly according to the customer questions and generates a large amount of behavior track and business flow data;
s4, through a machine learning technology, the RPA system identifies an automatically-operable business process from a large amount of behavior tracks and business process data generated in the step S3, and simultaneously generates an RPA automatic process corresponding to the intention and adds the RPA automatic process into an RPA process library;
step S4 includes the steps of:
s41, constructing a flow chart:
considering that the input of the flow generation is an operation log, the operation log needs to be converted into a flow chart, and the flow chart is constructed by a Petri Net model method and used for representing the concurrent flow;
the Petri Net model is a bipartite graph composed of a stay position place and a transition; the Petri Net model is a static network diagram;
s42, flow generation:
adopting Alpha algorithm to realize flow generation and setting L 1 For an example operation log document, A, B, C, D, E represents event names, respectively, the upper right of the event names represents the frequency of each flow, and the slave operation log document L is represented by petri_net 1 A process of the prior art.
2. The intelligent response and service processing method based on RPA and self-learning mechanism according to claim 1, wherein step S1 comprises the steps of:
s11, extracting basic features of a problem presented by a customer, wherein the basic features comprise word segmentation, semantic tag extraction, emotion analysis and named entity recognition;
s12, carrying out intention understanding on the questions presented by the clients, wherein the intention understanding comprises question field classification, intention recognition and attribute extraction;
s13, entering a dialogue management stage, including state tracking and dialogue decision-making; the state tracking specifically comprises the steps of determining the field and intention of a current customer for giving out a problem according to the context state, analyzing the intention of the customer, and matching corresponding response through the intention; the dialogue decision is specifically to decide the follow-up action according to the intention of the current customer to question.
3. The intelligent response and service processing method based on RPA and self-learning mechanism according to claim 1, wherein in step S2, the procedures stored in the RPA procedure library include a procedure manually preconfigured and a procedure generated by RPA acquisition and machine learning; each flow in the RPA flow library corresponds to a different intent.
4. The intelligent response and service processing method based on RPA and self-learning mechanism according to claim 3, wherein each flow in the RPA flow library is used as a flow template for providing reference and quick configuration when a new flow needs manual configuration.
5. The intelligent response and service processing method based on RPA and self-learning mechanism according to claim 1, wherein step S3 comprises the steps of:
s31, deploying an RPA platform to the terminal equipment, and grabbing behavior track and business flow data of customer service by an RPA system in a silent state;
s32, after the intervention of the manual customer service, the customer service responds and operates correspondingly according to the customer questioning;
s33, the RPA system collects operation process generation operation record data on the customer service computer.
6. The intelligent response and service processing method based on RPA and self-learning mechanism according to claim 1, wherein the Alpha algorithm defines an order relation of 4 activities based on the operation log document, specifically comprising:
directional connection, X > Y, X followed by Y;
causality, X- > Y, if and only if X is directly followed by Y, but Y is not directly followed by X;
parallel relationship, X Y, if and only if X is directly followed by Y, Y is also directly followed by X;
irrespective of the relationship, X#Y, if and only if X is not directly followed by Y, Y is also not directly followed by X;
wherein X and Y refer to activity in the log document;
based on the order relation of the 4 activities, 5 control flows are constructed, and the control flows are respectively as follows: sequential mode, parallel forking mode, parallel merging mode, select forking mode, select merging mode.
7. The intelligent response and service processing method based on the RPA and the self-learning mechanism according to claim 6, wherein the Alpha algorithm flow is specifically as follows:
s421, recording the log document L according to the operation 1 Constructing all active sets { A, B, C, D, E };
s422, determining a start position start (TI);
s423, determining a termination position end (TO);
s424, finding out all pairs (A, B), wherein each element belonging to the set A has a causal relationship with each element belonging to the set B, namely A- > B, and the elements in all sets A are independent of each other, and the elements in the set B are independent of each other; the pairs refer to event set pairs;
s425, deleting non-maximal bands to obtain final bandsmaximal pairs refer to the largest set, i.e., the (A, B) set retains the most fully contained events therein, and deletes the events that contain insufficiency;
s426, determining all the rest positions including p1, p2, p3 and p4;
s427, willThe pairs are connected to form a flow chart.
8. An intelligent response and service processing system based on an RPA and a self-learning mechanism, which is used for realizing the intelligent response and service processing method based on the RPA and the self-learning mechanism according to any one of claims 1-7, and is characterized in that the intelligent response and service processing system based on the RPA and the self-learning mechanism comprises;
the NLP intention analysis module is used for analyzing the problem raised by the client through natural language processing NLP to obtain the client intention;
the RPA flow library matching module is used for acquiring a result according to the RPA flow operation and feeding back the client if the client intention analyzed by the NLP is matched with the flow in the RPA flow library; if the NLP does not analyze the customer intention or does not match the flow in the RPA flow library, manual customer service intervention is performed;
the manual customer service operation module is used for carrying out corresponding response and operation according to the customer questioning after the manual customer service intervention, and generating a large amount of behavior track and business flow data;
the RPA process generation module is used for identifying an automatically-operable business process from a large amount of behavior tracks and business process data by the RPA system through a machine learning technology, generating an RPA automatic process corresponding to the intention, and adding the RPA automatic process into the RPA process library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310591238.5A CN116303982B (en) | 2023-05-24 | 2023-05-24 | Intelligent response and service processing method and system based on RPA and self-learning mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310591238.5A CN116303982B (en) | 2023-05-24 | 2023-05-24 | Intelligent response and service processing method and system based on RPA and self-learning mechanism |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116303982A CN116303982A (en) | 2023-06-23 |
| CN116303982B true CN116303982B (en) | 2023-08-22 |
Family
ID=86801814
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310591238.5A Active CN116303982B (en) | 2023-05-24 | 2023-05-24 | Intelligent response and service processing method and system based on RPA and self-learning mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116303982B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116882724B (en) * | 2023-07-13 | 2024-06-11 | 北京优特捷信息技术有限公司 | Method, device, equipment and medium for generating business process optimization scheme |
| CN116860952B (en) * | 2023-09-04 | 2023-11-03 | 富璟科技（深圳）有限公司 | RPA intelligent response processing method and system based on artificial intelligence |
| CN117311798A (en) * | 2023-11-28 | 2023-12-29 | 杭州实在智能科技有限公司 | RPA flow generation system and method based on large language model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919671A (en) * | 2019-02-27 | 2019-06-21 | 上海宝尊电子商务有限公司 | The high efficiency smart client service method and system of a kind of Electronic Commerce field customer orientation |
| WO2019153522A1 (en) * | 2018-02-09 | 2019-08-15 | 卫盈联信息技术（深圳）有限公司 | Intelligent interaction method, electronic device, and storage medium |
| CN111026843A (en) * | 2019-12-02 | 2020-04-17 | 北京智乐瑟维科技有限公司 | Artificial intelligent voice outbound method, system and storage medium |
| CN113704432A (en) * | 2021-08-31 | 2021-11-26 | 广州方舟信息科技有限公司 | Artificial intelligence customer service system construction method and device based on Internet hospital |
| WO2022095378A1 (en) * | 2020-11-03 | 2022-05-12 | 平安科技（深圳）有限公司 | Artificial-intelligence-based training method and apparatus, and computer device and storage medium |
| CN115599889A (en) * | 2022-10-31 | 2023-01-13 | 北京力码科技有限公司（Cn） | Online customer service intelligent answering method and system applied to financial insurance platform |
-
2023
- 2023-05-24 CN CN202310591238.5A patent/CN116303982B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019153522A1 (en) * | 2018-02-09 | 2019-08-15 | 卫盈联信息技术（深圳）有限公司 | Intelligent interaction method, electronic device, and storage medium |
| CN109919671A (en) * | 2019-02-27 | 2019-06-21 | 上海宝尊电子商务有限公司 | The high efficiency smart client service method and system of a kind of Electronic Commerce field customer orientation |
| CN111026843A (en) * | 2019-12-02 | 2020-04-17 | 北京智乐瑟维科技有限公司 | Artificial intelligent voice outbound method, system and storage medium |
| WO2022095378A1 (en) * | 2020-11-03 | 2022-05-12 | 平安科技（深圳）有限公司 | Artificial-intelligence-based training method and apparatus, and computer device and storage medium |
| CN113704432A (en) * | 2021-08-31 | 2021-11-26 | 广州方舟信息科技有限公司 | Artificial intelligence customer service system construction method and device based on Internet hospital |
| CN115599889A (en) * | 2022-10-31 | 2023-01-13 | 北京力码科技有限公司（Cn） | Online customer service intelligent answering method and system applied to financial insurance platform |
Non-Patent Citations (1)
| Title |
|---|
| 王强 ; 王筱涵 ; 刘刚 ; 耿慧欣 ; .飞机***安全性指标的PetriNet分配方法.国防科技大学学报.2015,(第06期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116303982A (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116303982B (en) | Intelligent response and service processing method and system based on RPA and self-learning mechanism | |
| CN107291867B (en) | Dialog processing method, device and equipment based on artificial intelligence and computer readable storage medium | |
| CN112035325B (en) | Text robot automatic monitoring method and device | |
| CN106911963A (en) | Method for information display and device in a kind of terminal device | |
| CN111488436B (en) | Multi-round dialogue model generation method, testing device and electronic equipment | |
| CN111679808B (en) | RPA robot application demand evaluation method and device | |
| CN107370823A (en) | Data acquisition and long-range control method, device and computer-readable recording medium | |
| CN111782635A (en) | Data processing method and apparatus, storage medium, and electronic apparatus | |
| CN112397061A (en) | Online interaction method, device, equipment and storage medium | |
| CN109118151B (en) | Work order transaction processing method and work order transaction processing system | |
| CN117311798A (en) | RPA flow generation system and method based on large language model | |
| CN106682910A (en) | Information processing method, system and related equipment | |
| CN116860938A (en) | Voice question-answering construction method, device and medium based on large language model | |
| Bettacchi et al. | Understanding production chain business process using process mining: a case study in the manufacturing scenario | |
| CN104123317A (en) | Service organization assessing and analyzing method based on knowledge base | |
| CN207099116U (en) | Data acquisition and remote control | |
| CN110062112A (en) | Data processing method, device, equipment and computer readable storage medium | |
| CN115794545A (en) | Automatic processing method of operation and maintenance data and related equipment thereof | |
| CN108491506A (en) | Method for pushing problem answers combination | |
| CN104123585A (en) | Service organization optimization analysis method based on service simulation | |
| KR101700838B1 (en) | Method and system for managing business project step by step | |
| do Amaral et al. | Attribute selection with filter and wrapper: an application on incident management process | |
| Baida et al. | E-business cases assessment: From business value to system feasibility | |
| CN111783391A (en) | Online artificial text marking system and method | |
| CN109471931A (en) | Corpus collection method, device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |