CN117221451A - Customer service response system and method based on artificial intelligence - Google Patents
Customer service response system and method based on artificial intelligence Download PDFInfo
- Publication number
- CN117221451A CN117221451A CN202311262679.7A CN202311262679A CN117221451A CN 117221451 A CN117221451 A CN 117221451A CN 202311262679 A CN202311262679 A CN 202311262679A CN 117221451 A CN117221451 A CN 117221451A
- Authority
- CN
- China
- Prior art keywords
- response
- client
- customer service
- customer
- module
- 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.)
- Granted
Links
- 230000004044 response Effects 0.000 title claims abstract description 225
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 38
- 230000003993 interaction Effects 0.000 claims abstract description 63
- 239000013598 vector Substances 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 12
- 230000002452 interceptive effect Effects 0.000 claims description 11
- 238000012790 confirmation Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 3
- 230000006399 behavior Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000001105 regulatory effect Effects 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000019771 cognition Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Abstract
The application relates to the technical field of Internet, and particularly discloses a customer service response system and method based on artificial intelligence, wherein the system comprises the following steps: the client management center is used for carrying out hierarchical division on all clients, respectively generating corresponding response templates for the clients of each hierarchy, and confirming the hierarchy to which the client belongs according to the historical interaction data of each client; the customer service response center is used for extracting response templates of corresponding levels according to the levels to which the customer belongs to respond, recording response data of the customer, and synchronizing the response data to the customer management center. According to the application, the clients are divided into the layers, the corresponding response templates are respectively set for each layer, the layer to which the clients belong is determined by combining the historical interaction data of the clients, and the corresponding response templates are used for responding, so that the method and the device can better meet the actual demands of the users and improve the customer service experience of the users.
Description
Technical Field
The application relates to the technical field of Internet, in particular to a customer service response system and method based on artificial intelligence.
Background
Along with the popularization of intelligent information processing and artificial intelligence, more and more fields begin to select to utilize artificial intelligence customer service to replace traditional artificial customer service, and the artificial intelligence customer service mode not only can save labor cost, but also can realize standardized response to customer demands, and is also a future development trend of customer service work.
Although the response given by the artificial intelligence customer service can effectively solve the problem of the customer in many times, and gives the customer a feeling of real person in conversation, the customer has a plurality of defects along with deep use of the customer, such as using the same response template in different use stages of the same customer, but the cognition and the demand of the customer on the product can be changed at this time, and if the same response template is still adopted, the customer is easy to bring bad service experience to the customer.
Disclosure of Invention
The application aims to provide an artificial intelligence-based customer service response system and method, which are characterized in that clients are divided into layers, a response template is set for each layer, and the user's historical interaction data is combined to select a proper response template to participate in response, so that the system and method better meet the actual demands of the users.
In a first aspect, the present application provides an artificial intelligence-based customer service response system, the system comprising a customer management center and a customer service response center;
the client management center is used for carrying out hierarchical division on all clients, respectively generating corresponding response templates for the clients of each hierarchy, and confirming the hierarchy to which the client belongs according to the historical interaction data of each client;
the customer service response center is used for extracting response templates of corresponding levels according to the levels to which the customer belongs to respond, recording response data of the customer, and synchronizing the response data to the customer management center.
Through the technical scheme, the clients are divided into the layers, corresponding response templates are set for each layer respectively, the layer to which the clients belong is determined by combining the historical interaction data of the clients, and the corresponding response templates are used for responding, so that the actual demands of the users can be met better, and the customer service experience of the users can be improved.
Optionally, the client management center comprises a hierarchy dividing module, a response template generating module and a client hierarchy confirming module;
the hierarchy dividing module is used for confirming a client dividing hierarchy;
the response template generation module is used for generating a corresponding response template for each level respectively;
the client hierarchy confirming module is used for carrying out matching through a preset knowledge graph model according to historical interaction data of clients and confirming the hierarchy to which the clients belong according to a matching result.
Optionally, the client level confirmation module includes a vector extraction unit, a semantic generation unit, a semantic matching unit, and a level determination unit;
the vector extraction unit is used for extracting keywords according to the historical interaction data and generating word vectors;
the semantic generation unit is used for generating a semantic network of the client through a preset knowledge graph model according to the word vector;
the semantic matching unit is used for carrying out similarity matching on the generated semantic network through a preset knowledge graph template and obtaining a semantic matching result;
the hierarchy judging unit is used for judging the hierarchy to which the client belongs through a preset threshold according to the semantic matching result.
Optionally, the client management center further includes an interaction data recording module and a client level updating module;
the interactive data recording module is used for recording behavior data of clients and receiving response data fed back by the customer service response center;
the client level updating module is used for updating the level to which the client belongs according to the newly added interaction data record.
Optionally, the customer service response center comprises an information acquisition module, a response template adjustment module and a customer service response module;
the information acquisition module is used for acquiring information of a client, wherein the information of the client comprises historical interaction data of the client and a hierarchy to which the client belongs;
the response template adjusting module is used for extracting a corresponding response template according to the hierarchy to which the client belongs and adjusting the response template according to the historical interaction data;
the customer service response module is used for responding according to the adjusted response template and synchronizing response data to the customer management center.
Optionally, the response template adjusting module comprises a style feature extracting unit and a style merging unit;
the style characteristic extraction unit is used for acquiring style characteristic information of the client through a preset method according to historical interaction data of the client;
the style fusion unit is used for fusing the client style characteristic information with the response template through a preset method to generate an adjusted response template.
Optionally, the customer service response center further comprises a response transfer module;
the response transfer module is used for outputting prompt information when the information submitted by the client contains preset keywords in the response process, and is used for prompting the response of the manual customer service catcher.
In a second aspect, the present application provides an artificial intelligence based customer service response method, including the steps of:
acquiring information of a current client, and judging whether the current client is a new client or not through preset database matching;
if yes, responding through a preset initial response template;
if not, confirming the hierarchy to which the client belongs through a preset database according to the client information, and acquiring a corresponding hierarchy response template;
extracting historical interaction data of the client through a preset database, and adjusting the acquired hierarchical response templates according to the historical interaction data to acquire adjusted response templates;
and responding through the adjusted response template, and simultaneously recording response data generated in the whole response process.
Optionally, the adjusting the obtained hierarchical response template according to the historical interaction data to obtain an adjusted response template includes:
according to the historical interaction data, through a preset method or style characteristic information of a client;
based on style characteristic information, fusing the style characteristic information with the current response template through a preset method to generate an adjusted response template.
In a third aspect, the present application provides a computer readable storage medium storing a computer program capable of being loaded by a processor and executing an artificial intelligence based customer service answering method as described above.
In summary, the familiarity with the client is firstly simulated according to the historical interaction data of the client, so that the client is hierarchically divided, and different response templates are respectively set for different hierarchies, so that the actual requirements of the user are met better, and the customer service experience of the client is improved; in addition, preference or style characteristics of the client are extracted according to historical interaction data of the client, so that a response template of the current and the client is adjusted, namely, the degree of understanding of the client is simulated, and the real response experience is better fitted; in addition, after each response is completed, the hierarchy to which the client belongs is updated according to the recorded response data, so that real-time tracking management is performed on the client, and the client is kept in an optimal state to face each client needing the response.
Drawings
FIG. 1 is a schematic diagram of an artificial intelligence based customer service answering system provided by an embodiment of the present application;
FIG. 2 is a schematic diagram of a customer management center provided by an embodiment of the present application;
FIG. 3 is a schematic diagram of a client-level validation module provided by an embodiment of the present application;
FIG. 4 is a schematic diagram of a customer service response center according to an embodiment of the present application;
fig. 5 is a flowchart of a customer service response method based on artificial intelligence according to an embodiment of the present application.
Detailed Description
The present application will be described in further detail with reference to fig. 1 to 5.
The application provides a customer service response system based on artificial intelligence, and referring to fig. 1, the system comprises a customer management center 10 and a customer service response center 20.
The client management center 10 is configured to perform hierarchical division on all clients, generate corresponding response templates for clients of each hierarchical level, and confirm the hierarchical level to which the client belongs according to historical interaction data of each client.
The customer service response center 20 is configured to extract a response template of a corresponding hierarchy according to the hierarchy to which the customer belongs to perform a response, record response data of the customer, and synchronize the response data to the customer management center 10.
The historical interaction data of the client comprise behavior data and communication data, wherein the behavior data is data such as browsing, purchasing and commenting of the client, and the communication data is communication data for responding between the client and customer service.
Because the existing artificial intelligence customer service generally uses the same response template for the same customer in different use stages when responding, but the cognition and the demand of the customer on the product may be changed at this time, if the same response template is still adopted, bad service experience is easily brought to the user.
For example, the user sends a page display of a product, and the artificial intelligence customer service responds to the simple introduction of the product according to the set template, if the customer is the first purchase, the problem does not exist, but if the customer is not the first purchase, it is obvious that the customer has a certain degree of knowledge about the product, and if the artificial intelligence customer service responds according to the previous template, the artificial intelligence customer service is obviously contrary to the actual requirement of the customer, so that a bad customer service experience is brought to the customer.
Therefore, for users in different stages, different response modes should be given, which is just like shopping in online physical stores, so that for the first visiting clients, the boss will have a corresponding conversation, and for acquaintances, the boss will have a conversation, this flexible mode is easy to pull the distance from the client, especially for acquaintances, and for online platforms, the client has a better experience, and it is very difficult for the client to have this experience, and the client may shop for many times in a certain online store, but artificial intelligence customer service will not usually treat you as an acquaintance, and if the client is offline, the boss will often get a cold greeting or even a family, and the two experiences are very different.
Therefore, the application aims to simulate the familiarity degree with the client according to the historical interaction data of the client, so as to divide the client into layers, and then respectively setting corresponding response templates for different layers.
In the embodiment of the present application, all clients are hierarchically divided by the client management center 10, corresponding response templates are generated for clients of each hierarchy, and the hierarchy to which the client belongs is confirmed according to the historical interaction data of each client.
Specifically, referring to fig. 2, the client management center 10 includes a hierarchy dividing module 11, a response template generating module 12, and a client hierarchy confirming module 13.
Wherein the hierarchy dividing module 11 is used for confirming the client division hierarchy.
The response template generation module 12 is configured to generate a corresponding response template for each hierarchy.
The client level confirmation module 13 is configured to perform matching according to historical interaction data of the client through a preset knowledge graph model, and confirm the level to which the client belongs according to the matching result.
Because the clients can be simply divided into two types, one type is that no historical interaction data exists, namely, the first visiting client, and for the first visiting client, an initial response template is used for responding; while the other class has historical interaction data, the division hierarchy is mainly aimed at the clients of the class.
First, the client division level is confirmed by the level division template 11, for example, the client division level can be simply divided into three levels, namely a first level, a second level and a third level, and abstract definition can be performed on each level, for example, the first level represents that the client has access or response records, and the first level can be defined as that the client has purchase intention; the second level indicates that the customer has purchased records; the third level indicates that the client has multiple purchase records, which is of course only an abstract representation and is not a standard of division, and the division of the level can be finer, and the application is not limited in particular.
After confirming the division levels of the clients, the response template generating module 12 can generate corresponding response templates, namely, a level response template, for each level, if the division modes of three levels are adopted, the generated level response templates can be respectively marked as a first response template, a second response template and a third response template, and naturally, an initial response template is also provided, and the actual level response template is also generated on the basis of the initial response template.
After confirming the response templates, when a client opens a session and needs to answer, the corresponding response templates can be extracted according to the hierarchy to which the client belongs, so that the hierarchy of the client needs to be confirmed, namely, the client is matched through a preset knowledge graph model according to the historical interaction data of the client through the client hierarchy confirming module 13, and the hierarchy to which the client belongs is confirmed according to the matching result.
Because the above description refers to simulating the familiarity with the customer according to the historical interaction data of the customer, and the familiarity needs to be defined by giving a quantifiable standard, a knowledge-graph-structured commodity network system can be constructed for each online store by means of the knowledge graph, each type of commodity is equivalent to an entity, each type of commodity has a correlation with each other, each commodity has a variety of attributes or labels, and each attribute or label has a variety of correlations with each other, so that a complete commodity network structure can be formed.
If the commodity network architecture of the store is used as a template, historical interaction data of the client is extracted through a knowledge graph and mapped onto the template, the number of the included entities and the number of the included relationship paths are included, quantization can be carried out, hierarchy division can be carried out by setting corresponding coverage thresholds, and the higher the coverage, the higher the familiarity of the user is, and the lower the familiarity is otherwise.
Specifically, referring to fig. 3, the client-level confirmation module 13 includes a vector extraction unit 131, a semantic generation unit 132, a semantic matching unit 133, and a level determination unit 134.
The vector extraction unit 131 is configured to extract keywords according to the historical interaction data, and generate word vectors.
The semantic generation unit 132 is configured to generate a semantic network of the client through a preset knowledge-graph model according to the word vector.
The semantic matching unit 133 is configured to perform similarity matching on the generated semantic network through a preset knowledge-graph template, and obtain a semantic matching result.
The hierarchy determining unit 134 is configured to determine, according to the semantic matching result, a hierarchy to which the client belongs through a preset threshold.
First, the keyword is extracted according to the historical interaction data by the vector extraction unit 131, and a word vector is generated, and since the above-mentioned historical interaction data of the client includes recording data such as browsing and purchasing of the user in addition to session data participating in the response, the keyword is extracted herein, and the keyword is extracted in combination with behavior data of the client, that is, browsing and purchasing data, except that different weights are correspondingly given, for example, more weights are given to the purchased information.
The semantic generation unit 132 can then generate the semantic network of the customer through a preset knowledge graph model according to the word vector, the semantic network actually refers to a target knowledge graph structure, which can be understood as a knowledge mapping, the information related to the customer is mapped into the network structure system of the store in the form of a knowledge graph, and the network structure system of the store refers to a preset knowledge graph template, so that the semantic matching unit 133 can perform similarity matching on the generated semantic network through the preset knowledge graph template, and obtain a semantic matching result, wherein the similarity matching result refers to the coverage degree of the knowledge surface mentioned in the foregoing, and the semantic matching result corresponds to a total score by defining a score for each entity, entity attribute and various relation paths in advance.
Finally, the hierarchy determining unit 134 may determine the hierarchy to which the client belongs according to the semantic matching result through a preset threshold, where the preset threshold is a threshold range corresponding to each hierarchy, and determine the hierarchy to which the client belongs by determining which threshold range the score corresponding to the semantic matching result falls.
On one hand, in the process of responding to the client, the communication data with the client, namely the response data, is recorded; on the other hand, the client may also synchronize and update the records of browsing access and the like of the client in real time, which is equivalent to that of the client, and the historical interaction data is newly added, and because one more interaction data, one more familiarity is possible, the current hierarchy of the client may need to be updated and adjusted.
Thus, in an embodiment of the present application, referring to FIG. 2, customer management center 10 further includes an interaction data recording module 14 and a customer level update module 15.
The interactive data recording module 14 is configured to record behavior data of a client and receive response data fed back by the customer service response center.
The client-level updating module 15 is configured to update a level to which a client belongs according to the newly added interaction data record.
Firstly, the interactive data recording module 14 is used for recording the behavior data of the client in real time and receiving the response data fed back by the customer service response center 20, namely, the newly added interactive data of the client, then the client level updating module 15 is used for updating the level to which the client belongs according to the newly added interactive data, and the client level confirmation module 13 is used for carrying out level confirmation again according to the current level of the client, namely, the latest historical interactive data, and then the level to which the client belongs is updated or reserved according to the level confirmation structure.
After confirming the belonging hierarchy for all clients, when the user opens the session and needs to answer, the belonging hierarchy of the clients can be confirmed first, and then the corresponding answer templates are extracted for answer.
Therefore, in the embodiment of the present application, the customer service response center 20 extracts the response templates of the corresponding levels according to the levels to which the customer belongs to perform the response, records the response data of the customer, and synchronizes the response data to the customer management center 10.
Specifically, referring to fig. 4, the customer service response center 20 includes an information acquisition module 21, a response template adjustment module 22, and a customer service response module 23.
Wherein the information acquisition module 21 is used for acquiring information of clients.
The response template adjusting module 22 is configured to extract a corresponding response template according to the hierarchy to which the client belongs, and adjust the response template according to the historical interaction data.
The customer service response module 23 is configured to respond according to the adjusted response template, and synchronize response data to the customer management center 10.
The client information comprises the historical interaction data of the client and the hierarchy to which the client belongs, and for the client information, a pre-established database is provided for storing the client data, and the user ID is used as a key field to extract the historical interaction data of the client and the current hierarchy to which the client belongs.
Firstly, after the session is started, the information of the client is acquired according to the information acquisition module 21, namely, the client is matched from the database according to the ID of the client, the hierarchy to which the client belongs is confirmed, if the user information fails to be matched with the information in the database, the client is the first visiting client, and then the client directly responds by adopting an initial response template; if the information can be matched, the historical interaction data of the client and the current hierarchical information of the client are directly extracted.
After confirming the current hierarchy of the client, the corresponding response template can be extracted, but each user has own preference or style, so even if the client is at one hierarchy, corresponding adjustment should be performed during response, namely, after the corresponding response template is extracted by the response template adjusting module 22, the response template is adjusted according to the client history interaction data, namely, the preference of the client is fused in the response process, and the customer service experience of the client can be better improved.
The above-mentioned description is that the response template simulates the familiarity degree of the client to the user, and the adjustment of the response template by the historical interaction data of the client simulates the knowledge degree of the user to the client, so that only two parties are familiar with each other, the response session at that time can better fit the real interaction state, thereby giving the client a better customer service experience.
Specifically, the answer template adjustment module 22 includes a style feature extraction unit 221 and a style fusion unit 222.
The style characteristic extraction unit 221 is configured to obtain style characteristic information of the client according to the historical interaction data of the client through a preset method.
The style fusion unit 222 is configured to fuse the client style feature information with the response template by a preset method, so as to generate an adjusted response template.
Firstly, style characteristic information of the client is obtained through a preset method according to historical interaction data of the client through a style characteristic extraction unit 221, wherein the style characteristic information is equivalent to the conventional method for recommending the client according to the historical browsing data of the client, and various notes, namely style characteristic information of the client, are attached to the client mainly through portrait of the client.
The customer style characteristic information is then fused with the response templates by the style fusion unit 222 using a preset method to generate an adjusted response template, since the response is actually a response and answer based on the questions posed by the customer, whereas the response template is actually trained by a large number of questions posed by the customer, and then a set of optimal response strategies given for each question is trained by a considerable algorithm model, for example, using RLHF (reinforced learning model based on human feedback), although the optimality here refers to the optimal outcome of the model training process, rather than to the fact that the final response template is optimal.
Therefore, style fusion is performed, that is, an algorithm model is actually adjusted, for example, training is performed according to the reinforcement learning model, when the reward weight is performed, the style characteristics of the client are referred to, that is, when the reward model is generated through training, the style characteristics of the client are fused into the whole reward model as a reward signal, so that the finally generated response model is inclined towards the preference or style of the client, and further better customer service experience is brought to the client.
After the current response template of the client is adjusted by the response template adjusting module 22, the client can start to participate in the response session, that is, the client can respond according to the adjusted response template by the customer service response module 23, and the response data is synchronized to the client management center 10, so that the interaction data of the client can be conveniently recorded and updated, and the current hierarchy of the client can be conveniently updated and adjusted.
In addition, during the process of using the artificial intelligence customer service, problems that are too subjective or that the artificial intelligence customer service does not answer well may be involved, such as price negotiations, or that the customer has a dissatisfaction, and thus, the artificial intelligence customer service is difficult to deal with, so that the artificial intelligence customer service is involved in this time.
Thus, in an embodiment of the present application, referring to FIG. 4, customer service response center 20 also includes response transfer module 24.
The response transferring module 24 is configured to output a prompt message for prompting the manual service to answer when the information submitted by the client includes a preset keyword in the response process.
Because the problem of not well processing artificial intelligence customer service is not well defined, only a practical scene can be combined, and some keywords, or keyword groups and the like are set, when a question of a customer or a response of the customer involves a preset keyword in the response process, corresponding prompt information is output through the response transfer module 24, and the prompt information prompts related artificial customer service to intervene, so that adverse effects caused by too relying on the artificial intelligence customer service are avoided to a certain extent.
The embodiment of the application also provides a customer service response method based on artificial intelligence, which is shown in fig. 5 and comprises the following steps:
s100, acquiring information of the current client, and judging whether the current client is a new client or not through preset database matching.
And S200, if the current client is a new client, responding through a preset initial response template.
And S300, if the current client is not a new client, confirming the hierarchy to which the client belongs through a preset database according to the client information, and acquiring a corresponding hierarchy response template.
S400, extracting historical interaction data of the client through a preset database, and adjusting the acquired hierarchical response templates according to the historical interaction data to acquire adjusted response templates.
S500, responding through the adjusted response template, and simultaneously recording response data generated in the whole response process.
Wherein the preset database comprises user ID, hierarchy corresponding to the user and clients
Historical interaction data.
When customer opens session and needs customer service response, user information is obtained, the information mainly represents ID information of the user, then if the current customer is a new customer, namely the customer is first visiting, at this time, the information of the customer is not in the preset database, so that response can be carried out through the preset initial response template.
If the current client is not a new user, that is, the client has remained interactive data, at this time, according to the ID information of the user, the historical interactive data of the user can be extracted through a preset database, the level to which the client belongs can be known, and then a corresponding level response template can be extracted for response.
Because each user has own preference or style, even if the user is in a hierarchy, corresponding adjustment should be performed when answering, so that in popular terms, people can speak, the preference of each user and the historical answer record of the user can be deduced, the personality of the user and the like can be obtained, and the styles can be integrated into the answer template, so that customer service experience of the user can be better improved.
Therefore, after the corresponding hierarchical response template is extracted according to the current client's hierarchical level, historical interaction data of the client is also extracted through a preset database, and the acquired hierarchical response template is adjusted according to the historical interaction data, so that the adjusted response template is acquired.
Specifically, the acquired hierarchical response templates are adjusted according to the historical interaction data to acquire adjusted response templates, and the method comprises the following steps:
s410, according to the historical interaction data, style characteristic information of the client is obtained through a preset method.
S420, fusing the current response template through a preset method based on style characteristic information to generate an adjusted response template.
The style characteristic information is equivalent to the conventional method for recommending the client according to the historical browsing data of the client, and is mainly formed by portrait of the client, and various notes, namely the style characteristic information of the client, are attached to the client.
After the style characteristic information of the client is obtained, the style characteristic information can be fused with the current hierarchical response template, namely, the style characteristic information of the client is added into an algorithm model for generating the response template, and the finally generated response template is inclined towards the preference or style of the client, so that the client is more fit with the real customer service experience.
Finally, after the answer template regulated by the style characteristics of the client is obtained, the regulated answer template can be used for answering, and answer data generated in the whole answer process are recorded at the same time, so that the interactive data of the client can be conveniently updated, and the current hierarchy of the client can be updated and regulated.
The embodiment of the application also provides a computer readable storage medium which stores a computer program capable of being loaded by a processor and executing any customer service response method based on artificial intelligence.
The embodiments of the present application are all preferred embodiments of the present application, and are not intended to limit the scope of the present application in this way, therefore: all equivalent changes according to the principles of the present application should be covered by the scope of the present application.
Claims (10)
1. A customer service response system based on artificial intelligence, which is characterized by comprising a customer management center and a customer service response center;
the client management center is used for carrying out hierarchical division on all clients, respectively generating corresponding response templates for the clients of each hierarchy, and confirming the hierarchy to which the client belongs according to the historical interaction data of each client;
the customer service response center is used for extracting response templates of corresponding levels according to the levels to which the customer belongs to respond, recording response data of the customer, and synchronizing the response data to the customer management center.
2. The customer service response system based on artificial intelligence according to claim 1, wherein the customer management center comprises a hierarchy dividing module, a response template generating module and a customer hierarchy confirming module;
the hierarchy dividing module is used for confirming a client dividing hierarchy;
the response template generation module is used for generating a corresponding response template for each level respectively;
the client hierarchy confirming module is used for carrying out matching through a preset knowledge graph model according to historical interaction data of clients and confirming the hierarchy to which the clients belong according to a matching result.
3. The customer service response system based on artificial intelligence according to claim 2, wherein the customer-level confirmation module comprises a vector extraction unit, a semantic generation unit, a semantic matching unit and a level determination unit;
the vector extraction unit is used for extracting keywords according to the historical interaction data and generating word vectors;
the semantic generation unit is used for generating a semantic network of the client through a preset knowledge graph model according to the word vector;
the semantic matching unit is used for carrying out similarity matching on the generated semantic network through a preset knowledge graph template and obtaining a semantic matching result;
the hierarchy judging unit is used for judging the hierarchy to which the client belongs through a preset threshold according to the semantic matching result.
4. The customer service response system based on artificial intelligence according to claim 1, wherein the customer management center further comprises an interactive data recording module and a customer level updating module;
the interactive data recording module is used for recording behavior data of clients and receiving response data fed back by the customer service response center;
the client level updating module is used for updating the level to which the client belongs according to the newly added interaction data record.
5. The customer service response system based on artificial intelligence according to claim 1, wherein the customer service response center comprises an information acquisition module, a response template adjustment module and a customer service response module;
the information acquisition module is used for acquiring information of a client, wherein the information of the client comprises historical interaction data of the client and a hierarchy to which the client belongs;
the response template adjusting module is used for extracting a corresponding response template according to the hierarchy to which the client belongs and adjusting the response template according to the historical interaction data;
the customer service response module is used for responding according to the adjusted response template and synchronizing response data to the customer management center.
6. The customer service response system based on artificial intelligence according to claim 5, wherein the response template adjusting module comprises a style feature extracting unit and a style merging unit;
the style characteristic extraction unit is used for acquiring style characteristic information of the client through a preset method according to historical interaction data of the client;
the style fusion unit is used for fusing the client style characteristic information with the response template through a preset method to generate an adjusted response template.
7. A customer service response system based on artificial intelligence according to claim 1, wherein the customer service response center further comprises a response transfer module;
the response transfer module is used for outputting prompt information when the information submitted by the client contains preset keywords in the response process, and is used for prompting the response of the manual customer service catcher.
8. The customer service response method based on artificial intelligence is characterized by comprising the following steps of:
acquiring information of a current client, and judging whether the current client is a new client or not through preset database matching;
if yes, responding through a preset initial response template;
if not, confirming the hierarchy to which the client belongs through a preset database according to the client information, and acquiring a corresponding hierarchy response template;
extracting historical interaction data of the client through a preset database, and adjusting the acquired hierarchical response templates according to the historical interaction data to acquire adjusted response templates;
and responding through the adjusted response template, and simultaneously recording response data generated in the whole response process.
9. The customer service response method based on artificial intelligence according to claim 7, wherein the adjusting the acquired hierarchical response templates according to the historical interaction data to acquire adjusted response templates comprises:
according to the historical interaction data, through a preset method or style characteristic information of a client;
based on style characteristic information, fusing the style characteristic information with the current response template through a preset method to generate an adjusted response template.
10. A computer readable storage medium storing a computer program capable of being loaded by a processor and executing an artificial intelligence based customer service answering method according to any one of claims 8 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311262679.7A CN117221451B (en) | 2023-09-27 | 2023-09-27 | Customer service response system and method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311262679.7A CN117221451B (en) | 2023-09-27 | 2023-09-27 | Customer service response system and method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117221451A true CN117221451A (en) | 2023-12-12 |
| CN117221451B CN117221451B (en) | 2024-04-26 |
Family
ID=89046063
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311262679.7A Active CN117221451B (en) | 2023-09-27 | 2023-09-27 | Customer service response system and method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117221451B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112214685A (en) * | 2020-09-27 | 2021-01-12 | 电子科技大学 | Knowledge graph-based personalized recommendation method |
| CN113313507A (en) * | 2020-02-27 | 2021-08-27 | 北京有限元科技有限公司 | Method, device and storage medium for improving marketing precision of telephone operation |
| TWM619682U (en) * | 2021-08-04 | 2021-11-11 | 中國信託商業銀行股份有限公司 | Digital marketing decision-making system based on product familiarity prediction |
| CN115114453A (en) * | 2022-06-29 | 2022-09-27 | 中国银行股份有限公司 | Intelligent customer service implementation method and device based on knowledge graph |
| CN116541504A (en) * | 2023-06-27 | 2023-08-04 | 北京聆心智能科技有限公司 | Dialog generation method, device, medium and computing equipment |
| CN116645134A (en) * | 2023-05-30 | 2023-08-25 | 中国银行股份有限公司 | Method, device, equipment and medium for recommending credit card in stages |
-
2023
- 2023-09-27 CN CN202311262679.7A patent/CN117221451B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113313507A (en) * | 2020-02-27 | 2021-08-27 | 北京有限元科技有限公司 | Method, device and storage medium for improving marketing precision of telephone operation |
| CN112214685A (en) * | 2020-09-27 | 2021-01-12 | 电子科技大学 | Knowledge graph-based personalized recommendation method |
| TWM619682U (en) * | 2021-08-04 | 2021-11-11 | 中國信託商業銀行股份有限公司 | Digital marketing decision-making system based on product familiarity prediction |
| CN115114453A (en) * | 2022-06-29 | 2022-09-27 | 中国银行股份有限公司 | Intelligent customer service implementation method and device based on knowledge graph |
| CN116645134A (en) * | 2023-05-30 | 2023-08-25 | 中国银行股份有限公司 | Method, device, equipment and medium for recommending credit card in stages |
| CN116541504A (en) * | 2023-06-27 | 2023-08-04 | 北京聆心智能科技有限公司 | Dialog generation method, device, medium and computing equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117221451B (en) | 2024-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5639544B2 (en) | Method for determining a relative preference of one or more consumers from a group of consumers for a form of product using a computer system including a processor and memory | |
| Park et al. | A Bayesian network approach to examining key success factors of mobile games | |
| CN107507612A (en) | A kind of method for recognizing sound-groove and device | |
| CN1980333A (en) | Method and device for generating media demo | |
| CN1980332A (en) | Method and device for controlling addition of personal hobby information into media demo | |
| CN107123057A (en) | User recommends method and device | |
| WO1998002835A1 (en) | A method and apparatus for expertly matching products, services, and consumers | |
| KR102592310B1 (en) | System and method for providing virtual reality service based on artificial intelligence | |
| CN108454114A (en) | A kind of customization platform and its method for customizing for 3D printing | |
| CN102201188A (en) | Building television advertisement system oriented intelligent control device and method | |
| CN110120001A (en) | The method and system that a kind of knowledge based spectrum library mentions point in conjunction with memory curve | |
| JPWO2017191696A1 (en) | Information processing system and information processing method | |
| CN111768239A (en) | Property recommendation method, device, system, server and storage medium | |
| CN111400525B (en) | Fashion clothing intelligent matching and recommending method based on vision combination relation learning | |
| KR20190103641A (en) | System and method for providing related services of immersive virtual fitting | |
| Mallalieu et al. | Understanding the role of consumer motivation and salesperson behavior in inducing positive cognitive and emotional responses during a sales encounter | |
| KR20220140226A (en) | Method, apparatus and computer program for determining contents | |
| US20220050972A1 (en) | Intelligent discovery multi-turn open dialogue agent | |
| Liu et al. | From llm to conversational agent: A memory enhanced architecture with fine-tuning of large language models | |
| CN117221451B (en) | Customer service response system and method based on artificial intelligence | |
| WO2021049254A1 (en) | Information processing method, information processing device, and program | |
| US20010023405A1 (en) | Method, apparatus, and computer program for generating a feeling in consideration of agent's self-confident degree | |
| CN112200615A (en) | Digital marketing service system based on big data | |
| US20060141426A1 (en) | Virtual scene interactive language learning system and method | |
| CN116129001A (en) | Virtual digital application method for electric power |
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 |