CN115544222A - Artificial intelligence interactive system - Google Patents

Abstract

An artificial intelligence interactive system comprises a visitor end, a response end and an artificial intelligence module, wherein the visitor end is used for being connected with the artificial intelligence module; the response end is used for connecting the artificial intelligence module; the artificial intelligence module is used for circularly executing the step R after the visitor end is successfully connected with the response end; wherein the step R comprises the following instructions: according to the scheme, the response of an operator at the response end can be facilitated, and the situation that the visitor directly receives the suggested response sentence and the artificial intelligence response sentence is too hard to influence the user experience is avoided.

Description

Artificial intelligence interactive system

Technical Field

The invention relates to the field of human-computer interaction, in particular to a human-computer interaction artificial intelligence auxiliary interaction system.

Background

The automatic customer service has gone deep into various industries, can carry out simple problem retrieval and display, still is not intelligent enough, and especially under some circumstances, what the user hoped to talk with oneself is more people rather than the machine, in order to promote the artificial intelligence's of this aspect problem answer level, and can reduce the work load of corresponding service personnel operation, need newly design an artificial intelligence's optimization method.

Disclosure of Invention

Therefore, an artificial intelligence interaction method capable of assisting in session question answering needs to be provided, relevant artificial intelligence conversation assistance work is carried out, and the problem that user experience is poor in the human-computer interaction process in the prior art is solved;

in order to achieve the above object, the inventor provides an artificial intelligence interactive system, which comprises a visitor terminal, a response terminal and an artificial intelligence module,

the visitor terminal is used for connecting the artificial intelligence module;

the response end is used for connecting the artificial intelligence module;

the artificial intelligence module is used for circularly executing the step R after the visitor end is successfully connected with the response end;

wherein the step R comprises the following instructions:

acquiring first visitor conversation information received by a visitor end, performing text analysis on the first visitor conversation information by the artificial intelligence module to generate a suggested answer sentence, sending the suggested answer sentence to a response end, presenting the suggested answer sentence by the response end, and providing an operation function of selecting the suggested answer sentence and/or editing the suggested answer sentence and/or sending the suggested answer sentence; and receiving a first editing result of the response end and sending the first editing result to the visitor end.

In some embodiments of the present application, the step R further includes the following instructions: receiving a first editing result of a response end, sending the first editing result into an artificial intelligence module, carrying out differential analysis on the first editing result and the suggested response sentence, and changing the state of artificial intelligence according to the differential analysis result.

In some embodiments of the application, the altering the state of the artificial intelligence comprises: altering the material library of the artificial intelligence and/or altering the model parameters of the artificial intelligence and/or altering the analysis dimensions of the artificial intelligence and/or altering the weights of the artificial intelligence.

In some embodiments of the application, the artificial intelligence module is specifically configured to perform the steps,

when a response end receives the non-selected operation result of the suggested response statement, combining the suggested response statement and the first visitor dialogue information into a counter-example matching data pair, adding an artificial intelligence neural network for training, and updating the artificial intelligence neural network model weight;

and when the answering end receives an operation result of editing the suggested answer sentence, the edited suggested answer sentence is used as the expected output of the artificial intelligent neural network, the first visitor dialogue information is used as the input of the neural network, a matching data pair is obtained and used as a training material, and the neural network model is trained or updated.

In some embodiments of the present application, the visitor session information includes voice information, and further includes a text conversion module, where the text conversion module is configured to perform voice breakpoint detection and voice recognition on the first visitor session information, convert the first visitor session information into a text, and then use the text to perform text analysis on the first visitor session information by using artificial intelligence.

In some embodiments of the application, the visitor session information includes picture information, and further includes a text conversion module, where the text conversion module is used for performing text extraction on the first visitor session information and then using the artificial intelligence to perform text analysis on the first visitor session information.

In some embodiments of the application, the guest includes a phone, a mobile application or a wechat applet, and the first guest dialog information is in a natural language.

In some embodiments of the present application, the artificial intelligence module is further configured to exit the loop of performing step R under a preset condition, where the preset condition includes that the guest closes the session.

In some embodiments of the application, the response end is further used for displaying the visitor session information and the suggested response sentence through different interaction paths, the interaction paths comprise visual character interaction and auditory voice interaction, and the artificial intelligence is obtained through the forwarding function of the response end.

In some embodiments of the application, the response end is further configured to receive a forwarding instruction of the visitor session information, and the forwarding instruction is received before forwarding the corresponding visitor session information to the artificial intelligence.

Through the scheme, the connection can be established between the visitor end and the answering end, the suggestion answering can be carried out on the first visitor conversation information through artificial intelligence, the suggestion answering sentence is sent to the answering end, meanwhile, the editing, selecting and sending functions of the suggestion answering sentence are provided, the visitor end can receive the edited content, the answering end can conveniently answer the operation personnel, and meanwhile, the situation that the visitor directly receives the suggestion answering sentence and the artificial intelligence response sentence is too hard to influence the user experience is avoided.

Drawings

FIG. 1 is a flowchart of an artificial intelligence interaction method according to an embodiment of the present invention;

FIG. 2 is a flowchart of a reverse query method according to an embodiment of the present invention;

FIG. 3 is a flowchart of a counter-example matching data method according to an embodiment of the present invention;

FIG. 4 is a diagram illustrating a method for updating weights of a neural network model according to an embodiment of the present invention;

fig. 5 is a flowchart of a guest dialog message conversion method according to an embodiment of the present invention;

fig. 6 is a schematic view of an operation interface of the response end according to the embodiment of the present invention;

FIG. 7 is a diagram of an artificial intelligence interaction system according to an embodiment of the present invention.

Detailed Description

In order to explain in detail possible application scenarios, technical principles, practical embodiments, and the like of the present application, the following detailed description is given with reference to the accompanying drawings in conjunction with the listed embodiments. The embodiments described herein are merely for more clearly illustrating the technical solutions of the present application, and therefore, the embodiments are only used as examples, and the scope of the present application is not limited thereby.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase "an embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or related to other embodiments specifically defined. In principle, in the present application, the technical features mentioned in the embodiments can be combined in any manner to form a corresponding implementable technical solution as long as there is no technical contradiction or conflict.

Unless defined otherwise, technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the use of relational terms herein is intended only to describe particular embodiments and is not intended to limit the present application.

In the description of the present application, the term "and/or" is a expression for describing a logical relationship between objects, meaning that three relationships may exist, for example a and/or B, meaning: there are three cases of A, B, and both A and B. In addition, the character "/" herein generally indicates that the former and latter associated objects are in a logical relationship of "or".

In this application, terms such as "first" and "second" are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.

Without further limitation, in this application, the use of "including," "comprising," "having," or other similar expressions in phrases and expressions of "including," "comprising," or "having," is intended to cover a non-exclusive inclusion, and such expressions do not exclude the presence of additional elements in a process, method, or article that includes the recited elements, such that a process, method, or article that includes a list of elements may include not only those elements but also other elements not expressly listed or inherent to such process, method, or article.

As is understood in the examination of the guidelines, the terms "greater than", "less than", "more than" and the like in this application are to be understood as excluding the number; the expressions "above", "below", "within" and the like are understood to include the present numbers. Furthermore, the description of embodiments herein of the present application of the term "plurality" means more than two (including two), and the analogous meaning of "plurality" is also to be understood, e.g., "plurality", etc., unless explicitly specified otherwise.

Referring to fig. 1, an artificial intelligence interaction method includes the following steps,

s10, performing visitor connection;

s11, executing response end connection;

s12, after detecting that the visitor end and the response end are successfully connected with the artificial intelligence module, circularly executing the step R;

wherein the step R comprises the following instructions:

s131, acquiring first visitor conversation information received by a visitor, performing text analysis on the first visitor conversation information by the artificial intelligence module to generate a suggested response sentence, S132 sending the suggested response sentence to a response end, S133 presenting the suggested response sentence by the response end, and providing an operation function of selecting the suggested response sentence and/or editing the suggested response sentence and/or sending the suggested response sentence; and receiving a first editing result of the response end and sending the first editing result to the visitor end.

The above scheme is used for adapting an artificial intelligence interactive system, such as an instant messaging window, a telephone voice chat and the like in the prior art, and the visitor end and the answering end can be a telephone, a telephone watch, a smart phone, a computer or other equipment. The user of the guest accesses one of the parties in the session process, such as the party who needs to provide a consultation question, or the input statement information of the user party. The first guest dialog information may be session content received by the guest. The response end is an equipment end for showing visitor conversation information to an operator, can be communicated with the user side equipment through a network, and the operator of the response end can be a response party. The artificial intelligence can be a text analysis neural network/deep learning model, and can output a corresponding suggested response sentence according to the input dialog information of the first visitor, or can be a neural network/deep learning model of voice-to-text analysis. The operation function provided by the response end can be an entity key, a virtual key such as an icon, and an enabled program identified by a method such as voice and gesture. Through the scheme, the technical effects of displaying the suggested answer sentences for reference, editing according to the operation actions of the operator and sending the edited answer sentences to the visitor end can be achieved.

In some specific embodiments, for example, a visitor user (party a) operates a visitor to connect to an artificial intelligence module, and to perform a question consultation, an operator using a response end connects to the artificial intelligence module for a consultant (party b). The response end in the conversation process can display the visitor conversation information and give out corresponding suggested response sentences, the suggested response sentences are output after the visitor conversation information is input into the artificial intelligent model, the suggested response sentences can be a plurality of output results or a single output result, and the results have different weights. One or more of the results may be presented to the operator. Meanwhile, in order to avoid the first party from obtaining error information, the response sentence is recommended not to be displayed to the first party. The answering terminal provides an operation function of selecting the suggested answer sentence and/or editing the suggested answer sentence and/or sending the suggested answer sentence, and the function can perform the following steps: and receiving the selection and/or editing operation of the second party on the suggested answer sentence, and then sending the first editing result. Therefore, the information received by the first party is ensured to be manually selected, and the false response caused by the direct response of artificial intelligence is avoided, or the first party is provided with a rigid feeling of talking with the robot. For example, the artificial intelligence model gives a suggested answer sentence A \ B \ C, and the ranking is the weight descending order. The second party can select B sentences, edit and modify B1 sentences, and finally reply to the first party with B sentences. The result of such an operation by party B also includes non-selected operations for the A and C statements.

In some embodiments of the present application, the step R further includes the following instructions: receiving a first editing result of the response end, sending the first editing result into an artificial intelligence module, carrying out differential analysis on the first editing result and the suggested response sentence, and changing the state of artificial intelligence according to the differential analysis result. For example, in the above scheme, the selection of the B statement is also used for state updating affecting artificial intelligence, and the first editing result B1 after the editing and modification of the B statement can also be used for state updating affecting artificial intelligence, and the first editing result can be immediately used as the input of the artificial intelligence module.

In other specific embodiments, the artificial intelligence model gives a suggested answer sentence a \ B \ C, the second party may not select any answer sentence, and finally gives an answer sentence D, so that the operation result of the second party further includes operations of non-selection and non-editing for the sentences A, B and C, and also includes operations and contents of replying the answer sentence D. The state updating system is also used for influencing artificial intelligence, and through the scheme, the answer content of the artificial intelligence can be continuously promoted in the conversation process of the two parties, namely the first party and the second party, so that the real-time promotion is realized, and the intelligent technical effect of the artificial intelligence is finally promoted better.

In other particular embodiments, updating the state of the artificial intelligence includes: altering the material library of the artificial intelligence and/or altering the model parameters of the artificial intelligence and/or altering the analysis dimensions of the artificial intelligence and/or altering the weights of the artificial intelligence. For example, when the operation result of the second party includes that the reply sentence is D, a new reply sentence D may be used to add the material library of artificial intelligence. When the operation structure of the second party comprises non-selection operations for the statements A and C, the weight reduction of A and C in the output result of the artificial intelligence model can be carried out according to the non-selection operations. When the operation result of the second party comprises the selection operation of the statement B, the weight of the statement B in the output result of the artificial intelligence model can be improved. For the artificial intelligence model, the analysis dimensionality of the artificial intelligence, such as the number of layers of an analysis layer, can be automatically adjusted according to the change of the weight of an output result, and the weight change between layers and between classifiers in the artificial intelligence can also be automatically adjusted. These automatic adjustments are determined by the characteristics of the artificial intelligence model, and human behavior may not be actively adjusted, but any step that seeks to alter the analysis dimensions of the artificial intelligence and/or alter the changes in the weights of the artificial intelligence, such as altering the training set, altering the output results, etc., are included in this example.

In some specific embodiments, the method further comprises the steps of:

when the responder receives the result of the non-selection operation on the proposed responder statement, the result may indicate that the input/output matching data pair related to the responder in the training data has an error. There are two possible ways of handling this. In one embodiment, as shown in fig. 2, the scheme may perform the steps of, S21, reversely querying the training data, finding out the matching data pairs in the training data that are close to the training data by using a method such as text similarity calculation, S22, directly deleting the matching data pairs or deleting the matching data pairs through manual verification, and then performing the step, S23, retraining the model, and updating the model weights. In other embodiments, as shown in fig. 3, steps may be performed, in which S31, directly extracts and combines the unselected proposed answer sentences and the corresponding consultation questions of the first party to form one or more (there are multiple cases depending on the sentence break arrangement and combination of the two) counter-example matching data pairs, and adds an artificial intelligence neural network to train, and S32, updates the artificial intelligence neural network model weight.

In other embodiments as shown in fig. 4, the scheme may further perform a step, when the responder receives an operation result of editing the proposed answer sentence, S41 uses the edited proposed answer sentence as an expected output of the artificial intelligence neural network, and uses a question corresponding to the proposed answer sentence as an input of the neural network, so as to form a set of input/output matching data pairs. By collecting multiple sets of matching data pairs, the training materials can be used to train or update the neural network model, i.e., update the connection weights in the neural network model. According to the scheme, the state of the artificial intelligence can be updated according to the action result of the corresponding answer end, so that the effect of improving the intelligence of the artificial intelligence session assistance in real time can be achieved.

In other embodiments, the guest includes a phone, a mobile application, or a wechat applet. The visitor can access through the telephone to carry out voice consultation, can also carry out networking through a mobile terminal application program, then accesses into the conversation process of the system to carry out consultation, can also carry out networking through modes such as a WeChat applet and the like, and accesses into the conversation process of the system to carry out consultation. The answering end of the second party may be a dialog window of a desktop end, such as a software program running in a windows and linux system, or any one of the above-mentioned phone, a mobile end application program or a wechat applet.

In some optional embodiments, the access is by telephone or chat, the visitor session information of the session flow includes voice information, and the artificial intelligence performs an artificial intelligence model for text analysis. As shown in fig. 5, a step is specifically performed in the present solution, after obtaining the visitor session information, S51 performs text conversion on the visitor session information, such as performing voice breakpoint detection and voice recognition, and then performs text analysis on the first visitor session information by using artificial intelligence. The artificial intelligence model has strong technical adaptability for analyzing and classifying texts, is low in deployment cost, and can meet the requirement of being more beneficial to artificial intelligence to analyze by uniformly converting visitor conversation information into a text form. In a specific application, speech recognition and speech-to-text conversion can be performed through a deep learning model or a neural network model.

In other specific embodiments, the input guest dialog message may include picture information, and the method further specifically performs step S52 of performing picture text extraction on the guest dialog message. The first guest dialog information is then text analyzed using artificial intelligence. The artificial intelligence model has strong technical adaptability for analyzing and classifying texts, is low in deployment cost, and can meet the requirement of being more beneficial to artificial intelligence to analyze by uniformly converting visitor conversation information into a text form. In a specific application, text recognition and graph-text conversion in pictures can be performed through a deep learning model or a neural network model.

In other embodiments, the visitor session information and the suggested response sentence are displayed through different interfaces of the response end. In the embodiment shown in fig. 6, an operation interface of the answering terminal is shown, and includes a plurality of different window interfaces, where the first window 60 may be used to show dialog information of a visitor, and the first window 60 is also used to receive an input sentence of the second party, for example, to put information to be sent typed by the second party in a to-be-sent column. While a second window 61 may be used to present suggested answer sentences. The second window 61 may be configured to receive a selection operation of the second party on the proposed answer sentence, and then display the selection operation on the information to be sent in the first window 60, and the second party may edit the information to be sent, and finally determine the sentence to be sent.

In some embodiments of the present application, the method further includes a preset condition of exiting the loop of executing step R, and the loop of exiting step R is executed when the preset condition is triggered. In some embodiments, the predetermined condition includes that the guest closes the session, and the loop of step R is exited. The preset condition can also be set as that the response end stops the artificial intelligence auxiliary function, if the response end provides a switch option for judging whether to carry out auxiliary response, if the response end starts the auxiliary response, the step R is executed circularly, the artificial intelligence gives a suggested response sentence to the first visitor session information, the artificial intelligence can continuously update the state of the artificial intelligence according to the editing result, and if the auxiliary response function is closed, the artificial intelligence does not respond to the first visitor session information and does not carry out learning update.

In other embodiments, the method further comprises the step of displaying the visitor conversation information and the suggested response sentence through different interaction paths of the response end, wherein the interaction paths comprise visual word interaction and auditory voice interaction. For example, at the response end, the visitor conversation information and the content replied by the consultant can be displayed in a mode of displaying visual character interactive information in a window, but an interactive path of the suggested response sentence is different from that of the visitor conversation information, the suggested response sentence is transmitted to an earphone of the response end in a voice mode to be played, and the consultant can autonomously determine the content replied according to voice prompt. For another example, at the response end, the counselor and the visitor can directly perform voice conversation interaction in an auditory voice interaction mode, and at the response end, the alternative result of the suggested response sentence can be displayed through an interaction path of the display screen, so that the counselor can see the suggested response sentence on the display screen. The consultant autonomously decides the reply content of the voice conversation to the guest user according to the prompt of the suggested reply sentence on the display screen. Through the scheme, visitor conversation information and suggested response sentences can be displayed to the consultant through different communication routes, misoperation caused by the fact that the consultant is confused due to carelessness or fatigue is avoided, and the technical effect of reducing possibility of misoperation through simple fool prevention is achieved.

In some other specific embodiments of the present disclosure, the artificial intelligence obtains the visitor session information through a forwarding function of the response end. For example, artificial intelligence can deploy at the high in the clouds server, and the networking route can directly be networked response end and user, and the response end is given the high in the clouds through transmitting visitor dialogue information, then receives the result that returns of high in the clouds server, and this kind of mode can save the size of the shared memory of response end, can arrange the response end more easily. Meanwhile, the permission whether to forward the visitor session information can be left at the answering end, so that the server is prevented from acquiring excessive personal privacy.

In a further specific implementation, the response end receives a forwarding instruction for the visitor session information, and forwards the corresponding visitor session information to the artificial intelligence after receiving the forwarding instruction. The forwarding instruction may generally be operated by a consultant. In some embodiments, the response terminal can communicate normally during conversation, consultants can send a forwarding instruction in a mode of pressing keys, clicking a mouse and the like when artificial intelligence intervention is needed, and the response terminal can send corresponding visitor conversation information to the artificial intelligence after receiving the forwarding instruction. The artificial intelligence will then proceed with the subsequent step of presenting the proposed answer sentence. Through the scheme, the technical effect of carrying out artificial intelligence access according to the demands of the consultants can be achieved.

In the embodiment shown in fig. 7, there is also described an artificial intelligence interaction system, comprising a guest 70, a responder 71, an artificial intelligence module 72,

the visitor end 70 is used for connecting an artificial intelligence module 72;

the response end 71 is used for connecting an artificial intelligence module 72;

the artificial intelligence module 72 is configured to execute the step R in a circulating manner after the guest 70 and the responder 71 are successfully connected; ,

wherein the step R comprises the following instructions:

acquiring first visitor conversation information received by a visitor terminal 70, performing text analysis on the first visitor conversation information by the artificial intelligence module 72 to generate a suggested answer sentence, sending the suggested answer sentence to an answer terminal 71, presenting the suggested answer sentence by the answer terminal 71, and providing an operation function of selecting the suggested answer sentence and/or editing the suggested answer sentence and/or sending the suggested answer sentence; the first editing result of the answering end is received and sent to the guest end 70.

The system can establish connection between the visitor end 70 and the answering end 71, can also propose to the first visitor dialogue information through the artificial intelligence module 72 to answer, send the proposed answering sentence to the answering end 71, and simultaneously provide editing, selecting and sending functions to the proposed answering sentence, and the visitor end 70 can receive the edited content, so that the visitor end 71 can conveniently answer, and meanwhile, the visitor can be prevented from directly receiving the proposed answering sentence and the artificial intelligence answering sentence is too hard to influence the user experience.

Optionally, the step R further includes the following instructions: receiving a first editing result of the response end 71, sending the first editing result to the artificial intelligence module 72, performing differential analysis on the first editing result and the suggested response sentence, and changing the state of artificial intelligence according to the differential analysis result.

Optionally, altering the state of the artificial intelligence comprises: altering the material library of the artificial intelligence and/or altering the model parameters of the artificial intelligence and/or altering the analysis dimensions of the artificial intelligence and/or altering the weights of the artificial intelligence.

Optionally, the artificial intelligence module 72 is specifically configured to:

when the response end 71 receives the non-selected operation result of the suggested response sentence, the suggested response sentence and the first visitor session information are combined into a counterexample matching data pair, and an artificial intelligence neural network is added for training to update the artificial intelligence neural network model weight;

when the response end 71 receives the operation result of editing the suggested response sentence, the edited suggested response sentence is used as the expected output of the artificial intelligent neural network, the first visitor session information is used as the input of the neural network, a matching data pair is obtained and used as a training material, and the neural network model is trained or updated.

Optionally, the guest 70 includes a phone, a mobile application, or a wechat applet.

Optionally, the visitor session information includes a voice message, and further includes a voice recognition module, where the voice recognition module is configured to perform text conversion on the first visitor session information and send the first visitor session information to the artificial intelligence module 72.

Further, the visitor session information includes picture information, and further includes a picture recognition module, where the picture recognition module is configured to extract a picture text of the first visitor session information and send the extracted picture text to the artificial intelligence module 72.

Optionally, the response end 71 further includes different interfaces, and the response end 71 is configured to respectively display the visitor dialog message and the suggested answer sentence through the different interfaces.

Optionally, the response end 71 is further configured to display the visitor dialog message and the suggested response sentence through different interaction paths, where the interaction paths include visual text interaction and auditory voice interaction.

Optionally, the answering terminal 71 further comprises a forwarding function for forwarding the guest dialog information, and the artificial intelligence module 72 is configured to receive the guest dialog information.

Optionally, the response end 71 is configured to receive a forwarding instruction for the guest dialog message, and further configured to forward the corresponding guest dialog message to the artificial intelligence after receiving the forwarding instruction.

The artificial intelligence auxiliary system can obtain the operation of selecting or not selecting the consultation sentences by the user as feedback to adjust the response strategy of the artificial intelligence, so that a better artificial intelligence auxiliary effect is obtained.

It should be noted that, although the above embodiments have been described herein, the scope of the present invention is not limited thereby. Therefore, based on the innovative concepts of the present invention, the technical solutions of the present invention can be directly or indirectly applied to other related technical fields by making changes and modifications to the embodiments described herein, or by using equivalent structures or equivalent processes performed in the content of the present specification and the attached drawings, which are included in the scope of the present invention.

Claims (10)

1. An artificial intelligence interactive system is characterized in that the system comprises a visitor end, a response end and an artificial intelligence module,

the visitor terminal is used for connecting the artificial intelligence module;

the response end is used for connecting the artificial intelligence module;

the artificial intelligence module is used for circularly executing the step R after the visitor end is successfully connected with the response end; ,

wherein the step R comprises the following instructions:

acquiring first visitor conversation information received by a visitor end, performing text analysis on the first visitor conversation information by the artificial intelligence module to generate a suggested response sentence, sending the suggested response sentence to a response end, presenting the suggested response sentence by the response end, and providing an operation function of selecting the suggested response sentence and/or editing the suggested response sentence and/or sending the suggested response sentence; and receiving a first editing result of the response end and sending the first editing result to the visitor end.

2. The artificial intelligence interaction system of claim 1, wherein step R further comprises instructions to: receiving a first editing result of a response end, sending the first editing result into an artificial intelligence module, carrying out differential analysis on the first editing result and the suggested response sentence, and changing the state of artificial intelligence according to the differential analysis result.

3. The artificial intelligence interaction system of claim 2, wherein the altering of the state of artificial intelligence comprises: altering the material base of the artificial intelligence and/or altering the model parameters of the artificial intelligence and/or altering the analysis dimension of the artificial intelligence and/or altering the weighting of the artificial intelligence.

4. The artificial intelligence interaction system of claim 2, wherein the artificial intelligence module is specifically configured to perform the steps,

when a response end receives a non-selected operation result of the suggested response statement, combining the suggested response statement and the first visitor session information into a counterexample matching data pair, adding an artificial intelligence neural network for training, and updating the artificial intelligence neural network model weight;

and when the answering end receives an operation result of editing the suggested answer sentence, the edited suggested answer sentence is used as the expected output of the artificial intelligent neural network, the first visitor dialogue information is used as the input of the neural network, a matching data pair is obtained and used as a training material, and the neural network model is trained or updated.

5. The system of claim 1, wherein the visitor session information comprises voice information, and further comprising a text conversion module, wherein the text conversion module is configured to perform voice breakpoint detection and voice recognition on the first visitor session information, and to perform text analysis on the first visitor session information using the artificial intelligence after converting the first visitor session information into text.

6. The artificial intelligence interaction system of claim 1, wherein the guest dialog information comprises picture information, further comprising a text conversion module, the text conversion module configured to perform picture text extraction on the first guest dialog information and then perform text analysis on the first guest dialog information using the artificial intelligence.

7. The system of claim 1, wherein the guest comprises a phone, a mobile application, or a wechat applet, and the first guest dialog message is in a natural language.

8. The system according to claim 1, wherein the artificial intelligence module is further configured to exit the loop of performing step R under a predetermined condition, the predetermined condition comprising a guest closing the session.

9. The interactive artificial intelligence system of claim 1, wherein the response end is further configured to display the visitor session information and the suggested response sentence through different interaction paths, the interaction paths include visual text interaction and auditory voice interaction, and the artificial intelligence obtains the visitor session information through a forwarding function of the response end.

10. The artificial intelligence interaction system of claim 1, wherein the answering terminal is further configured to receive a forwarding instruction for the guest dialog message, and forward the corresponding guest dialog message to artificial intelligence after receiving the forwarding instruction.

Images