Human-computer dialogue process testing method and system

Abstract

The present invention discloses a human-computer dialogue process testing system, including a robot management device, a testing device, and a recording device; the robot management device creates a voice robot and configures the voice robot; the robot management device sends the voice robot information to the testing device; the testing device tests the human-computer dialogue process of the voice robot and generates corresponding test results, and sends the test result feedback to the dialogue robot management device, and the dialogue robot management device optimizes the configuration of the dialogue robot according to the test results, thereby forming a test closed loop; the recording device records the test results and data. In addition, the present invention also discloses a human-computer dialogue process testing method. The present invention makes the test process of the voice robot clear and visible, and has high reusability, and can timely discover and locate problems that occur in the robot during the construction and intention understanding process, greatly improving the test efficiency of the voice robot.

Description

Technical Field

[0001] The present invention relates to the technical field of call centers, and in particular to a method and system for testing a human-computer dialogue process. Background Art

[0002] There are two major defects in the testing of human-computer dialogue processes in the existing technology. First, the verification and optimization of the human-computer dialogue process is usually completed by real-time scene dialogue. The testing process is not transparent, and it is impossible to view the variable assignment, understanding judgment and other information of the voice robot in each round of dialogue in real time; secondly, in terms of understanding testing, the existing technology does not support test case retention. After each optimization of the voice robot, the test case needs to be rewritten to verify the optimization results. Summary of the invention

[0003] Based on this, in order to solve the technical problems in the existing technology, a human-computer dialogue process testing method is proposed, including:

[0004] A voice robot is created through a robot management device, and after creation, the voice robot is configured to complete the construction of the voice robot; the robot management device sends the information of the completed voice robot to a test device connected thereto, and the test device tests the human-machine dialogue process of the voice robot;

[0005] The human-machine dialogue process of the voice robot is tested by a testing device and a corresponding test result is generated. The testing device sends the test result and data to a recording device connected thereto; the testing device includes a process tester and an understanding tester; the process tester is used to check whether there are errors in the syntax of the dialogue flow chart, and to detect the process execution, node jump, and variable assignment of the voice robot in the human-machine dialogue process; the understanding tester is used to detect whether the voice robot accurately understands the customer's intention in the human-machine dialogue process;

[0006] The testing device sends the test result feedback to the conversational robot management device, and the conversational robot management device optimizes the configuration of the conversational robot according to the test result, thereby forming a test closed loop;

[0007] The recording device records the test results and data, and saves the correct number and accuracy rate of the conversational robot's understanding of the customer's conversation response content calculated after each test is completed in the test results.

[0008] In one embodiment, the flow tester includes a dialog flow chart verifier, which verifies the configuration and syntax of the dialog flow chart; in the dialog flow chart verification, it is verified whether the dialog flow chart contains configuration errors or syntax errors to ensure that the dialog flow will not be interrupted; after verification, the dialog flow chart verifier obtains the name, location, description, and repair suggestion of the error or warning in the dialog flow chart;

[0009] Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result;

[0010] The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces. The description of an error or warning refers to a detailed description of the error or warning. The suggestion about an error or warning refers to a suggestion for fixing the error or warning.

[0011] In one embodiment, the process tester includes a dialogue tester, which simulates human-machine conversations and detects the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations; the content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay;

[0012] Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process;

[0013] The test starting point indicates which node in the current process should be selected for testing through the dialogue tester;

[0014] The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node.

[0015] Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain;

[0016] Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling;

[0017] The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call;

[0018] The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester;

[0019] Dialogue replay means observing the voice robot's intention judgment in the current node through the dialogue tester, and observing the voice robot's judgment results for each intention through dialogue replay until the customer intention in the node is found that is the same as the recognition result and the next step is executed.

[0020] In one embodiment, the understanding tester includes a quick understanding tester, which quickly tests the voice robot's understanding of the content of the customer conversation response by inputting a test text; the quick understanding tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the voice robot's correct understanding number and understanding accuracy rate regarding the intention understanding after the quick test is completed.

[0021] In one embodiment, the understanding tester includes a test set tester, which tests the intention understanding of the voice robot in the human-computer dialogue process through a test set;

[0022] In test set testing, the test set tester uploads a reusable test set before testing. The test set includes one or more test cases.

[0023] Whenever the voice robot is optimized, the test set tester references the uploaded test set to test the voice robot's understanding ability in the human-machine dialogue process; the test set tester displays and compares the expected and actual results of intent recognition for each test case, and calculates the number of correct understandings and the accuracy rate of intent understanding;

[0024] After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage.

[0025] In addition, in order to solve the technical problems in the prior art, a human-machine dialogue process testing system is proposed, which includes a robot management device, a testing device, and a recording device; the robot management device is connected to the testing device; the testing device is connected to the recording device;

[0026] The robot management device is used to create a voice robot, configure the voice robot after creation, and complete the construction of the voice robot; the robot management device sends the information of the completed voice robot to the testing device, and the testing device tests the human-computer dialogue process of the voice robot;

[0027] The testing device is used to test the human-computer dialogue process of the voice robot and generate corresponding test results, and send the test results and data to the recording device; the testing device includes a process tester and an understanding tester; the process tester is used to check whether there are errors in the syntax of the dialogue flow chart, and detect the process execution, node jump, and variable assignment of the voice robot in the human-computer dialogue process; the understanding tester is used to detect whether the voice robot accurately understands the customer's intention in the human-computer dialogue process;

[0028] The testing device sends the test result feedback to the conversational robot management device, and the conversational robot management device optimizes the configuration of the conversational robot according to the test result, thereby forming a test closed loop;

[0029] Among them, the recording device is used to record the test results and data, and the test results save the correct number and accuracy rate of the conversational robot's understanding of the customer's conversation response content calculated after each test is completed.

[0030] In one embodiment, the process tester includes a dialog flow chart verifier;

[0031] The dialog flow chart verifier is used to verify the configuration and syntax of the dialog flow chart. In the dialog flow chart verification, the dialog flow chart is verified to see whether it contains configuration errors or syntax errors to ensure that the dialog flow is not interrupted. After verification, the dialog flow chart verifier obtains the name, location, description, and repair suggestion of the error or warning in the dialog flow chart.

[0032] Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result;

[0033] The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces. The description of an error or warning refers to a detailed description of the error or warning. The suggestion about an error or warning refers to a suggestion for fixing the error or warning.

[0034] In one embodiment, the process tester includes a dialog tester;

[0035] The dialogue tester is used to simulate human-machine conversations and detect the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations; the content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay;

[0036] Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process;

[0037] The test starting point indicates which node in the current process should be selected for testing through the dialogue tester;

[0038] The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node.

[0039] Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain;

[0040] Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling;

[0041] The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call;

[0042] The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester;

[0043] Dialogue replay means observing the voice robot's intention judgment in the current node through the dialogue tester, and observing the voice robot's judgment results for each intention through dialogue replay until the customer intention in the node is found that is the same as the recognition result and the next step is executed.

[0044] In one embodiment, the comprehension tester comprises a rapid comprehension tester;

[0045] The quick understanding tester quickly tests the voice robot's understanding of the content of the customer's conversation response by inputting test text; the quick understanding tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the number of correct understandings and the accuracy rate of the voice robot's intention understanding after the quick test is completed.

[0046] In one embodiment, the comprehension tester comprises a test set tester;

[0047] The test set tester is used to test the intention understanding of the voice robot in the human-computer dialogue process through the test set. In the test set test, the test set tester uploads a reusable test set before the test. The test set includes one or more test cases. Whenever the voice robot is optimized, the test set tester references the uploaded test set to test the understanding ability of the voice robot in the human-computer dialogue process. The test set tester displays and compares the expected and actual results of intention recognition for each test case, and calculates the correct understanding number and understanding accuracy rate of intention understanding.

[0048] After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage.

[0049] Implementing the embodiments of the present invention will have the following beneficial effects:

[0050] Through the present invention, the testing process of the voice robot becomes clear and visible, and has high reusability. Problems that arise during the robot's construction and intention understanding process can be discovered and located in a timely manner, thereby greatly improving the testing efficiency of the voice robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] in:

[0053] Figure 1 It is a flowchart of the human-computer dialogue process testing method in the present invention;

[0054] Figure 2 A schematic diagram of a human-computer dialogue process testing system in the present invention; DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] In order to solve the problems existing in the prior art, the present invention proposes a human-computer dialogue process testing method for a human-computer dialogue scenario, comprising:

[0057] A voice robot is created through a robot management device, and after creation, the voice robot is configured to complete the construction of the voice robot; the robot management device sends the information of the completed voice robot to a test device connected thereto, and the test device tests the human-machine dialogue process of the voice robot;

[0058] The content of configuring the voice robot includes configuring the dialogue flow chart, understanding model, understanding intention, and common questions of the voice robot; the dialogue flow chart is a logic diagram for the voice robot to jump according to the customer's dialogue response, and the dialogue flow chart includes one or more nodes;

[0059] The test device tests the human-machine dialogue process of the voice robot and generates corresponding test results, and the test device sends the test results and data to a recording device connected thereto;

[0060] Among them, the test of human-computer dialogue process includes process test and understanding test;

[0061] Process testing includes dialogue flow chart verification and dialogue testing; comprehension testing includes quick comprehension testing and test set testing;

[0062] The testing device includes a flow tester and a comprehension tester;

[0063] Use the process tester to check whether there are any errors in the syntax of the dialogue flow chart, and detect the process execution, node jump, and variable assignment of the voice robot in the human-computer dialogue process;

[0064] Use the comprehension tester to detect whether the voice robot accurately understands the customer's intentions during the human-computer dialogue process;

[0065] The process tester includes a dialog flow chart verifier, which verifies the configuration and syntax of the dialog flow chart. In the dialog flow chart verification, the dialog flow chart is verified to see whether it contains configuration errors or syntax errors to ensure that the dialog flow is not interrupted. After verification, the dialog flow chart verifier obtains the name, location, description, and repair suggestion of the error or warning in the dialog flow chart.

[0066] Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result;

[0067] The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces; the description of an error or warning refers to a detailed description of the error or warning; suggestions for errors or warnings refer to suggestions for fixing errors or warnings;

[0068] The process tester includes a dialogue tester, which simulates human-machine conversations and detects the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations. The content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay.

[0069] Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process;

[0070] The test starting point indicates which node in the current process should be selected for testing through the dialogue tester;

[0071] The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node.

[0072] Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain;

[0073] Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling;

[0074] The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call;

[0075] The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester;

[0076] Dialogue replay means observing the intention judgment of the voice robot in the current node through the dialogue tester, and observing the judgment results of the voice robot for each intention through dialogue replay until the customer intention in the node that is the same as the recognition result is found and the next operation is performed;

[0077] Use the comprehension tester to detect whether the voice robot accurately understands the customer's intentions during the human-computer dialogue process;

[0078] The understanding tester includes a quick understanding tester, which quickly tests the voice robot's understanding of the content of the customer's dialogue response by inputting a test text; the quick understanding tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the number of correct understandings and the accuracy rate of the voice robot's intention understanding after the quick test is completed;

[0079] The understanding tester includes a test set tester, which uses a test set to test the voice robot's understanding of the intention in the human-computer dialogue process;

[0080] In test set testing, the test set tester uploads a reusable test set before testing. The test set includes one or more test cases.

[0081] Whenever the voice robot is optimized, the test set tester references the uploaded test set to test the voice robot's understanding ability in the human-machine dialogue process; the test set tester displays and compares the expected and actual results of intent recognition for each test case, and calculates the number of correct understandings and the accuracy rate of intent understanding;

[0082] After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage;

[0083] The testing device sends the test result feedback to the conversational robot management device, and the conversational robot management device optimizes the configuration of the conversational robot according to the test result, thereby forming a test closed loop;

[0084] The recording device records the test results and data, and saves the correct number and accuracy rate of the conversational robot's understanding of the customer's conversation response content calculated after each test is completed in the test results.

[0085] In addition, the present invention also proposes a human-machine dialogue process testing system, comprising a robot management device, a testing device, and a recording device; the robot management device is connected to the testing device; the testing device is connected to the data recording device;

[0086] The robot management device is used to create a voice robot, configure the voice robot after creation, and complete the construction of the voice robot; the robot management device sends the information of the completed voice robot to the testing device, and the testing device tests the human-machine dialogue process of the voice robot;

[0087] The content of configuring the voice robot includes configuring the dialogue flow chart, understanding model, understanding intention, and common questions of the voice robot; the dialogue flow chart is a logic diagram for the voice robot to jump according to the customer's dialogue response, and the dialogue flow chart includes one or more nodes;

[0088] The testing device is used to test the human-machine dialogue process of the voice robot and generate corresponding test results, and send the test results and data to the recording device;

[0089] Among them, the test of human-computer dialogue process includes process test and understanding test; process test includes dialogue flow chart verification and dialogue test; understanding test includes quick understanding test and test set test;

[0090] The testing device includes a flow tester and a comprehension tester;

[0091] The process tester is used to check the syntax of the dialogue flow chart, detect the process execution, node jump, and variable assignment of the voice robot in the human-computer dialogue process;

[0092] Wherein, the process tester includes a dialogue flow chart verifier;

[0093] The dialog flow chart verifier is used to verify the configuration and syntax of the dialog flow chart. In the dialog flow chart verification, the dialog flow chart is verified to see whether it contains configuration errors or syntax errors to ensure that the dialog flow is not interrupted. After verification, the dialog flow chart verifier obtains the name, location, description, and repair suggestion of the error or warning in the dialog flow chart.

[0094] Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result;

[0095] The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces; the description of an error or warning refers to a detailed description of the error or warning; suggestions for errors or warnings refer to suggestions for fixing errors or warnings;

[0096] Wherein, the process tester includes a dialogue tester;

[0097] The dialogue tester is used to simulate human-machine conversations and detect the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations; the content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay;

[0098] Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process;

[0099] The test starting point indicates which node in the current process should be selected for testing through the dialogue tester;

[0100] The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node.

[0101] Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain;

[0102] Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling;

[0103] The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call;

[0104] The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester;

[0105] Dialogue replay means observing the intention judgment of the voice robot in the current node through the dialogue tester, and observing the judgment results of the voice robot for each intention through dialogue replay until the customer intention in the node that is the same as the recognition result is found and the next operation is performed;

[0106] The understanding tester is used to detect whether the voice robot accurately understands the customer's intention in the human-computer dialogue process;

[0107] Wherein, the comprehension tester comprises a rapid comprehension tester;

[0108] The quick understanding tester quickly tests the voice robot's understanding of the content of the customer's dialogue response by inputting test text; the quick tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the number of correct understandings and the accuracy rate of the voice robot's intention understanding after the quick test is completed;

[0109] Wherein, the comprehension tester comprises a test set tester;

[0110] The test set tester is used to test the intention understanding of the voice robot in the human-computer dialogue process through the test set. In the test set test, the test set tester uploads a reusable test set before the test. The test set includes one or more test cases. Whenever the voice robot is optimized, the uploaded test set is referenced to test the voice robot's understanding ability in the human-computer dialogue process.

[0111] After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage;

[0112] The test set tester displays and compares the expected and actual results of intent recognition for each test case, and calculates the number of correct understandings and the correct understanding rate of intent understanding;

[0113] The testing device sends the test result feedback to the robot management device, and the robot management device optimizes the voice robot according to the test result, thereby forming a test closed loop;

[0114] Among them, the test recording device is used to record the test results and data, and the test results save the correct number and accuracy rate of the voice robot's understanding of the customer's conversation response content calculated after each test is completed.

[0115] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A human-computer dialogue process testing system, It is characterized in that It includes a robot management device, a testing device, and a recording device; the robot management device is connected to the testing device; the testing device is connected to the recording device; The robot management device is used to create a voice robot, configure the voice robot after creation, and complete the construction of the voice robot; the robot management device sends the information of the completed voice robot to the testing device, and the testing device tests the human-computer dialogue process of the voice robot; The content of configuring the voice robot includes configuring the dialogue flow chart, understanding model, understanding intention, and common questions of the voice robot; the dialogue flow chart is a logic diagram for the voice robot to jump according to the customer's dialogue response, and the dialogue flow chart includes one or more nodes; The testing device is used to test the human-computer dialogue process of the voice robot and generate corresponding test results, and send the test results and data to the recording device; the testing device includes a process tester and an understanding tester; the process tester is used to check whether there are errors in the syntax of the dialogue flow chart, and detect the process execution, node jump, and variable assignment of the voice robot in the human-computer dialogue process; the understanding tester is used to detect whether the voice robot accurately understands the customer's intention in the human-computer dialogue process; The testing device sends the test result feedback to the conversational robot management device, and the conversational robot management device optimizes the configuration of the conversational robot according to the test result, thereby forming a test closed loop; Among them, the recording device is used to record the test results and data, and the test results save the correct number and accuracy rate of the conversational robot's understanding of the customer's conversation response content calculated after each test is completed.

2. The human-computer dialogue process testing system according to claim 1 ， It is characterized in that Wherein, the process tester includes a dialogue flow chart verifier; The dialog flow chart verifier is used to verify the configuration and syntax of the dialog flow chart. In the dialog flow chart verification, the dialog flow chart is verified to see whether it contains configuration errors or syntax errors to ensure that the dialog flow is not interrupted. After verification, the dialog flow chart verifier obtains the name, location, description, and repair suggestion of the error or warning in the dialog flow chart. Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result; The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces. The description of an error or warning refers to a detailed description of the error or warning. The suggestion about an error or warning refers to a suggestion for fixing the error or warning.

3. The human-computer dialogue process testing system according to claim 1 ， It is characterized in that Wherein, the process tester includes a dialogue tester; The dialogue tester is used to simulate human-machine conversations and detect the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations; the content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay; Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process; The test starting point indicates which node in the current process should be selected for testing through the dialogue tester; The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node. Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain; Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling; The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call; The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester; Dialogue replay means observing the voice robot's intention judgment in the current node through the dialogue tester, and observing the voice robot's judgment results for each intention through dialogue replay until the customer intention in the node is found that is the same as the recognition result and the next step is executed.

4. The human-computer dialogue process testing system according to claim 1, characterized in that: in, The comprehension tester includes a rapid comprehension tester; The quick understanding tester quickly tests the voice robot's understanding of the content of the customer's conversation response by inputting test text; the quick understanding tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the number of correct understandings and the accuracy rate of the voice robot's intention understanding after the quick test is completed.

5. The human-computer dialogue process testing system according to claim 1 ， It is characterized in that in, The comprehension tester includes a test set tester; The test set tester is used to test the intention understanding of the voice robot in the human-computer dialogue process through the test set. In the test set test, the test set tester uploads a reusable test set before the test. The test set includes one or more test cases. Whenever the voice robot is optimized, the test set tester references the uploaded test set to test the understanding ability of the voice robot in the human-computer dialogue process. The test set tester displays and compares the expected and actual results of intention recognition for each test case, and calculates the correct understanding number and understanding accuracy rate of intention understanding. After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage.

6. A human-computer dialogue process testing method, It is characterized in that include: Create a voice robot through the robot management device, configure the voice robot after creation, and complete the construction of the voice robot; The robot management device sends the information of the built voice robot to the test device connected to it, and the test device tests the human-machine dialogue process of the voice robot; The content of configuring the voice robot includes configuring the dialogue flow chart, understanding model, understanding intention, and common questions of the voice robot; the dialogue flow chart is a logic diagram for the voice robot to jump according to the customer's dialogue response, and the dialogue flow chart includes one or more nodes; The human-machine dialogue process of the voice robot is tested by a testing device and a corresponding test result is generated. The testing device sends the test result and data to a recording device connected thereto; the testing device includes a process tester and an understanding tester; the process tester is used to check whether there are errors in the syntax of the dialogue flow chart, and to detect the process execution, node jump, and variable assignment of the voice robot in the human-machine dialogue process; the understanding tester is used to detect whether the voice robot accurately understands the customer's intention in the human-machine dialogue process; The testing device sends the test result feedback to the conversational robot management device, and the conversational robot management device optimizes the configuration of the conversational robot according to the test result, thereby forming a test closed loop; The recording device records the test results and data, and saves the correct number and accuracy rate of the conversational robot's understanding of the customer's conversation response content calculated after each test is completed in the test results.

7. The human-computer dialogue process testing method according to claim 6, It is characterized in that The process tester includes a dialog flow chart verifier, which verifies the configuration and syntax of the dialog flow chart. In the dialog flow chart verification, it is verified whether the dialog flow chart contains configuration errors or syntax errors to ensure that the dialog flow will not be interrupted. The dialog flow chart verifier obtains the name, location, description, and repair suggestions of the error or warning in the dialog flow chart after verification; Among them, the error of the dialogue flow chart refers to the problem that cannot be continued during the configuration of the dialogue flow chart, and the warning refers to the problem that does not affect the execution of the dialogue flow but affects the execution result; The location of an error or warning refers to the location where the error or warning occurs, including nodes, configuration items, and interfaces. The description of an error or warning refers to a detailed description of the error or warning. The suggestion about an error or warning refers to a suggestion for fixing the error or warning.

8. The human-computer dialogue process testing method according to claim 6, It is characterized in that The process tester includes a dialogue tester, which simulates human-machine conversations and detects the process execution, node jumps, and variable assignments of the voice robot during the simulation of human-machine conversations. The content of the dialogue test includes test process selection, test starting point, node jump status, intention feedback, logic execution, initial parameters, fact variable values, and dialogue replay. Test process selection means selecting which process to execute for testing through the dialogue tester. You can choose the main process or its sub-process; The test starting point indicates which node in the current process should be selected for testing through the dialogue tester; The node jump situation means that the conversation tester simulates the customer's conversation response through text, and the voice robot executes the node jump in the process according to the customer's conversation response content. After observing which node the voice robot enters after the customer's conversation response, the current node ends after completing the action in the node. Intent feedback means that the conversation tester checks the customer intent output by the voice robot after recognizing the customer's conversation response content at the current node; customer intent includes affirmative, negative, and uncertain; Logical execution means that the dialogue tester checks the next action taken by the voice robot after judging the intention; logical execution includes jumping out of branches, triggering FAQs, and triggering exception handling; The initial parameters are the voice robot input parameters uploaded by the conversation tester when simulating a call; The fact variable value is the change information of the parameter value of the voice robot before and after the node jump observed by the dialogue tester; Dialogue replay means observing the voice robot's intention judgment in the current node through the dialogue tester, and observing the voice robot's judgment results for each intention through dialogue replay until the customer intention in the node is found that is the same as the recognition result and the next step is executed.

9. The human-computer dialogue process testing method according to claim 6, It is characterized in that The understanding tester includes a quick understanding tester, which quickly tests the voice robot's understanding of the content of the customer's conversation response by inputting test text; the quick understanding tester temporarily tests the voice robot's intention understanding performance after the voice robot is created and configured or optimized, and calculates the voice robot's correct understanding number and understanding accuracy rate regarding the intention understanding after the quick test is completed.

10. The human-computer dialogue process testing method according to claim 6, It is characterized in that The understanding tester includes a test set tester, which uses a test set to test the voice robot's understanding of the intention in the human-computer dialogue process; In test set testing, the test set tester uploads a reusable test set before testing. The test set includes one or more test cases. Whenever the voice robot is optimized, the test set tester references the uploaded test set to test the voice robot's understanding ability in the human-machine dialogue process; the test set tester displays and compares the expected and actual results of intent recognition for each test case, and calculates the number of correct understandings and the accuracy rate of intent understanding; After the test set tester completes the test set test, the test results of the test set and the corresponding test data are sent to the test recording device for storage.

Images