Method for training mediator, electronic equipment and storage medium

Abstract

The invention provides a method for training a mediator, electronic equipment and a storage medium. The method comprises the steps that a target exercise script is configured, the script comprises party statements and mediator statements which alternately appear, and for any party statement, the next mediator statement located in the party statement is a target mediator statement matched with the party statement; at least performing voice conversion on sentences of the parties in the target exercise script to obtain voice of the parties; executing a man-machine drilling operation; outputting a party voice corresponding to at least one party statement in the target drilling script; after the voice of the party is output each time, receiving the voice of the user; calculating a statement score of the user voice obtained each time based on the user voice received after the user voice is output each time and a target mediator statement corresponding to the user voice in the target exercise script; and determining a verbal skill score based on all statement scores. According to the method, the training effect of the mediator can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of dispute mediation, and more specifically to a method, electronic device and storage medium for training mediators. Background Art

[0002] As the central government continues to implement the transformation of the conflict resolution model to "mediation as the center, litigation as the ultimate guarantee" to improve the legal level of conflict prevention and resolution, the number of court mediation cases has also increased sharply. In the process of financial dispute mediation, the telephone communication between the mediator and the parties is the core link of mediation communication. The mediator needs to communicate with the parties a lot by phone to negotiate the mediation plan. In this process, the effectiveness of mediation work depends largely on the professionalism of the mediator. In order to improve the professionalism of mediators, professional training is usually provided to them. However, the current training mainly relies on traditional face-to-face teaching methods, which are less effective, limit the improvement of mediators' professional capabilities, and affect the effective operation of the financial dispute mediation mechanism.

[0003] In view of this, the present invention is proposed. Summary of the invention

[0004] The present invention is proposed in view of the above problems. According to one aspect of the present invention, a method for training mediators is provided, comprising: configuring a target rehearsal script, wherein the target rehearsal script includes party statements and mediator statements, and the party statements and mediator statements in the target rehearsal script appear alternately, and for any party statement in the target rehearsal script, the next mediator statement located after the party statement is a target mediator statement matching the party statement; performing voice conversion on at least the party statements in the target rehearsal script to obtain party voices; performing the following human-computer rehearsal operations: outputting party voices corresponding to at least one party statement in the target rehearsal script; receiving user voices after each output of the party voices; calculating the sentence scores of the user voices obtained each time based on the user voices received after each output of the party voices and the target mediator statements corresponding to the party voices in the target rehearsal script; determining the speech score of the current user based on all sentence scores obtained in the human-computer rehearsal operation.

[0005] Exemplarily, before performing speech conversion on at least the party sentences in the target rehearsal script, the method further includes: for each party sentence in the target rehearsal script, rewriting the party sentence using a large language model to obtain a sentence set corresponding to the party sentence, wherein the sentence set includes the party sentence and at least one candidate sentence with the same meaning as the party sentence; performing speech conversion on at least the party sentences in the target rehearsal script includes: for each party sentence in the target rehearsal script, converting all sentences in the sentence set corresponding to the party sentence into speech to obtain party speech corresponding to each sentence in the sentence set one by one; wherein, for any party sentence in the target rehearsal script, when outputting the party speech corresponding to the party sentence, randomly selecting a sentence from the sentence set where the party sentence is located, and outputting the speech obtained by converting the selected sentence.

[0006] Exemplarily, the outputting of the party voice corresponding to at least one party sentence in the target rehearsal script includes: outputting the party voice corresponding to each party sentence in the target rehearsal script in sequence; or, randomly outputting the party voice corresponding to any one or several party sentences in the target rehearsal script.

[0007] Exemplarily, configuring the target drill script includes: creating the target drill script in response to user input information; or, determining the target drill script from multiple preset drill scripts based on a first selection instruction input by the user; or, selecting a target communication scenario from multiple preset communication scenarios based on a second selection instruction input by the user; and determining a target drill script corresponding to the target communication scenario.

[0008] Exemplarily, for each preset communication scenario among the multiple preset communication scenarios, the preset communication scenario corresponds to at least one preset drill script; determining the target drill script corresponding to the target communication scenario includes: randomly selecting a preset drill script from at least one preset drill script corresponding to the target communication scenario as the target drill script; or, according to a third selection instruction input by the user, determining the target drill script from at least one preset drill script corresponding to the target communication scenario.

[0009] Exemplarily, the sentence score of the user voice obtained each time is calculated based on the user voice received after each output of the party's voice and the target mediator sentence corresponding to the party's voice in the rehearsal script, including: for any user voice received after the party's voice is output, converting the user voice into text to obtain a user sentence; calculating the semantic similarity between the user sentence and the target mediator sentence corresponding to the party's voice in the rehearsal script to obtain the sentence score of the user voice.

[0010] Exemplarily, determining the current user's speech score based on all sentence scores obtained in the human-computer rehearsal operation includes: calculating the average of all sentence scores obtained in the human-computer rehearsal operation to determine the speech score; or determining the median of all sentence scores obtained in the human-computer rehearsal operation as the speech score.

[0011] Exemplarily, after determining the speech score of the current user, the method further includes: generating a training evaluation result according to the speech score; or generating a training evaluation result according to a chat record, wherein the chat record includes the output voice of the party concerned and the received user voice.

[0012] According to another aspect of the present invention, there is provided an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and the processor is used to execute the computer program to implement the above method.

[0013] According to another aspect of the present invention, a computer-readable storage medium is provided, storing a computer program / instruction, and the computer program / instruction implements the above method when executed by a processor.

[0014] In the above technical solution, the voice of the parties can be output according to the predetermined target exercise script, and the sentence score and speech score of the current user can be measured according to the received user voice and the mediator's sentence in the target exercise script. On the one hand, this method of training users by human-computer training can break the limitations of time and place, so that users can be trained at any time and any place, which helps to improve the efficiency and coverage of training. In particular, this training method can be carried out in small and medium-sized cities and rural areas, which helps to further improve the coverage and accessibility of mediation services. On the other hand, this human-computer training method can simulate real financial dispute mediation scenarios, allowing users to perform actual operations in a simulated environment, enhance the interactivity, practicality, practicability and effectiveness of training, and thus further improve the training effect. On the other hand, this method can provide systematic training of the same quality for different users, which helps to improve the professional background and industry experience of mediators, optimize the structure of the mediator team, and improve the overall mediation quality. On the other hand, this training method is more efficient and less costly than the traditional manual teaching method, which helps to optimize the allocation of mediation resources. In summary, the above-mentioned technical solutions can help enhance the professional skills of financial dispute mediators, improve mediation efficiency and quality, optimize the allocation of mediation resources, and reduce mediation costs, thereby helping to achieve more efficient and professional financial dispute mediation results.

[0015] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other purposes, features and advantages of the present invention will become more apparent by describing the embodiments of the present invention in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 A schematic flow chart showing a method for training mediators according to an embodiment of the present invention; Figure 2 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the present invention more obvious, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described in the present invention, all other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present invention.

[0019] As mentioned above, the current training of mediators mainly relies on traditional face-to-face teaching. This method has the following shortcomings: 1. Uneven distribution of training resources: Currently, the training of financial dispute mediators is mostly organized by courts or relevant financial management agencies. Such training resources are often concentrated in large cities and developed areas, while mediators in small and medium-sized cities and rural areas may find it difficult to obtain training resources of the same quality; 2. Single training content and format: Existing training content is mostly focused on legal knowledge, mediation skills, etc., lacking the cultivation of soft skills such as mediators' communication skills and psychological tactics. In addition, the training format is relatively single, mostly lecture-style teaching, lacking opportunities for interaction and practical operation; 3. Insufficient professional training: Financial disputes are highly professional and involve complex financial products and legal knowledge. The existing mediator team structure is not reasonable, lacks professional background and industry experience, and some mediators lack the necessary legal knowledge and communication skills; 4. The quality of the mediators is uneven: The quality of the mediators is uneven. Some mediators lack the necessary legal knowledge and communication skills, which directly affects the efficiency and effectiveness of the mediation work; 5. Insufficient technological empowerment: The technical and information levels of financial dispute mediation are still relatively low, and modern information technology has not been fully utilized for training and mediation.

[0020] In summary, the existing financial dispute mediator training technology has problems such as uneven resource distribution, single content and form, insufficient professional training, uneven quality of the mediator team, and insufficient technological empowerment. These problems limit the improvement of the professional ability of mediators and affect the effective operation of the financial dispute mediation mechanism. In view of this, the present invention provides a method, electronic device and storage medium for training mediators. The method can train mediators in the form of human-computer duels. This training method is efficient and low-cost, and can effectively improve the professional ability of mediators and promote the effective operation of the financial dispute mediation mechanism. The method, electronic device and storage medium are described in detail below.

[0021] According to one aspect of an embodiment of the present invention, a method for training a mediator is provided. Figure 1FIG. 1 is a schematic flow chart of a method for training mediators according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps S110, S120, S130 and S140.

[0022] In step S110, a target exercise script is configured, wherein the target exercise script includes statements of the parties and statements of the mediator, and the statements of the parties and the statements of the mediator in the target exercise script appear alternately. For any statement of the parties in the target exercise script, the statement of the mediator next to the statement of the parties is a target mediator statement that matches the statement of the parties.

[0023] The target drill script is the drill script used for the current user (ie, mediator) to conduct the current human-machine training. The target drill script can be created directly by the user, or can be selected by the user from multiple preset drill scripts, which will not be described in detail.

[0024] In the scheme of this example, the statements of the parties and the statements of the mediator appear alternately in the target exercise script. That is, in the target exercise script, the statements of the parties and the statements of the mediator appear alternately in the order of context. For ease of description, the target exercise script in a specific embodiment is used for explanation below. In this embodiment, the target exercise script is as follows: "Mediator: Hello, is this XXX? Party: Yes, what’s the matter? Mediator: I am a mediator from XX Court, with work number XXX. I am here to mediate on behalf of XXX. I noticed that your Huabei payment has been overdue for some time. Is this true? Person involved: Yes, but I have no choice. I was laid off, I have no job, and no money.

[0025] Mediator: I understand your current situation. You have been overdue for a long time. Can you ask relatives or friends for help? " The party's statement is represented by A, and the mediator's statement is represented by Q. In the above exercise script, the entire dialogue is structured as Q0, A1, Q1, A2, Q2, and the mediator's statements and the party's statements appear alternately. And the target mediator's statement that matches each party's statement is the next mediator's statement of the party. For example, the party's statement A1 "Yes, what's the matter?" corresponds to the target mediator's statement Q1 "I am a mediator of XX Court, work number XXX, entrusted by XXX to do mediation. I see that your Huabei has been overdue for a while, is it true?"

[0026] It is understood that the specific target exercise scripts shown above are only examples. The actual target exercise scripts can be configured or selected as needed and will not be described in detail.

[0027] In step S120, at least the party's sentences in the target practice script are converted into voice to obtain the party's voice.

[0028] After the target exercise script is configured, at least the party's sentences in the target exercise script can be converted into voice. The voice conversion method can be implemented by any existing or future developed voice-to-text model, which will not be described in detail.

[0029] In step S130, a human-machine rehearsal operation is performed, which may specifically include the following steps S131, S132, and S133.

[0030] In step S131, the party's speech corresponding to at least one party's sentence in the target practice script is output.

[0031] In step S132, after each output of the party's voice, the user's voice is received.

[0032] In step S133, based on the user voice received after each output of the party's voice and the target mediator's sentence corresponding to the party's voice in the target rehearsal script, the sentence score of the user voice obtained each time is calculated.

[0033] In the solution of this example, the system executing the method for training mediators can output the party's voice corresponding to the party's sentence, and after each output of the party's voice, receive the user's voice through the user's microphone. The received user voice at this time is the user's response to the played party's voice. In a specific embodiment, when the role type is "party", the system can output a party's voice. When the role type is "mediator", the system can receive the user's voice from the user's microphone.

[0034] As described above, the target mediator sentence matched by each party's sentence is the next mediator sentence of the party's sentence. That is, the target mediator sentence corresponding to the party's sentence in the target rehearsal script can be regarded as a reply template for responding to the party's sentence. In the scheme of this example, the target mediator sentence corresponding to the party's voice is the mediator sentence matched by the party's sentence corresponding to the party's voice in the target rehearsal script, and the user voice received after each output of the party's voice is the user's actual reply. After obtaining the user's voice, the user's voice can be compared with the reply template, and the comparison result can be used to determine the sentence score of the user's voice.

[0035] In the scheme of this example, a user voice is received each time after the party's voice is output. Step S133 can be executed after all the party's voices (referring to the party's voices that need to be output for this human-machine exercise operation) are output and all user voices are received. For example, if this human-machine exercise operation needs to output three party voices, in this embodiment, step S133 can be executed after the third party's voice is output and the corresponding user voice is received, so as to determine the sentence score of each received user voice. Of course, step S133 can also be executed each time a user voice is received. This execution method helps to improve the overall training efficiency.

[0036] In step S140, the speech score of the current user is determined based on the scores of all sentences obtained in the human-computer practice operation.

[0037] After the human-machine rehearsal operation is completed, the sentence scores corresponding to all the user voices received during the human-machine rehearsal operation can be obtained. These sentence scores can be used to characterize the professionalism of each user's reply. In the scheme of this example, the speech score of the current user can be determined based on all sentence scores. For example, the speech score can be determined based on the average, median, maximum, minimum, etc. of all sentence scores.

[0038] The above technical solution can output the voice of the parties according to the predetermined target exercise script, and measure the sentence score and speech score of the current user according to the received user voice and the mediator's sentence in the target exercise script. On the one hand, this method of training users by human-computer training can break the limitations of time and place, so that users can be trained at any time and any place, which helps to improve the efficiency and coverage of training. In particular, this training method can be carried out in small and medium-sized cities and rural areas, which helps to further improve the coverage and accessibility of mediation services. On the other hand, this human-computer training method can simulate real financial dispute mediation scenarios, allowing users to perform actual operations in a simulated environment, enhance the interactivity, practicality, practicability and effectiveness of training, and further improve the training effect. On the other hand, this method can provide systematic training of the same quality for different users, which helps to improve the professional background and industry experience of mediators, optimize the structure of the mediator team, and improve the overall mediation quality. On the other hand, this training method is more efficient and less costly than the traditional manual teaching method, which helps to optimize the allocation of mediation resources. In summary, the above-mentioned technical solutions can help enhance the professional skills of financial dispute mediators, improve mediation efficiency and quality, optimize the allocation of mediation resources, and reduce mediation costs, thereby helping to achieve more efficient and professional financial dispute mediation results.

[0039] Exemplarily, in step S120, before performing speech conversion on at least the party statements in the target rehearsal script, the method may further include the following steps: for each party statement in the target rehearsal script, rewrite the party statement using a large language model to obtain a sentence set corresponding to the party statement, the sentence set including the party statement and at least one candidate sentence having the same meaning as the party statement.

[0040] Step S120, performing voice conversion on at least the party's sentences in the target rehearsal script, including: for each party's sentence in the target rehearsal script, converting all sentences in the sentence set corresponding to the party's sentence into voice, so as to obtain party voices corresponding to each sentence in the sentence set; wherein, for any party's sentence in the target rehearsal script, when outputting the party's voice corresponding to the party's sentence, randomly select a sentence from the sentence set where the party's sentence is located, and output the voice obtained by converting the selected sentence.

[0041] In the solution of this example, after obtaining the target exercise script, each party's sentence can be expanded using a large language model to obtain multiple sentences with the same meaning as the party's sentence but with different expressions. The large language model can use any open source or closed source large language model that is currently available or will be developed in the future. For example, the large language model can include but is not limited to GPT-3.5, Tongyi Qianwen, Wenxin Yiyan, LLAMA, ChatGLM, etc.

[0042] Optionally, rewriting the party's statement using the large language model may include the following steps: adding the party's statement to a prompt word template to obtain a target prompt word; inputting the target prompt word into the large language model to obtain at least one candidate statement with the same meaning as the party's statement. The at least one candidate statement with the same meaning as the party's statement and the party's statement may constitute a statement set corresponding to the party's statement.

[0043] In a specific embodiment, the prompt word template may be: "The following sentence is a sentence in the call record. Please give me 3 sentences with the same meaning as this sentence. The form should be more colloquial, and no extra meaning should be extended. Separate by 1.2.3.: {input}". The input is used to add the party's sentence. According to the prompt word template, three candidate sentences can be generated. It can be understood that the prompt word template is only an example and not a limitation of the present invention. The specific content of the prompt word template and the number of candidate sentences can be set as needed and will not be repeated.

[0044] For ease of description, the solution of this example is described by taking the target rehearsal script in the above embodiment as an example. For the statement A2 of the party, the large language model can be first used for amplification to obtain 3 candidate statements with the same semantics as A2. At this time, the statement set corresponding to A2 can be: 1. Yeah, but I'm also helpless. I got fired and now I have no job and no money in my pocket.

[0045] 2. Well, I have no choice either. I lost my job, I'm short of money, and I don't have a single cent.

[0046] 3. That's true, but I'm just anxious. I was laid off and now I don't have a job and I've run out of money.

[0047] 4. Yes, but I have no way out either. I was laid off, have no job, and have no money.

[0048] After obtaining the statement set corresponding to A2, all the statements in the statement set can be converted into voices. In this case, it can be considered that the voices corresponding to all the statements in the statement set are the voices of the parties corresponding to A2. When it is necessary to output A2, a statement can be randomly selected from the statement set corresponding to A2, and the voice obtained by converting the selected statement can be output. For example, the voice obtained by converting the 4th statement in the statement set can be output.

[0049] In the above technical solution, by expanding each statement of the party and randomly selecting any statement in the statement set corresponding to the party statement when playing the party statement, this method can expand the expression ways of the party statement, increase the diversity and unpredictability during the human-machine practice, and make the human-machine practice closer to the dialogue scenario in reality. Training the mediator in this way helps to improve the mediator's ability to handle various situations and enhance their communication skills and psychological tactics.

[0050] Exemplarily, step S131, outputting the voice of the party corresponding to at least one statement of the party in the target rehearsal script includes: sequentially outputting the voices of the parties corresponding to each statement in the target rehearsal script; or, randomly outputting the voices of the parties corresponding to any one or several statements in the target rehearsal script.

[0051] In some embodiments, the voices of the parties corresponding to the statements of the parties in the target rehearsal script can be sequentially output according to the text order of the statements of the parties. Still taking the target rehearsal script in the above embodiment as an example for description. In this embodiment, the voice corresponding to A1 and the voice corresponding to A2 can be sequentially output. This sequential output method can completely simulate the communication scenario corresponding to the target rehearsal script, thereby ensuring the continuity of the communication and improving the training effect.

[0052] In other embodiments, the party's voice corresponding to any one or several party's sentences in the target drill script can also be randomly output. For example, only A1 or A2 can be output, or A2 can be output first, and then A1. This random output method can train the user's ability to respond to sudden questions, improve their psychological endurance, and avoid users relying solely on mechanical memory to conduct human-computer training, thereby ensuring the user's training effect. In addition, this method does not require the output of the voice corresponding to all the party's sentences in the target drill script. When there are too many party sentences in the target drill script, only the party's voice corresponding to one or several party sentences can be randomly output. This can facilitate the user to control the training time and improve the flexibility of the training time.

[0053] Exemplarily, configuring a target drill script includes: creating a target drill script in response to user input information; or, determining a target drill script from multiple preset drill scripts based on a first selection instruction input by the user; or, selecting a target communication scenario from multiple preset communication scenarios based on a second selection instruction input by the user; and determining a target drill script corresponding to the target communication scenario.

[0054] Optionally, configuring the target drill script includes: creating a target drill script in response to user input information. In the solution of this embodiment, the user can first create a script currently used for the drill. This method can facilitate the user to create a drill script according to the typical communication scenario required, thereby improving the mediation ability of training in the typical communication scenario.

[0055] Optionally, configuring the target drill script includes: determining the target drill script from a plurality of preset drill scripts according to a first selection instruction input by the user. In this embodiment, a plurality of preset drill scripts may be preconfigured. When human-machine training is required, the user may select from a plurality of preset drill scripts. This direct selection method does not require the user to create a script separately, which can improve training efficiency.

[0056] Optionally, configuring a target drill script includes: selecting a target communication scenario from a plurality of preset communication scenarios according to a second selection instruction input by the user; and determining a target drill script corresponding to the target communication scenario. In this embodiment, drill scripts for different preset communication scenarios can be configured separately, and the number of drill scripts for each preset communication scenario can be one or more. When human-computer sparring training is required, the user can select the required preset communication scenario, and then determine the target drill script corresponding to the target communication scenario. Preset communication scenarios include, but are not limited to, communication scenarios in which the parties have objections to the bank loan interest rate, credit card dispute communication scenarios, scenarios in which the parties have disputes over loan terms, and the like. This solution can facilitate users to conduct human-computer sparring training in the required communication scenarios, which helps to improve the user's mediation ability in specific communication scenarios.

[0057] Exemplarily, for each preset communication scenario among multiple preset communication scenarios, the preset communication scenario corresponds to at least one preset drill script; determining a target drill script corresponding to the target communication scenario includes: randomly selecting a preset drill script from at least one preset drill script corresponding to the target communication scenario as the target drill script; or, according to a third selection instruction input by the user, determining the target drill script from at least one preset drill script corresponding to the target communication scenario.

[0058] Optionally, for each preset communication scenario in a plurality of preset communication scenarios, the preset communication scenario corresponds to at least one preset drill script; determining the target drill script corresponding to the target communication scenario includes: randomly selecting a preset drill script from at least one preset drill script corresponding to the target communication scenario as the target drill script. In the scheme of this embodiment, after the user determines the target communication scenario, any one of the preset drill scripts included in the target communication scenario can be randomly selected as the target drill script. This random selection method can increase the diversity and unpredictability of human-computer duels, and improve the mediator's ability to deal with various party issues in the target communication scenario.

[0059] Optionally, for each preset communication scenario in a plurality of preset communication scenarios, the preset communication scenario corresponds to at least one preset drill script; determining the target drill script corresponding to the target communication scenario includes: determining the target drill script in at least one preset drill script corresponding to the target communication scenario according to a third selection instruction input by the user. In the scheme of this embodiment, after determining the target communication scenario, the user can further select a target drill script from the preset drill scripts included in the target communication scenario. Thus, the entire process of configuring the target drill script is entirely implemented by user selection. This method can facilitate users to use specific drill scripts for training.

[0060] Exemplarily, based on the user voice received after each output of the party's voice and the target mediator sentence corresponding to the party's voice in the rehearsal script, the sentence score of the user voice obtained each time is calculated, including: for any user voice received after the party's voice is output, the user voice is converted into text to obtain a user sentence; the semantic similarity between the user sentence and the target mediator sentence corresponding to the party's voice in the rehearsal script is calculated to obtain the sentence score of the user voice.

[0061] Optionally, the user's speech may be converted into text using any existing or future developed speech-to-text model, and the present invention does not impose any limitation on this.

[0062] As described above, the user voice is the actual response of the current user to the outputted party voice, and the target mediator's sentence corresponding to the party voice in the exercise script can be regarded as a response template. In the scheme of this example, the similarity between the actual response and the response template can be used as the sentence score of the user voice. For example, the Jaccard similarity coefficient or cosine similarity of the user sentence and the target mediator's sentence corresponding to the party voice in the exercise script can be calculated to determine the sentence score of the user voice.

[0063] In some embodiments, the semantic similarity between the user's statement and the target mediator's statement corresponding to the party's speech in the exercise script can be calculated by the following formula:

[0064] In the formula, S 1 and S 2 Represent the target mediator statement and user statement respectively, w i1 and w i2 Respectively represent the weight of the i-th word in the target mediator sentence and the weight of the i-th word in the user sentence. In this embodiment, the user sentence and the target mediation sentence can be segmented by the Chinese word segmentation method, and the two sentences can be divided into n words. Then, the semantic similarity between the two sentences can be obtained by calculating through the above formula.

[0065] In the above technical solution, by calculating the semantic similarity between two sentences, the sentence score of each user's voice can be evaluated more accurately, thereby providing a more accurate basis for determining the user's speech score.

[0066] Exemplarily, step S140 determines the current user's speech score based on all sentence scores obtained in the human-computer rehearsal operation, including: calculating the average of all sentence scores obtained in the human-computer rehearsal operation to determine the speech score; or, determining the median of all sentence scores obtained in the human-computer rehearsal operation as the speech score.

[0067] Optionally, step S140 includes: calculating the average of all sentence scores obtained in the human-machine rehearsal operation to determine the speech score. In the solution of this embodiment, the average of all sentence scores can be used as the speech score of the current user in the current human-machine rehearsal training process. This method of determining the speech score fully considers the sentence score of each user's reply, and the result can accurately measure the mediation professionalism of the current user.

[0068] Optionally, step S140 includes: determining the median of all sentence scores obtained in the human-machine rehearsal operation as the speech score. In the scheme of this embodiment, the median is selected as the speech score. This method can avoid the influence of abnormal sentence scores and more accurately evaluate the mediation expertise of the current user in this training process.

[0069] Exemplarily, after determining the speech score of the current user, the method further includes: generating a training evaluation result based on the speech score.

[0070] In the solution of this example, a training evaluation result can be generated based on the speech score to serve as an overall evaluation of the human-machine training. For example, different score segments can be pre-divided, and different score segments correspond to different evaluations. Then, the score segment to which the speech score belongs can be output as the training evaluation result.

[0071] The above technical solution can generate training evaluation results more accurately based on the speech score. This training evaluation result can more accurately evaluate the current user's professionalism in the current human-computer practice process, thereby providing an accurate basis for the user to continue to improve his or her professional mediation ability.

[0072] Exemplarily, after determining the speech score of the current user, the method further includes: generating a training evaluation result according to the chat record, wherein the chat record includes the output voice of the party concerned and the received user voice.

[0073] Optionally, before generating the training evaluation results based on the chat records, the method may further include the following steps: generating a chat record. In this embodiment, a chat record of the human-computer duel can be generated based on the output voice of the parties and the received user voice, and then a training evaluation result is generated based on the chat record. For example, the chat record can be input into a large language model to generate a training evaluation result. In a specific embodiment, the large language model can output the following training evaluation result: "The mediator performed well in this exercise, was able to effectively guide the parties to express their demands, and maintained the fluency of the conversation." The above technical solution generates training evaluation results based on the chat records during the human-computer training process. This method fully considers the performance of the current user in the entire human-computer training process, and the generated training evaluation results are more objective and accurate.

[0074] According to yet another aspect of the embodiments of the present invention, an electronic device is provided. Figure 2 FIG. 2 shows a schematic block diagram of an electronic device according to an embodiment of the present invention. Figure 2 As shown, the electronic device 200 includes: a processor 210 and a memory 220. The memory 220 stores a computer program, and the processor 210 is used to execute the computer program to implement the above method.

[0075] In some embodiments, the processor 210 may include a drill configuration module and a human-machine drill module. The drill configuration module is used to: configure a target drill script; perform voice conversion on at least the party's sentences in the target drill script to obtain the party's voice. The human-machine drill module is used to: perform a human-machine drill operation; and determine the current user's speech score based on all sentence scores obtained in the human-machine drill operation.

[0076] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided. The storage medium stores a computer program / instruction, and the computer program / instruction implements the above method when executed by a processor. The storage medium may include, for example, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disk read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0077] A person skilled in the art can easily understand the implementation structure, working principle and beneficial effects of the electronic device and the computer-readable storage medium by reading the above method, and will not be described in detail for the sake of brevity.

[0078] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely exemplary and are not intended to limit the scope of the present invention thereto. Various changes and modifications may be made therein by one of ordinary skill in the art without departing from the scope and spirit of the present invention. All such changes and modifications are intended to be included within the scope of the present invention as required by the appended claims.

[0079] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0080] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0081] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0082] Similarly, it should be understood that in order to streamline the present invention and help understand one or more of the various inventive aspects, in the description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, the method of the present invention should not be interpreted as reflecting the following intention: the claimed invention requires more features than the features explicitly stated in each claim. More specifically, as reflected in the corresponding claims, the inventive point is that the corresponding technical problem can be solved with less than all the features of a single disclosed embodiment. Therefore, the claims following the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself serves as a separate embodiment of the present invention.

[0083] Those skilled in the art will understand that, except for mutually exclusive features, all features disclosed in this specification (including the accompanying claims, abstract and drawings) and all processes or units of any method or device disclosed in this specification may be combined in any combination. Unless otherwise explicitly stated, each feature disclosed in this specification (including the accompanying claims, abstract and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose.

[0084] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, in the claims, any one of the claimed embodiments may be used in any combination.

[0085] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. It should be understood by those skilled in the art that a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all functions of some modules in an electronic device according to an embodiment of the present invention. The present invention may also be implemented as a device program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.

[0086] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc., does not indicate any order. These words may be interpreted as names.

[0087] The above is only a specific embodiment of the present invention or an explanation of a specific embodiment. The protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for training mediators, characterized in that include: Configuring a target exercise script, wherein the target exercise script includes statements of the parties and statements of the mediator, and the statements of the parties and the statements of the mediator in the target exercise script appear alternately, and for any statement of the parties in the target exercise script, the statement of the mediator next to the statement of the parties is a target mediator statement matching the statement of the parties; Performing voice conversion on at least the sentences of the parties in the target exercise script to obtain the voice of the parties; Perform the following man-machine drill operations: Outputting a party's speech corresponding to at least one party's sentence in the target exercise script; After outputting the party's voice each time, receiving the user's voice; Calculate the sentence score of the user voice obtained each time based on the user voice received after outputting the party's voice each time and the target mediator sentence corresponding to the party's voice in the target exercise script; Based on all sentence scores obtained in the human-machine rehearsal operation, the speech score of the current user is determined.

2. The method according to claim 1, characterized in that Before performing voice conversion on at least the party's sentences in the target exercise script, the method further includes: For each party statement in the target exercise script, rewrite the party statement using the large language model to obtain a statement set corresponding to the party statement, wherein the statement set includes the party statement and at least one candidate statement having the same meaning as the party statement; The at least performing voice conversion on the party's sentences in the target exercise script includes: For each party's statement in the target exercise script, convert all the statements in the statement set corresponding to the party's statement into speech, so as to obtain party speech corresponding to each statement in the statement set; Among them, for any party statement in the target rehearsal script, when outputting the party voice corresponding to the party statement, a sentence is randomly selected from the sentence set where the party statement is located, and the voice obtained by converting the selected sentence is output.

3. The method according to claim 1, characterized in that The outputting of the party's speech corresponding to at least one party's sentence in the target exercise script includes: Outputting the party's speech corresponding to each party's sentence in the target exercise script in sequence; or, Randomly output the party's voice corresponding to any one or several party's sentences in the target exercise script.

4. The method according to any one of claims 1 to 3, characterized in that: The configuration target exercise script includes: In response to user input information, creating the target drill script; or, Determining the target exercise script from a plurality of preset exercise scripts according to a first selection instruction input by a user; or, According to a second selection instruction input by the user, a target communication scene is selected from a plurality of preset communication scenes; Determine the target drill script that corresponds to this target communication scenario.

5. The method according to claim 4, characterized in that For each preset communication scenario in the plurality of preset communication scenarios, the preset communication scenario corresponds to at least one preset drill script; Determining a target exercise script corresponding to the target communication scenario includes: Randomly selecting a preset drill script from at least one preset drill script corresponding to the target communication scenario as the target drill script; or, According to a third selection instruction input by the user, the target drill script is determined in at least one preset drill script corresponding to the target communication scenario.

6. The method according to any one of claims 1 to 3, characterized in that: The step of calculating the sentence score of the user voice obtained each time based on the user voice received after outputting the party's voice each time and the target mediator sentence corresponding to the party's voice in the exercise script includes: For any user voice received after outputting the party's voice, Convert the user's voice into text to obtain a user sentence; The semantic similarity between the user sentence and the target mediator sentence corresponding to the party's speech in the exercise script is calculated to obtain a sentence score for the user's speech.

7. The method according to any one of claims 1 to 3, characterized in that: The determining of the current user's speech score based on all sentence scores obtained in the human-machine rehearsal operation includes: Calculate the average of all sentence scores obtained in the human-machine rehearsal operation to determine the speech score; or, The median of all sentence scores obtained in the human-machine rehearsal operation is determined as the speech score.

8. The method according to any one of claims 1 to 3, characterized in that: After determining the speech score of the current user, the method further includes: Generating a training evaluation result according to the speech score; or, A training evaluation result is generated based on the chat record, wherein the chat record includes the output voice of the parties and the received user voice.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor is used to execute the computer program to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: A computer program / instruction is stored, and when the computer program / instruction is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images