Response information printing method and device, electronic equipment and computer readable medium

Abstract

The embodiment of the invention discloses a reply information printing method and device, electronic equipment and a computer readable medium. A specific embodiment of the method comprises the following steps: in response to an obtained reply content, obtaining an initial question stem prompt template; generating a prompt filling content set according to the reply content; filling the prompt filling content set into the initial question stem prompt template to generate a question stem prompt template; inputting the question stem prompt template into a question and answer model based on a large language model to generate reply score information and score basis information for the reply content; displaying the reply score information and the score basis information; and in response to determining that the confirmation information is received and the reply content is content in a paper form, controlling the target mechanical device to print literal content for the reply score information and the score basis information. According to the embodiment, on the premise that the question and answer model based on the large language model is utilized, the reply score information and score basis information of the reply content can be accurately and efficiently generated.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and in particular to a reply information printing method, device, electronic device, and computer-readable medium. Background Art

[0002] At present, in the fields of education, examination and assessment, the generation of test score results has always been an important and complex task. For the generation of the answer score information corresponding to the test questions, the usual method is: the traditional test score now mainly relies on manual online or offline marking, and the answer score information corresponding to the test questions is generated manually.

[0003] However, when the above method is used to generate the answer score information corresponding to the test questions, the following technical problems often occur: Online or offline scoring methods are inefficient and may be affected by the subjective factors of the scorer, resulting in inconsistent and unfair scoring results.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art. Summary of the invention

[0005] The content of this disclosure is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The content of this disclosure is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0006] Some embodiments of the present disclosure propose a reply information printing method, apparatus, electronic device, and computer-readable medium to solve one or more of the technical problems mentioned in the above background technology section.

[0007] In a first aspect, some embodiments of the present disclosure provide a reply information printing method, comprising: in response to obtaining reply content for a target question, obtaining an initial question prompt template; based on the reply content, generating a prompt filling content set for the target question to be filled in the initial question prompt template; filling the prompt filling content set into the initial question prompt template to generate a question prompt template for the target question; inputting the question prompt template into a question-answering model based on a large language model to generate reply score information and score basis information for the reply content; displaying the reply score information and the score basis information on a reply score information display interface; in response to determining that confirmation information for the display content corresponding to the reply score information display interface is received and the reply content is in paper form, controlling the target mechanical device to print text content for the reply score information and the score basis information at a corresponding position in the reply test paper corresponding to the reply content.

[0008] In a second aspect, some embodiments of the present disclosure provide a reply information printing device, comprising: an acquisition unit, configured to acquire an initial question stem prompt template in response to acquiring a reply content for a target question stem; a generation unit, configured to generate a prompt filling content set for the target question stem to be filled in the initial question stem prompt template according to the above reply content; a filling unit, configured to fill the above prompt filling content set into the initial question stem prompt template to generate a question stem prompt template for the target question stem; an input unit, configured to input the above question stem prompt template into a question and answer model based on a large language model to generate reply score information and score basis information for the above reply content; a display unit, configured to display the above reply score information and the above score basis information on a reply score information display interface; a control unit, configured to control a target mechanical device in response to determining that confirmation information for the display content corresponding to the above reply score information display interface is received and the above reply content is in paper form, so as to print text content for the above reply score information and the above score basis information at a corresponding position in the reply test paper corresponding to the above reply content.

[0009] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0010] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner in the first aspect is implemented.

[0011] The above-mentioned embodiments of the present disclosure have the following beneficial effects: through the reply information printing method of some embodiments of the present disclosure, under the premise of using the question-answering model based on the large language model, the reply score information and score basis information of the reply content can be accurately and efficiently generated. Specifically, the reason why the relevant reply score information is not accurate and efficient is that the scoring efficiency of the online or offline scoring method is low, and it may be affected by the subjective factors of the scorer, resulting in inconsistent and unfair scoring results. Based on this, the reply information printing method of some embodiments of the present disclosure, first, in response to obtaining the reply content for the target question stem, obtains the initial question stem prompt template. Here, the obtained initial question stem prompt template is used as the input template of the question-answering model later, so that the large language model can understand the content to be output. Then, according to the above reply content, a prompt filling content set to be filled in the above initial question stem prompt template for the above target question stem can be accurately generated, so as to obtain a more accurate question stem prompt template later. Secondly, the above prompt filling content set is filled into the above initial question stem prompt template to generate a more accurate question stem prompt template for the above target question stem. Then, the above-mentioned question stem prompt template is input into the question-answering model based on the large language model to accurately and efficiently generate the reply score information and score basis information for the above-mentioned reply content. Then, the above-mentioned reply score information and the above-mentioned score basis information are displayed on the reply score information display interface. Finally, in response to determining that the confirmation information corresponding to the display content of the above-mentioned reply score information display interface is received and the above-mentioned reply content is in paper form, the target mechanical device is controlled to print the text content of the above-mentioned reply score information and the above-mentioned score basis information at the corresponding position in the reply test paper corresponding to the above-mentioned reply content. In summary, by generating an accurate question stem prompt template, the subsequent question-answering model can clearly understand the content required to be output. On this basis, through the model accuracy and output efficiency of the question-answering model based on the large language model, the reply score information and the score basis information are accurately and efficiently output, so as to avoid the occurrence of inconsistent and unfair scoring results due to the influence of subjective factors of the scorer. In addition, by controlling the target mechanical device, the mechanized and automated supplementation of the reply score information and the score basis information can be achieved, which greatly improves the reply efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0013] Figure 1is a flowchart of some embodiments of the reply information printing method according to the present disclosure; Figure 2 is a schematic diagram of the structure of some embodiments of the reply information printing device according to the present disclosure; Figure 3 It is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0014] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0015] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure can be combined with each other.

[0016] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0017] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0018] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0019] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0020] refer to Figure 1 , shows a process 100 of some embodiments of the reply information printing method according to the present disclosure. The reply information printing method comprises the following steps: Step 101, in response to obtaining the answer content for the target question stem, obtain an initial question stem prompt template.

[0021] In some embodiments, in response to obtaining the answer content for the target question stem, the execution subject (for example, an electronic device) of the above-mentioned answer information printing method can obtain the initial question stem prompt template through a wired connection method or a wireless connection method. Among them, the target question stem can be the question stem to be answered. The answer content can be the content of the answer to the question corresponding to the target question stem. For example, the target question stem can be "Please write down the "Quatrain"". The answer content can be "Quatrain: Two orioles sing in the green willows, and a line of egrets fly up to the blue sky. The window contains the thousand-year snow of the Western Ridge, and the door is moored with the Dongwu Wanli ship.". The initial question stem prompt template can be a content template that represents the demand prompt content for scoring processing for the target question stem. In practice, the initial question stem prompt template can be a prompt example of a demand prompt content set that has not been filled in. The demand prompt content can be a prompt text of the desired demand content. The demand prompt content is used to allow the question-answering model based on the large language model to learn more prompt text content to output the demand content (that is, the scoring-related content). That is, the demand prompt content set is a data set for learning features for the question-answering model based on the large language model. Among them, the above-mentioned initial question prompt template can be pre-set for various answer scenarios.

[0022] For example, the initial prompt template could be: "Short answer question: ##Role <Content to be filled in>.

[0023] ##topic <Content to be filled in>.

[0024] ##Reference answer <Content to be filled in>.

[0025] ## Scoring Rules <Content to be filled in>. ”.

[0026] Step 102, based on the above-mentioned reply content, generate a set of prompt filling content for the above-mentioned target question stem, to be filled in the above-mentioned initial question stem prompt template.

[0027] In some embodiments, the execution subject may generate a set of prompt filling contents for the target question stem to be filled into the prompt template of the initial question stem according to the reply content. The prompt filling contents in the prompt filling contents may be required prompt contents. The number of prompt filling contents included in the prompt filling content set may be at least one.

[0028] For example, the prompt content set may include: "##Role: You are a marking teacher who needs to judge the answer level of each candidate and give corresponding scores and scoring basis", "##Topic: Please explain what the Internet of Things is, and give examples of the application of the Internet of Things in daily life", "##Reference answer: The Internet of Things (IoT) refers to a network that uses information sensing devices such as radio frequency identification, infrared sensors, global positioning systems, laser scanners, etc. to exchange and communicate information for any object in accordance with agreed protocols to achieve intelligent identification, positioning, tracking, monitoring and management. The Internet of Things (IoT) refers to a network that uses information sensing devices such as radio frequency identification, infrared sensors, global positioning systems, laser scanners, etc. to exchange and communicate information for any object in accordance with agreed protocols to achieve intelligent identification, positioning, tracking, monitoring and management.

[0029] Examples of IoT applications in daily life: Smart home: Through the Internet of Things technology, we can control the lights, air conditioners, televisions and other equipment in our homes, realize intelligent management, and improve the quality of life.

[0030] Smart health: Devices such as smart bracelets and smart watches can monitor our health status, such as heart rate, sleep, etc. in real time, and transmit the data to mobile phones or computers for analysis, helping us pay more attention to our health.

[0031] Smart transportation: Through IoT technology, we can achieve smart traffic management, such as smart traffic lights, smart parking systems, etc., to improve traffic efficiency and reduce traffic congestion. "

[0032] In some optional implementations of some embodiments, generating a set of prompt filling content for the target question stem to be filled into the initial question stem prompt template according to the reply content may include the following steps: In the first step, the execution subject may obtain at least one prompt filling content pre-bound to the target question stem, wherein the prompt filling content in the at least one prompt filling content may be a filling content pre-generated for the target question stem.

[0033] As an example, first, the execution subject may determine the stem type corresponding to the target stem. Then, the execution subject may obtain at least one prompt filling content corresponding to the stem type from the target prompt filling content database. The stem type may be a question type that characterizes the question corresponding to the stem. For example, the stem type may be, but is not limited to, one of the following: multiple choice type, question and answer type, filling type. The target prompt filling content database stores an association relationship table corresponding to the stem type and at least one prompt filling content.

[0034] In a second step, the execution entity may determine the reply content and the at least one prompt filling content as the prompt filling content set.

[0035] In some optional implementations of some embodiments, the at least one prompt to fill in content is generated by the following steps: The first step is to obtain a prompt filling content requirement page for the target question stem, wherein the prompt filling content requirement page may be a requirement page for filling in various prompt filling contents for the target question stem.

[0036] The second step is to receive at least one prompt filling content for the above-mentioned target question stem filled in or input by voice on the above-mentioned prompt filling content requirement page, wherein the above-mentioned prompt filling content requirement page includes: at least one content generation column for at least one prompt filling content. Each prompt filling content in the at least one prompt filling content has a corresponding content generation column and / or at least one content input voice control. Specifically, the content generation column can be a content selection column or a content filling column. The content input voice control can be a clickable page control for inputting relevant voice content.

[0037] In some optional implementations of some embodiments, the at least one prompt to fill in content includes at least one of the following: role content, instruction content, question content, reference answer content, scoring basis content, question answer content, and score calculation content. Among them, the role content can be the filling content related to the role. The instruction content can be the instruction content of the instruction representing what content to output. The question content can be the substantive content of the question. The reference answer content can be the answer content of the reference answer. The scoring basis content can be the basis content of the scoring basis. The question answer content can be the answer content of the answer to the question. The score calculation content can represent the calculation logic of how to calculate the score for the target question.

[0038] Optionally, at least one prompt fill-in content is generated by the following steps; The first step is to obtain the above target question stem.

[0039] The second step is to generate a prompt filling content requirement template for the above target question stem. The template content corresponding to the prompt filling content requirement template includes: the scoring key point requirement for the above target question stem and the reference answer requirement for the above target question stem. The prompt filling content requirement template can be a requirement template that represents how to obtain at least one prompt filling content for the target question stem. The scoring key point requirement can be information that represents the scoring key points corresponding to the target question stem. The reference answer requirement can be information that represents the reference answer corresponding to the target question stem.

[0040] The third step is to input the above-mentioned prompt filling content requirement template into the above-mentioned question-answering model based on the large language model to generate at least one prompt filling content for the above-mentioned target question stem.

[0041] In some optional implementations of some embodiments, the above-mentioned reply content is content in paper form or content in image form. The content in paper form may be the reply content in a paper test paper. The content in image form may be the reply content in an image after taking a picture of the paper test paper.

[0042] Optionally, the generating of a set of prompt content for the target question stem to be filled into the initial question stem prompt template according to the reply content may include the following steps: In the first step, a target scanning device is used to perform content scanning or text extraction on the above-mentioned answer content to generate target stem content and stem answer content. Content scanning can be a scan of the test paper content in a paper test paper. Text extraction can be image text extraction of an image after the paper test paper is photographed. Specifically, text extraction can be achieved through a related text extraction model. The target stem content can be the actual content of the target stem. The stem answer content can be the answer content of the test question fed back to the target stem.

[0043] The second step is to generate a set of prompts to fill in the content based on the above target question content and the above question answer content.

[0044] As an example, the execution entity may use the target question content and the question answer content as prompts to fill in the content set.

[0045] As another example, first, the role content, instruction content, score calculation content and scoring basis content for the target question are generated. Then, the target question content, the answer content, the role content, instruction content, score calculation content and scoring basis content of the target question are used as the prompt to fill in the content set.

[0046] Step 103, fill the above-mentioned prompt filling content set into the above-mentioned initial question stem prompt template to generate a question stem prompt template for the above-mentioned target question stem.

[0047] In some embodiments, the execution subject may fill in the initial question prompt template with the prompt content set to generate a question prompt template for the target question. The question prompt template may be a prompt example with the required prompt content set filled in.

[0048] Step 104, input the above question prompt template into the question-answering model based on the large language model to generate answer score information and score basis information for the above answer content.

[0049] In some embodiments, the above-mentioned execution entity may input the above-mentioned question prompt template into a question-answering model based on a large language model to generate reply score information and score basis information for the above-mentioned reply content. For example, the question-answering model based on a large language model may be a model product for question answering. For example, the question-answering model based on a large language model may be ChatGPT (Chat Generative Pre-trained Transformer). The reply score information may be the score information of the reply content. For example, the reply score information may be "the score is 4 points". The score basis information may be the reason why the reply content corresponds to the score of the reply score information. For example, the score basis information may be "1. The definition of the Internet of Things is given, which is clear and accurate - 2 points, 2. The description of how the Internet of Things devices operate and exchange data is relatively complete - 2 points".

[0050] Considering the problem that the conventional solution "using the question-answering model based on the large language model to generate the answer score information and the score basis information is not accurate enough, and the question-answering model of the large language model does not output the main features.", combined with the advantages / technical status, we can decide to adopt the following solution: In some optional implementations of some embodiments, the question-answering model based on the large language model includes: a template content division model, at least one semantic information generation model and at least one keyword extraction model, and an encoding and decoding model based on a temporal neural network model. Among them, the template content division model can be a neural network model that divides the content of each template content in the input question stem prompt template. In practice, the template content division model can be a content segmentation model based on a target symbol. For example, the target symbol can be "##". The semantic information generation model in at least one semantic information generation model can be a neural network model that generates content semantic information for the divided template content. The content semantic information can represent the main semantics of the template content. The question stem prompt template corresponds to each divided template content. Among them, the divided template content in each divided template content has a one-to-one corresponding input relationship with the semantic information generation model in at least one semantic information generation model. For example, the semantic information generation model can be a recurrent neural network model. The keyword extraction model in at least one keyword extraction model has a one-to-one corresponding input relationship with the divided template content in each divided template content. The keyword extraction model can be a neural network model that extracts a set of keywords in the corresponding divided template content. The encoding and decoding model based on the time series neural network model can be a time series neural network model that outputs the answer score information and the score basis information. Here, the template content division model, at least one semantic information generation model and at least one keyword extraction model, and the encoding and decoding model based on the time series neural network model are trained in a common conventional manner (for example, a training method based on back propagation) based on a large number of pre-acquired training samples.

[0051] Optionally, the question prompt template is input into a question-answering model based on a large language model to generate answer score information and score basis information for the answer content, including the following steps: In the first step, the question prompt template is input into the above-mentioned template content division model to generate each divided template content. Among them, the divided template content in each divided template content and the prompt filling content in the prompt filling content set have a one-to-one correspondence. Among them, the template content division model is a model for division based on template content keywords. The template content keyword can be a prompt keyword corresponding to the prompt filling content. For example, the template content keyword can be but is not limited to at least one of the following: role content, instruction content, reference answer.

[0052] In the second step, each of the divided template contents is input into a corresponding semantic information generation model in the at least one semantic information generation model to generate content semantic information, thereby obtaining each piece of content semantic information.

[0053] The third step is to input each of the divided template contents into a corresponding keyword extraction model in the at least one keyword extraction model to generate a keyword set, thereby obtaining each keyword set.

[0054] The fourth step is to generate each content semantic feature information for each content semantic information, and generate each keyword feature information for each of the above keyword sets. Among them, the content semantic feature information in each content semantic feature information has a one-to-one correspondence with the content semantic information in each content semantic information. The keyword feature information in each keyword feature information has a one-to-one correspondence with the keyword set in each keyword set. Among them, the content semantic feature information can represent the overall semantic feature information of the content semantic information. The keyword feature information can represent the overall keyword feature information of each keyword in the keyword set.

[0055] The fifth step is to perform corresponding feature information combinations on the above-mentioned content semantic feature information and the above-mentioned keyword feature information to generate each combination feature information.

[0056] In the sixth step, each combination feature information is input into the encoding and decoding model based on the temporal neural network model to generate the reply score information and score basis information for the above reply content.

[0057] The above-mentioned content of "in some optional implementation methods of some embodiments" is one of the inventive points of the present disclosure, which solves the technical problem mentioned in the background technology that "the use of a question-answering model based on a large language model to generate answer score information and score basis information is not accurate enough." Based on this, the question-answering model based on a large language model of the present disclosure includes a template content division model, at least one semantic information generation model and at least one keyword extraction model, and an encoding and decoding model based on a temporal neural network model, to extract rich features from the perspective of subject semantics and keywords in many aspects, and on the basis of the encoding and decoding model based on the temporal neural network model, the answer score information and score basis information can be accurately generated.

[0058] In some optional implementations of some embodiments, after step 104, the steps further include: In the first step, for the above-mentioned score basis information and the above-mentioned answer score information, the following first generation step is performed: Sub-step 1, generating a set of judgment prompts for filling in content that represents the judgment of the scoring basis information. The set of judgment prompts for filling in content includes: judgment role content, judgment instruction content, question stem content, scoring basis content, question stem answer content, score calculation content and scoring basis information. The scoring basis judgment can be to judge whether the scoring basis is accurate.

[0059] Sub-step 2, fill the above-mentioned judgment prompt filling content set into the above-mentioned initial judgment question prompt template to generate a judgment question prompt template for the above-mentioned score basis information. Among them, the judgment question prompt template can include the score basis information.

[0060] Sub-step 3, input the above-mentioned judgment question prompt template into the judgment model based on the large language model to generate judgment score information and judgment basis information for the above-mentioned scoring basis information, wherein the above-mentioned judgment model based on the large language model and the above-mentioned question-answering model based on the large language model are synchronously trained. The judgment model based on the large language model can be a model that uses a large language model to judge the scoring basis information. That is, the judgment model based on the large language model determines the accuracy of the scoring basis information. The judgment score information can characterize the information accuracy of the scoring basis information. The judgment basis information can be the reason for the judgment score information. The judgment model based on the large language model and the question-answering model based on the large language model can be two models with different large language model structures.

[0061] Sub-step 4, in response to determining that the judgment score information is not lower than the target score information, generating confirmation information indicating that the response score information and the score basis information are correct. The target score information may be a preset score value.

[0062] Sub-step 5, in response to determining that the determination score information is lower than the target score information, executing the following second generation step: The first sub-step is to fill in the above-mentioned prompt filling content set and the above-mentioned judgment basis information into the above-mentioned initial question stem prompt template to generate a supplementary question stem prompt template for the above-mentioned target question stem.

[0063] The second sub-step is to input the above-mentioned supplementary question prompt template into the question-answering model based on the large language model to generate supplementary reply score information and supplementary score basis information for the above-mentioned reply content.

[0064] In the third sub-step, the supplementary score basis information and the supplementary reply score information are used as the score basis information and the reply score information respectively, and the first generation step is continued.

[0065] And the above method also includes: The first step is to obtain the actual answer score information and the actual score basis information corresponding to the above-mentioned question prompt template. The actual answer score information can be the actual score information after at least one relevant technical personnel manually calculates the score of the answer content. The actual score basis information can be the score basis information after at least one relevant technical personnel manually calculates the score of the answer content.

[0066] In the second step, a reply sample is generated based on the obtained score basis information set, reply score information set, judgment score information set and judgment basis information set, the above-mentioned actual reply score information and the above-mentioned actual score basis information.

[0067] As an example, the above-mentioned execution entity can directly package the obtained score basis information set, reply score information set, judgment score information set and judgment basis information set, the above-mentioned actual reply score information and the above-mentioned actual score basis information to generate a reply sample.

[0068] As another example, first, the execution subject may combine the corresponding score basis information, reply score information, determination score information and determination basis information for each first generation step to generate single combination information. Then, the obtained single combination information set, actual reply score information and the actual score basis information are sample combined to generate a reply sample.

[0069] The third step is to store the reply samples in a sample database, where the sample database may store samples for each large model training.

[0070] In step 4, in response to determining that the model training time has been reached, a response sample set is obtained from the sample database. The model training time may be the training time for training the decision model based on the large language model and the question-answering model based on the large language model. The model training time may be set regularly.

[0071] The fifth step is to retrain the judgment model based on the large language model and the question-answering model based on the large language model according to the above-mentioned response sample set to generate a trained judgment model and a trained question-answering model.

[0072] As an example, the above-mentioned execution entity can retrain the judgment model based on the large language model and the above-mentioned question-answering model based on the large language model according to the response sample set and based on the training method of the generative adversarial model to generate a post-training judgment model and a post-training question-answering model.

[0073] In the sixth step, the post-training judgment model and the post-training question-answering model are respectively determined as a judgment model based on a large language model and a question-answering model based on a large language model to put the model online.

[0074] The above-mentioned content of "in some optional implementation methods of some embodiments" as one of the inventive points of the present disclosure solves the technical problem mentioned in the background technology that "the use of a question-answering model based on a large language model to generate reply score information and score basis information cannot be accurately guaranteed. If the question-answering model based on a large language model is not accurate enough, a large amount of computing resources will be wasted on the basis of inaccurate scores." Based on this, the present disclosure can effectively determine the accuracy of the output content of the question-answering model based on a large language model through a judgment model based on a large language model to ensure the accuracy of the reply score information and the score basis information. On this basis, by jointly training the question-answering model of the large language model and the judgment model of the large language model, the accuracy of the score output of the question-answering model and the judgment accuracy of the judgment model can be guaranteed.

[0075] Step 105, displaying the above-mentioned reply score information and the above-mentioned score basis information on the reply score information display interface.

[0076] In some embodiments, the execution entity may display the reply score information and the score basis information on a reply score information display interface, wherein the reply score information display interface may be an interface for displaying the reply score information and the score basis information.

[0077] Step 106, in response to determining that confirmation information for the display content corresponding to the above-mentioned answer score information display interface is received and the above-mentioned answer content is in paper form, control the target mechanical device to print the text content for the above-mentioned answer score information and the above-mentioned score basis information at the corresponding position in the answer test paper corresponding to the above-mentioned answer content. Among them, the confirmation information can represent information that the display content is confirmed to be correct. The target mechanical device can be a device for printing the corresponding score information and the score basis information. In practice, the target mechanical device can be controlled to implement a text printing operation in a paper-based answer test paper. The corresponding position can be the position of the answer column corresponding to the target question stem.

[0078] In some embodiments, in response to determining that confirmation information has been received regarding the display content corresponding to the above-mentioned reply score information display interface and the above-mentioned reply content is in paper form, the above-mentioned execution entity can control the target mechanical device to print the text content of the above-mentioned reply score information and the above-mentioned score basis information at the corresponding position in the reply test paper corresponding to the above-mentioned reply content.

[0079] The above-mentioned embodiments of the present disclosure have the following beneficial effects: through the reply information printing method of some embodiments of the present disclosure, under the premise of using the question-answering model based on the large language model, the reply score information and score basis information of the reply content can be accurately and efficiently generated. Specifically, the reason why the relevant reply score information is not accurate and efficient is that the scoring efficiency of the online or offline scoring method is low, and it may be affected by the subjective factors of the scorer, resulting in inconsistent and unfair scoring results. Based on this, the reply information printing method of some embodiments of the present disclosure, first, in response to obtaining the reply content for the target question stem, obtains the initial question stem prompt template. Here, the obtained initial question stem prompt template is used as the input template of the question-answering model later, so that the large language model can understand the content to be output. Then, according to the above reply content, a prompt filling content set to be filled in the above initial question stem prompt template for the above target question stem can be accurately generated, so as to obtain a more accurate question stem prompt template later. Secondly, the above prompt filling content set is filled into the above initial question stem prompt template to generate a more accurate question stem prompt template for the above target question stem. Then, the above-mentioned question stem prompt template is input into the question-answering model based on the large language model to accurately and efficiently generate the reply score information and score basis information for the above-mentioned reply content. Then, the above-mentioned reply score information and the above-mentioned score basis information are displayed on the reply score information display interface. Finally, in response to determining that the confirmation information corresponding to the display content of the above-mentioned reply score information display interface is received and the above-mentioned reply content is in paper form, the target mechanical device is controlled to print the text content of the above-mentioned reply score information and the above-mentioned score basis information at the corresponding position in the reply test paper corresponding to the above-mentioned reply content. In summary, by generating an accurate question stem prompt template, the subsequent question-answering model can clearly understand the content required to be output. On this basis, through the model accuracy and output efficiency of the question-answering model based on the large language model, the reply score information and the score basis information are accurately and efficiently output, so as to avoid the occurrence of inconsistent and unfair scoring results due to the influence of subjective factors of the scorer. In addition, by controlling the target mechanical device, the mechanized and automated supplementation of the reply score information and the score basis information can be achieved, which greatly improves the reply efficiency.

[0080] Further references Figure 2 As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a reply information printing device, and these device embodiments are similar to Figure 1 Corresponding to the method embodiments shown, the reply information printing device can be specifically applied to various electronic devices.

[0081] like Figure 2As shown, a reply information printing device 200 includes: an acquisition unit 201, a generation unit 202, a filling unit 203, an input unit 204, a display unit 205 and a control unit 206. Among them, the acquisition unit 201 is configured to obtain an initial question stem prompt template in response to obtaining a reply content for a target question stem; the generation unit 202 is configured to generate a prompt filling content set for the target question stem to be filled in the initial question stem prompt template according to the above reply content; the filling unit 203 is configured to fill the above prompt filling content set into the initial question stem prompt template to generate a question stem prompt template for the target question stem; the input unit 204 is configured to input the above question stem prompt template into a question-answering model based on a large language model to generate a reply score information and score basis information for the above reply content; the display unit 205 is configured to display the above reply score information and the above score basis information on the reply score information display interface. The control unit 206 is configured to control the target mechanical device to print the text content of the above-mentioned reply score information and the above-mentioned score basis information at the corresponding position in the reply test paper corresponding to the above-mentioned reply content in response to determining that confirmation information corresponding to the display content of the above-mentioned reply score information display interface has been received and the above-mentioned reply content is in paper form.

[0082] It is understood that the units described in the reply information printing device 200 are similar to those described in the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the reply information printing device 200 and the units included therein, and will not be described in detail here.

[0083] Reference below Figure 3 , which shows a structural schematic diagram of an electronic device (eg, an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0084] like Figure 3 As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 to a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0085] Typically, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0086] In particular, according to some embodiments of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the method of some embodiments of the present disclosure are executed.

[0087] It should be noted that the computer-readable medium mentioned above in some embodiments of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0088] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0089] The computer-readable medium may be included in the electronic device; or it may exist independently without being assembled into the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: in response to obtaining the answer content for the target question, obtain the initial question prompt template; based on the answer content, generate a prompt filling content set for the target question to be filled in the initial question prompt template; fill the prompt filling content set into the initial question prompt template to generate a question prompt template for the target question; input the question prompt template into the question-answering model based on the large language model to generate the answer score information and score basis information for the answer content; display the answer score information and the score basis information on the answer score information display interface; in response to determining that the confirmation information for the corresponding display content of the answer score information display interface is received and the answer content is in paper form, control the target mechanical device to print the text content for the answer score information and the score basis information at the corresponding position in the answer test paper corresponding to the answer content.

[0090] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0091] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0092] The units described in some embodiments of the present disclosure may be implemented by software or by hardware. The described units may also be provided in a processor, for example, may be described as: a processor including an acquisition unit, a generation unit, a filling unit, an input unit, a display unit and a control unit. The names of these units do not constitute limitations on the units themselves in certain circumstances, for example, the acquisition unit may also be described as "a unit for acquiring an initial question prompt template in response to acquiring the answer content for the target question stem".

[0093] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0094] The above descriptions are only some preferred embodiments of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A method for printing a reply message, comprising: In response to obtaining the answer content for the target question stem, obtaining an initial question stem prompt template; Generating a set of prompt filling content for the target question stem to be filled into the initial question stem prompt template according to the reply content; Filling the prompt filling content set into the initial question stem prompt template to generate a question stem prompt template for the target question stem; Inputting the question prompt template into a question-answering model based on a large language model to generate answer score information and score basis information for the answer content; The answer score information and the score basis information are displayed on the answer score information display interface; In response to determining that confirmation information corresponding to the display content of the reply score information display interface has been received and the reply content is in paper form, the target mechanical device is controlled to print the text content of the reply score information and the score basis information at a corresponding position in the reply test paper corresponding to the reply content.

2. The method according to claim 1, wherein: The step of generating a prompt content set for the target question stem to be filled into the initial question stem prompt template according to the reply content includes: Acquire at least one prompt filling content pre-bound to the target question; The reply content and the at least one prompt filling content are determined as the prompt filling content set.

3. The method according to claim 1, wherein: The reply content is in paper form or image form; as well as The step of generating a prompt content set for the target question stem to be filled into the initial question stem prompt template according to the reply content includes: Using a target scanning device, scanning or extracting text from the reply content to generate target stem content and stem reply content; A set of prompt filling content is generated based on the target question content and the question answer content.

4. The method according to claim 1, wherein: The reply content is obtained by following the steps below: In response to receiving answer submission information for an online answering interface, extracting answer content for the target question in the online answering interface; The reply content is sent to the reply score information generating terminal.

5. The method according to claim 2, wherein: The at least one prompt to fill in content is generated by the following steps; Obtain a prompt filling content requirement page for the target question stem; Receive at least one prompt filling content for the target question stem filled in or voice-input on the prompt filling content requirement page, wherein the prompt filling content requirement page includes: at least one content generation field and / or at least one content input voice control for at least one prompt filling content.

6. The method according to claim 2, wherein: The at least one prompt content includes at least one of the following: role content, instruction content, question content, reference answer content, scoring basis content, question answer content, and score calculation content; And at least one prompt fill-in content is generated by the following steps; Obtaining the target question stem; Generate a prompt filling content requirement template for the target question stem, wherein the prompt filling content requirement template corresponds to the template content including: the scoring key point requirement for the target question stem and the reference answer requirement for the target question stem; The prompt filling content requirement template is input into the question-answering model based on the large language model to generate at least one prompt filling content for the target question stem.

7. A reply information printing device, comprising: An acquisition unit, configured to acquire an initial question prompt template in response to acquiring a response content for a target question; A generating unit is configured to generate a prompt filling content set for the target question stem to be filled into the initial question stem prompt template according to the answer content; A filling unit is configured to fill the prompt filling content set into the initial question stem prompt template to generate a question stem prompt template for the target question stem; An input unit, configured to input the question prompt template into a question-answering model based on a large language model to generate answer score information and score basis information for the answer content; A display unit, configured to display the reply score information and the score basis information on a reply score information display interface; The control unit is configured to control the target mechanical device to print the text content of the reply score information and the score basis information at the corresponding position in the reply test paper corresponding to the reply content in response to determining that confirmation information corresponding to the display content of the reply score information display interface has been received and the reply content is in paper form.

8. An electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images