Experiment report marking method, device and equipment and computer readable storage medium

Abstract

The invention discloses an experiment report marking method, device and equipment and a computer readable storage medium. The method comprises the following steps: receiving an experiment report file and storing the experiment report file into a reviewing queue; reading the experiment report file from a reading queue; analyzing the experiment report file to obtain text content and picture content; inputting a first prompt word and the text content into a text big language model, so that the text big language model generates a first evaluation for the text content; inputting a second prompt word and the picture content into a multi-modal model, so that the multi-modal model generates a second evaluation for the picture content; and integrating the first evaluation and the second evaluation to obtain a reviewing result of the experiment report file. According to the method and the device, the test report file is automatically read by utilizing the text large language model and the multi-modal model, so that the reading efficiency is improved, and the reading standard consistency is also ensured.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, device, and computer-readable storage medium for reviewing experimental reports. Background Art

[0002] In current university experimental courses, students are required to submit a large number of experimental reports that contain a mixture of text and images, usually in PDF format. Traditional review methods rely on teachers to manually complete the review, which has the following technical problems: Inefficiency: Teachers need to read each report, manually grade and write comments, and processing a single report takes an average of 5-10 minutes; Poor consistency in grading: Manual grading is easily affected by subjective factors, and the grading standards of different teachers or the same teacher for different reports may fluctuate. Summary of the Invention

[0003] To solve the above technical problems, the present application provides a method, apparatus, device and computer-readable storage medium for reviewing experimental reports.

[0004] In a first aspect, an embodiment of the present application provides a method for reviewing an experimental report, the method comprising: Receive experimental report files and store them in the review queue; Read the experimental report file from the review queue; Parsing the experimental report file to obtain text content and image content; Inputting the first prompt word and the text content into a text language model, so that the text language model generates a first evaluation of the text content; Inputting the second prompt word and the picture content into the multimodal model, so that the multimodal model generates a second evaluation of the picture content; The review result of the experimental report file is obtained by combining the first evaluation and the second evaluation.

[0005] In conjunction with the first aspect, in one embodiment, the experimental report review method includes: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

[0006] In conjunction with the first aspect, in one embodiment, the first evaluation includes a first score, the second evaluation includes a second score, and the review result of the experimental report file obtained by combining the first evaluation and the second evaluation includes: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

[0007] In conjunction with the first aspect, in one embodiment, after obtaining the review result of the experimental report file by integrating the first evaluation and the second evaluation, the method further includes: Feedback the review result to the review end of the experimental report file; When an approval instruction is received from the review end of the experimental report file, the review result of the experimental report file is sent to the submission end of the experimental report file.

[0008] In combination with the first aspect, in one implementation, the large text language model and the multimodal model are deployed in a private deployment manner.

[0009] In a second aspect, an embodiment of the present application provides a device for reviewing an experimental report, the device comprising: Receiving module, used to receive experimental report files and store them in the review queue; A reading module, used for reading the experimental report file from the review queue; A parsing module, used to parse the experimental report file to obtain text content and image content; A first evaluation module, configured to input a first prompt word and the text content into a text language model, so that the text language model generates a first evaluation of the text content; a second evaluation module, configured to input a second prompt word and the image content into the multimodal model, so that the multimodal model generates a second evaluation of the image content; The comprehensive evaluation module is used to comprehensively evaluate the first evaluation and the second evaluation to obtain the review result of the experimental report file.

[0010] In conjunction with the second aspect, in one embodiment, the apparatus for reviewing experimental reports further includes a generating module configured to: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

[0011] In conjunction with the second aspect, in one embodiment, the first evaluation includes a first score, the second evaluation includes a second score, and the comprehensive evaluation module is configured to: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

[0012] In a third aspect, an embodiment of the present application provides an experimental report review device, which includes a processor, a memory, and an experimental report review program stored on the memory and executable by the processor, wherein when the experimental report review program is executed by the processor, the steps of the experimental report review method described in the first aspect are implemented.

[0013] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which an experimental report review program is stored, wherein when the experimental report review program is executed by a processor, the steps of the experimental report review method described in the first aspect are implemented.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present application include: In an embodiment of the present application, an experimental report file is received and stored in a review queue; the experimental report file is read from the review queue; the experimental report file is parsed to obtain text content and image content; a first prompt word and the text content are input into a text large language model, so that the text large language model generates a first evaluation of the text content; a second prompt word and the image content are input into a multimodal model, so that the multimodal model generates a second evaluation of the image content; and the review result of the experimental report file is obtained by combining the first evaluation and the second evaluation. Through the embodiment of the present application, the review of the experimental report file is automatically completed using the text large language model and the multimodal model, which improves the review efficiency and ensures the consistency of the review standards. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of an embodiment of the method for reviewing an experimental report of this application; Figure 2 A schematic diagram of a scenario of an embodiment of the method for reviewing an experimental report of this application; Figure 3 This is a schematic diagram of the functional modules of an embodiment of the experimental report review device of this application; Figure 4 This is a schematic diagram of the hardware structure of the experimental report review equipment involved in the embodiment of this application. DETAILED DESCRIPTION

[0016] In the following, the technical solutions in the embodiments of the present application will be described clearly and completely in connection with the drawings in the embodiments of the present application, so that those skilled in the art can better understand the solutions in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0017] In order to make the purposes, technical solutions and advantages of the present application clearer, the embodiments of the present application will be described in further detail below in connection with the drawings.

[0018] In a first aspect, the embodiments of the present application provide an experimental report grading method.

[0019] In an embodiment, the experimental report grading method comprises the following steps. Figure 1 Figure 1 The flowchart of the first embodiment of the experimental report grading method of the present application is shown in FIG. 1. As shown in FIG. 1, the experimental report grading method comprises the following steps. Figure 1 Step S10, receiving an experimental report file and storing it in a grading queue; Step S20, reading the experimental report file from the grading queue; In the embodiment, the laboratory management platform can receive the original experimental report file submitted by the student end, then convert the original experimental report file into a PDF format experimental report file, and transmit it to the execution subject executing the method of the embodiment, so that the execution subject receives the experimental report file and stores it in the grading queue, and then reads the experimental report file from the grading queue.

[0020] Step S30, parsing the experimental report file to obtain text content and picture content; In the embodiment, in combination with the above description, the experimental report file is in PDF format, and the experimental report file can be parsed by calling Unstructed to split out the text content and picture content. The text content and picture content can be respectively stored in the memory and local disk, and in order to ensure data security, the text content and picture content can also be encrypted.

[0021] Step S40, inputting the first prompt word and the text content into a text large language model, so that the text large language model generates a first evaluation for the text content; Step S50, inputting the second prompt word and the picture content into a multi-modal model, so that the multi-modal model generates a second evaluation for the picture content; ​​In this embodiment, considering that the review process needs to call the text large language model and the multimodal model, it will consume a large amount of CPU and GPU resources on the execution entity. In order to avoid the normal progress of other tasks on the execution entity, it is possible to choose to call the text large language model and the multimodal model to perform the review task during a period when there are sufficient idle CPU and GPU resources.

[0022] In this embodiment, the large text language model and multimodal model are selected based on actual needs. For example, DeepSeek is used for the large text language model, and MiniCPM is used for the multimodal model. The selected large text language model and multimodal model can also be replaced, updated, and fine-tuned based on actual needs.

[0023] By calling the text large language model and inputting the first prompt word and text content, the text large language model outputs the corresponding first evaluation based on the text content according to the first prompt word; by calling the multimodal model and inputting the second prompt word and picture content, the multimodal model outputs the corresponding second evaluation based on the picture content according to the second prompt word.

[0024] Furthermore, in one embodiment, the experimental report review method includes: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

[0025] In this embodiment, the reviewer / teacher posts a lab report submission task on the laboratory management platform, enables the AI ​​automatic assisted review function, and sets review rules. The review rules set by each reviewer are read, and the first and second prompt words corresponding to each reviewer / teacher are generated.

[0026] When automatically reviewing the experimental report file, the corresponding reviewer / teacher is first determined, and then the corresponding first prompt word and second prompt word are read for performing steps S40 and S50.

[0027] Step S60: Combining the first evaluation and the second evaluation to obtain the review result of the experimental report file.

[0028] In this embodiment, the first evaluation and the second evaluation are fused, and the fusion result is used as the review result of the experimental report file.

[0029] Furthermore, in one embodiment, the first evaluation includes a first score, and the second evaluation includes a second score. Step S60 includes: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

[0030] In this embodiment, the accuracy of the large text language model and the multimodal model is determined by verifying model performance, for example, using a validation set. Assuming the accuracy of the large text language model is A and the accuracy of the multimodal model is B, the weight of the first score is α1 = A / (A + B), and the weight of the second score is α2 = B / (A + B).

[0031] In this embodiment, refer to Figure 2 , Figure 2 This is a scene diagram of an embodiment of the experimental report review method of this application. Figure 2 As shown, the text large language model DeepSeek and the multimodal model MiniCPM are integrated through LangChain; by sending text and text prompt words (i.e., the first prompt words) to DeepSeek, the text score S1 and comment C1 fed back by DeepSeek are obtained; by sending the image path and image prompt words (i.e., the second prompt words) to MiniCPM, the image score S2 and comment C2 fed back by MiniCPM are obtained; then the text score S1, comment C1, image score S2, and comment C2 are provided to the result fusion module for fusion processing, that is, the text score S1 and the image score S2 are weighted to obtain S, and the comment C1 and comment C2 are fused into the overall comment C, that is, the review result of the experimental report file (including S and C) is obtained.

[0032] On a single server with 24GB of video memory, it was verified that reviewing a 6-page PDF lab report with images and text took approximately 110 seconds, and reviewing a 4-page PDF lab report with images and text took approximately 90 seconds. The teacher's comments and scores for 40 lab reports were generally consistent with the teacher's actual review style, with the score difference within 5 points. Based on the time consumption assessment, a single server can batch process approximately 700 documents per day, significantly improving lab report review efficiency and reducing labor costs.

[0033] In the embodiment of the present application, the experimental report file is received and stored in the review queue; the experimental report file is read from the review queue; the text content and the picture content are obtained by analyzing the experimental report file; the first prompt word and the text content are input into the large language model to generate the first evaluation for the text content; the second prompt word and the picture content are input into the multi-modal model to generate the second evaluation for the picture content; and the review result of the experimental report file is obtained by synthesizing the first evaluation and the second evaluation. Through the embodiment of the present application, the review of the experimental report file is automatically completed by using the large language model and the multi-modal model, which improves the review efficiency and ensures the consistency of the review standard.

[0034] Further, in an embodiment, after step S60, the method further comprises: feeding back the review result to the review end of the experimental report file; when receiving the consent instruction fed back by the review end of the experimental report file, sending the review result of the experimental report file to the submission end of the experimental report file.

[0035] In the embodiment, after obtaining the review result, the review result is first fed back to the corresponding review end through the laboratory management platform for the reviewer to audit. If the audit is passed, the consent instruction is fed back through the related operation to trigger the laboratory management platform, and then sent to the execution subject of the embodiment. Finally, the execution subject of the embodiment sends the review result to the submission end of the experimental report file through the laboratory management platform. In addition, the laboratory management platform can also be directly used as the execution subject of the embodiment.

[0036] Further, when the reviewer is not satisfied with the review result, the review result can be modified for sending the modified review result to the submission end of the experimental report file.

[0037] Further, in an embodiment, the large language model and the multi-modal model are deployed by a private deployment method.

[0038] In the embodiment, the large language model and the multi-modal model are deployed by a private deployment method through a local model deployment tool (such as Ollama), so that the model and the student data are stored locally, and the data privacy and security are guaranteed.

[0039] In a second aspect, the embodiment of the present application also provides an experimental report review device.

[0040] In an embodiment, the experimental report review device comprises a review result obtaining unit, a review result synthesizing unit, and a review result feeding back unit. Figure 3 , Figure 3 FIG. 1 is a schematic diagram of the functional modules of an embodiment of the experimental report review device of the present application. As shown in FIG. 1, the experimental report review device comprises a review result obtaining unit 10, a review result synthesizing unit 20, and a review result feeding back unit 30. Figure 3As shown, the experimental report review device includes: Receiving module 10, for receiving experimental report files and storing them in the review queue; A reading module 20 is used to read the experimental report file from the review queue; The parsing module 30 is used to parse the experimental report file to obtain text content and image content; A first evaluation module 40 is configured to input the first prompt word and the text content into a text language model so that the text language model generates a first evaluation of the text content; A second evaluation module 50 is configured to input the second prompt word and the image content into the multimodal model so that the multimodal model generates a second evaluation of the image content; The comprehensive evaluation module 60 is used to comprehensively evaluate the first evaluation and the second evaluation to obtain the review result of the experimental report file.

[0041] Furthermore, in one embodiment, the apparatus for reviewing experimental reports further includes a generating module for: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

[0042] Furthermore, in one embodiment, the first evaluation includes a first score, the second evaluation includes a second score, and the comprehensive evaluation module 60 is configured to: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

[0043] Furthermore, in one embodiment, the apparatus for reviewing experimental reports further includes a feedback module for: Feedback the review result to the review end of the experimental report file; When an approval instruction is received from the review end of the experimental report file, the review result of the experimental report file is sent to the submission end of the experimental report file.

[0044] Furthermore, in one embodiment, the text large language model and the multimodal model are deployed through a private deployment method.

[0045] Among them, the functional implementation of each module in the above-mentioned experimental report reviewing device corresponds to the various steps in the above-mentioned experimental report reviewing method embodiment, and its functions and implementation processes will not be repeated here one by one.

[0046] In a third aspect, an embodiment of the present application provides an experimental report reviewing device, which may be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0047] Reference Figure 4 , Figure 4 Schematic diagram of the hardware structure of the experimental report review device involved in the embodiment of the present application. In the embodiment of the present application, the experimental report review device may include a processor, a memory, a communication interface and a communication bus.

[0048] The communication bus may be of any type and is used to interconnect the processor, memory, and communication interface.

[0049] Communication interfaces include input / output (I / O) interfaces, physical interfaces, and logical interfaces. These interfaces interconnect components within the lab report review system, as well as interfaces that connect the lab report review system to other devices (such as other computing devices or user devices). Physical interfaces can include Ethernet, fiber optic, and ATM interfaces; user devices can include displays and keyboards.

[0050] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0051] The processor may be a general-purpose processor that can call a lab report review program stored in a memory and execute the lab report review method provided in the embodiments of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the lab report review program is called may refer to the various embodiments of the lab report review method of the present application and will not be described in detail here.

[0052] Those skilled in the art will understand that Figure 4 The hardware structure shown in the figure does not constitute a limitation to the present application and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0053] In a fourth aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0054] The computer-readable storage medium of the present application stores an experimental report review program, wherein when the experimental report review program is executed by a processor, the steps of the experimental report review method as described above are implemented.

[0055] Among them, the method implemented when the experimental report review program is executed can refer to the various embodiments of the experimental report review method of this application, and will not be repeated here.

[0056] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0057] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0058] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0059] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0060] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0061] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of this application.

[0062] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for reviewing an experimental report, characterized in that: The experimental report review method includes: Receive experimental report files and store them in the review queue; Read the experimental report file from the review queue; Parsing the experimental report file to obtain text content and image content; Inputting the first prompt word and the text content into a text language model, so that the text language model generates a first evaluation of the text content; Inputting the second prompt word and the picture content into the multimodal model, so that the multimodal model generates a second evaluation of the picture content; The review result of the experimental report file is obtained by combining the first evaluation and the second evaluation.

2. The method for reviewing an experimental report according to claim 1, wherein: The experimental report review method includes: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

3. The method for reviewing an experimental report according to claim 1, wherein: The first evaluation includes a first score, the second evaluation includes a second score, and the review result of the experimental report file obtained by combining the first evaluation and the second evaluation includes: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

4. The method for reviewing an experimental report according to claim 1, wherein: After obtaining the review result of the experimental report file by integrating the first evaluation and the second evaluation, the method further includes: Feedback the review result to the review end of the experimental report file; When an approval instruction is received from the review end of the experimental report file, the review result of the experimental report file is sent to the submission end of the experimental report file.

5. The method for reviewing an experimental report according to claim 1, wherein: The text large language model and the multimodal model are deployed through a private deployment method.

6. A device for reviewing experimental reports, characterized in that: The experimental report review device comprises: Receiving module, used to receive experimental report files and store them in the review queue; A reading module, used for reading the experimental report file from the review queue; A parsing module, used to parse the experimental report file to obtain text content and image content; A first evaluation module, configured to input a first prompt word and the text content into a text language model, so that the text language model generates a first evaluation of the text content; a second evaluation module, configured to input a second prompt word and the image content into the multimodal model, so that the multimodal model generates a second evaluation of the image content; The comprehensive evaluation module is used to comprehensively evaluate the first evaluation and the second evaluation to obtain the review result of the experimental report file.

7. The experimental report reviewing device according to claim 6, characterized in that: The experimental report review device further includes a generating module for: Obtaining the review rules set by the reviewer of the experimental report file; A first prompt word and a second prompt word are generated based on the review rule.

8. The experimental report reviewing device according to claim 6, characterized in that: The first evaluation includes a first score, the second evaluation includes a second score, and the comprehensive evaluation module is used to: A weighted sum is performed on the first score and the second score to obtain a total score of the experimental report file, wherein the weights of the first score and the second score are determined based on the accuracy of the text large language model and the multimodal model.

9. An experimental report review device, characterized in that: The experimental report review device includes a processor, a memory, and an experimental report review program stored in the memory and executable by the processor, wherein when the experimental report review program is executed by the processor, the steps of the experimental report review method as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an experimental report review program, wherein when the experimental report review program is executed by a processor, the steps of the experimental report review method according to any one of claims 1 to 5 are implemented.

Images