Paper review method and equipment based on multidimensional data fusion

Abstract

The present application provides a paper review method and device based on multidimensional data fusion, which relates to the field of electronic digital data processing. The method includes: extracting target record data matching the target paper from a software system used to store research record data and punch-in information, obtaining the time dimension score, spatial variables and text relevance coefficient of the target paper; determining the content quality feature vector of the target paper based on natural language processing technology, and determining the text quantity index of the target paper; inputting the time dimension score, text relevance coefficient, text quantity index, spatial variables and content quality feature vector into a multidimensional data fusion paper review model based on an interactive attention mechanism to output the paper review score. The present application can improve the accuracy of the temporary patchwork judgment of the paper review, thereby effectively reducing the resource utilization and storage cost of the equipment performing the paper review, and improving the rationality and effectiveness of the paper review score.

Description

Technical Field

[0001] The present application relates to the technical field of electronic digital data processing, and in particular to a paper review method and device based on multi-dimensional data fusion. Background Art

[0002] Paper review refers to the process of examining and evaluating the content and structure of academic papers. Its purpose is to ensure the scholarly nature, standardization, and integrity of the paper, and to improve its quality and level. Traditional review and assessment methods involve experts conducting reviews and assigning scores to the papers. However, this manual review process is not only time-consuming, but also, due to its over-reliance on templated requirements, may cause authors to sacrifice the coherence of the content to meet formatting standards. Furthermore, the subjectivity of manual review makes it difficult to effectively identify the authenticity of the research process, trace actual behavior within the research cycle, and determine whether the paper content is derived from literature compilation or actual research accumulation. Therefore, the automation of paper review has become a research focus.

[0003] Currently, among existing automated paper review methods, one approach involves AI-powered paper quality profiling. This involves AI-powered collaborative review systems training a "paper quality profiling" model and combining it with historical review data to predict the risk of paper retraction. Other approaches utilize blockchain technology to store and trace experimental data related to papers, storing experimental data and revision records on-chain to ensure data immutability. Alternatively, smart contracts can be used to automatically identify relationships between authors and funders, preventing conflicts of interest.

[0004] However, AI-based paper quality profiling relies on historical data to train the model. When encountering emerging research areas, the model may have little or no historical data to refer to, leading to a high rate of misjudgment. While blockchain technology can ensure data immutability and store experimental data and modification records on-chain, it faces significant challenges, including high equipment resource utilization, high storage costs, and technical barriers to entry.

[0005] Therefore, there is an urgent need to design a method that can improve the rationality and effectiveness of automated paper review and reduce the equipment resource usage in the technology of determining whether a paper is temporarily assembled. Summary of the Invention

[0006] In view of this, the embodiments of the present application provide a paper review method and device based on multidimensional data fusion to eliminate or improve one or more defects in the prior art.

[0007] One aspect of the present application provides a paper review method based on multidimensional data fusion, comprising:

[0008] Extracting research record data that matches the current target paper from a software system for storing research record data and punch-in information periodically submitted by users as target record data, and obtaining, based on the target record data and the paper text data corresponding to the target paper, a time dimension score corresponding to the target paper for representing the time distribution of paper research, a spatial variable for representing the spatial distribution of paper research, and a text correlation coefficient for representing the similarity between the paper text data and the target record data;

[0009] Furthermore, based on natural language processing technology, a content quality feature vector corresponding to the target paper for representing the quality of the paper content is determined according to the paper text data, and a text quantity index corresponding to the target paper for representing the proportion of quantitative dimension information in the paper is determined according to the paper text data;

[0010] The time dimension score, the text relevance coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper are input into a multidimensional data fusion paper review model based on an interactive attention mechanism, so that the paper review model outputs a paper review score corresponding to the target paper.

[0011] In some embodiments of the present application, extracting the research record data that matches the current target paper as the target record data from the software system for storing the research record data and check-in information periodically submitted by the user includes:

[0012] Extracting each research record data periodically submitted by the user who is the author of the target paper from a software system for storing research record data and check-in information periodically submitted by the user;

[0013] Extracting the paper text data corresponding to the target paper and the keywords corresponding to each of the research record data respectively, and calculating the frequency and inverse document frequency of each of the keywords in the paper text data or the research record data;

[0014] According to the frequencies and inverse document frequencies corresponding to the respective keywords, a word frequency inverse document frequency algorithm is used to search for research record data that matches the keywords in the target paper in the respective research record data to serve as target record data.

[0015] In some embodiments of the present application, the target record data and the paper text data corresponding to the target paper are used to obtain the time dimension score corresponding to the target paper for representing the time distribution of paper research, the spatial variable for representing the spatial distribution of paper research, and the text correlation coefficient for representing the similarity between the paper text data and the target record data, including:

[0016] Extracting the submission time corresponding to the target record data from the software system to obtain the submission time period corresponding to all the target record data; determining the time dimension score corresponding to the target paper for representing the research time distribution of the paper according to the submission time period and a preset empirical threshold;

[0017] According to the target record data and the paper text data corresponding to the target paper, a spatial variable corresponding to the target paper is obtained for representing the spatial distribution of the paper research;

[0018] Furthermore, based on the target record data and the paper text data corresponding to the target paper, a text correlation coefficient corresponding to the target paper is obtained, which is used to represent the similarity between the paper text data and the target record data.

[0019] In some embodiments of the present application, obtaining, based on the target record data and the paper text data corresponding to the target paper, a spatial variable corresponding to the target paper for representing the spatial distribution of paper research includes:

[0020] Obtaining the check-in information corresponding to each target record data submitted by the user who is the author of the target paper in the software system, and determining the geographical location information corresponding to each check-in information according to the geographic information system;

[0021] The geographical location information is periodically encoded to obtain a spatial variable corresponding to the target paper for representing the spatial distribution of the paper research.

[0022] In some embodiments of the present application, obtaining, based on the target record data and the paper text data corresponding to the target paper, a text correlation coefficient corresponding to the target paper for indicating the similarity between the paper text data and the target record data includes:

[0023] Generate summary text data corresponding to the target record data based on the large language model;

[0024] Inputting the summary text data and the paper text data into a BERT model respectively, so that the BERT model outputs a summary feature vector corresponding to the summary text data and a paper feature vector corresponding to the paper text data respectively;

[0025] The cosine similarity is used to calculate the similarity between the abstract feature vector and the paper feature vector, so as to serve as a text correlation coefficient corresponding to the target paper and used to represent the similarity between the paper text data and the target record data.

[0026] In some embodiments of the present application, determining the content quality feature vector corresponding to the target paper and used to represent the content quality of the paper based on the paper text data based on natural language processing technology includes:

[0027] Preprocessing the paper text data; wherein the preprocessing includes: removing noise, sentence segmentation and word segmentation;

[0028] Based on the corresponding word segmenter of natural language processing technology, obtain the word embedding vector sequence corresponding to the preprocessed paper text data;

[0029] Linear mapping is performed on the hidden layer feature vectors corresponding to the word embedding vector sequence to obtain a content quality feature vector corresponding to the target paper for representing the content quality of the paper.

[0030] In some embodiments of the present application, determining the text quantity index corresponding to the target paper and used to represent the proportion of quantity dimension information in the paper based on the paper text data includes:

[0031] Obtain the data volume, number of experiments and number of cited literature corresponding to the paper text data;

[0032] Standardizing the data volume, the number of experiments, and the number of cited documents, respectively, to obtain a first standard value corresponding to the data volume, a second standard value corresponding to the number of experiments, and a third standard value corresponding to the number of cited documents;

[0033] Calculate the original quantity index corresponding to the paper text data according to the first standard value, the second standard value, the third standard value and their corresponding weights;

[0034] The original quantity index is linearly transformed to convert it to the interval [0, 10] to obtain the text quantity index corresponding to the target paper, which is used to represent the proportion of quantity dimension information in the paper.

[0035] In some embodiments of the present application, the multidimensional data fusion paper review model based on the interactive attention mechanism includes:

[0036] A scalar embedding layer is used to perform scalar embedding on the time dimension score, the text relevance coefficient, and the text quantity index corresponding to the target paper to obtain a time feature vector corresponding to the time dimension score, a text relevance feature vector corresponding to the text relevance coefficient, and a text quantity feature vector corresponding to the text quantity index;

[0037] A vector projection layer, configured to perform vector projection on the spatial variable and the content quality feature vector corresponding to the target paper to obtain a projected spatial variable corresponding to the spatial variable and a projected content quality feature vector corresponding to the content quality feature vector;

[0038] An input sequence construction layer, configured to construct a corresponding input sequence according to the time feature vector, the text relevance feature vector, the text quantity feature vector, the projected spatial variable, and the projected content quality feature vector;

[0039] a spatial and content cross attention layer, configured to use the projected content quality feature vector in the input sequence as a query vector and the projected spatial variable as a key-value pair vector to perform a cross attention calculation to obtain a corresponding spatial and content fusion feature vector;

[0040] a quantity and quality cross attention layer, configured to concatenate the projected content quality feature vector in the input sequence with the spatial and content fusion feature vector to obtain a corresponding first concatenated vector; use the first concatenated vector as a query vector and the text quantity feature vector in the input sequence as a key-value pair vector to perform cross attention calculation to obtain a corresponding quantity and quality fusion feature vector;

[0041] a vector splicing layer, configured to splice the projected content quality feature vector, the projected spatial variable, and the quantity and quality fusion feature vector in the input sequence to obtain a corresponding second splicing vector;

[0042] a dynamic fusion layer, configured to dynamically fuse the second concatenated vector using the text relevance feature vector in the input sequence as a modulation factor;

[0043] a three-dimensional attention routing layer, configured to obtain a target output representation corresponding to the target paper based on the self-interaction result data between the time feature vector, the text relevance feature vector, and the text quantity feature vector, the self-interaction result data between the projected content quality feature vector and the projected spatial variable, and the second splicing vector after dynamic fusion;

[0044] A linear layer is used to map the target output representation to a paper review score corresponding to the target paper.

[0045] Another aspect of the present application provides a paper review device based on multidimensional data fusion, comprising:

[0046] A first multidimensional data acquisition module is used to extract research record data that matches the current target paper from a software system for storing research record data and punch-in information periodically submitted by users as target record data, and to obtain, based on the target record data and the paper text data corresponding to the target paper, a time dimension score corresponding to the target paper for representing the time distribution of paper research, a spatial variable for representing the spatial distribution of paper research, and a text correlation coefficient for representing the similarity between the paper text data and the target record data;

[0047] And, a second multidimensional data acquisition module is used to determine, based on natural language processing technology, a content quality feature vector corresponding to the target paper for representing the quality of the paper content according to the paper text data, and to determine, based on the paper text data, a text quantity index corresponding to the target paper for representing the proportion of quantity dimension information in the paper;

[0048] The paper review scoring module is used to input the time dimension score, the text relevance coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper into a multidimensional data fusion paper review model based on the interactive attention mechanism, so that the paper review model outputs the paper review score corresponding to the target paper.

[0049] The third aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the paper review method based on multidimensional data fusion when executing the computer program.

[0050] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the paper review method based on multidimensional data fusion.

[0051] The fifth aspect of the present application provides a computer program product, including a computer program, which implements the paper review method based on multidimensional data fusion when executed by a processor.

[0052] The paper review method based on multidimensional data fusion provided in the present application extracts research record data matching the current target paper as target record data from a software system for storing research record data and check-in information periodically submitted by users, and obtains the time dimension score corresponding to the target paper for representing the time distribution of paper research, the spatial variable for representing the spatial distribution of paper research, and the text correlation coefficient for representing the similarity between the paper text data and the target record data according to the target record data and the paper text data corresponding to the target paper respectively; and, based on natural language processing technology, determines the content quality feature vector corresponding to the target paper for representing the content quality of the paper according to the paper text data, and determines the text quantity index corresponding to the target paper for representing the proportion of quantity dimension information in the paper according to the paper text data; inputs the time dimension score, the text correlation coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper into a multidimensional data fusion paper review model based on the interactive attention mechanism, so that the paper review model outputs the paper review corresponding to the target paper. Score; through the software system for storing the research record data and check-in information periodically submitted by users, the research record data periodically submitted by users that matches the current target paper is extracted as the target record data, and according to the target record data and the paper text data corresponding to the target paper, the time dimension score, spatial variable and text correlation coefficient corresponding to the target paper are respectively obtained, which can effectively trace the user's daily paper research time, research space distribution and similarity, and can effectively improve the accuracy of identifying whether the paper content is temporarily assembled; and, by automatically scoring the paper review score by using the time dimension score, the text correlation coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper as multidimensional data, the rationality and effectiveness of the paper review score can be improved; that is, the present application can improve the accuracy of the temporary assembly judgment of the paper review without adopting blockchain technology, thereby effectively reducing the resource occupancy and storage cost of the equipment performing the paper review and lowering the technical threshold; and compared with the method of adopting AI paper quality portrait, it can effectively improve the rationality and effectiveness of the paper review score and reduce the misjudgment rate.

[0053] Additional advantages, purposes, and features of the present application will be described in part in the following description and will become apparent to those skilled in the art upon study of the following or may be learned from practice of the present application. The purposes and other advantages of the present application may be achieved and obtained by the structures specifically pointed out in the specification and drawings.

[0054] Those skilled in the art will understand that the purposes and advantages that can be achieved by the present application are not limited to the above specific description, and the above and other purposes that can be achieved by the present application will be more clearly understood based on the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The drawings described herein are intended to provide a further understanding of the present application, constitute a part of the present application, and do not constitute a limitation of the present application. The components in the drawings are not drawn to scale, but are only for the purpose of illustrating the principles of the present application. In order to facilitate the illustration and description of some parts of the present application, the corresponding parts in the drawings may be enlarged, that is, they may become larger than other components in the exemplary device actually manufactured according to the present application. In the drawings:

[0056] Figure 1 This is a first flow chart of a paper review method based on multidimensional data fusion in one embodiment of the present application.

[0057] Figure 2 This is a second flow chart of the paper review method based on multidimensional data fusion in one embodiment of the present application.

[0058] Figure 3 This is a third flow chart of the paper review method based on multidimensional data fusion in one embodiment of the present application.

[0059] Figure 4 This is a schematic diagram of the architecture of a multi-dimensional data fusion paper review model based on an interactive attention mechanism in one embodiment of the present application.

[0060] Figure 5 This is a structural diagram of a paper review device based on multidimensional data fusion in one embodiment of the present application.

[0061] Figure 6 This is a schematic diagram of the application of a multi-dimensional data fusion paper review model based on an interactive attention mechanism in an application example of this application.

[0062] Figure 7 This is a flowchart of the graduate student monthly report system in an application example of this application.

[0063] Figure 8 A flowchart illustrating the process of submitting monthly or annual report information for students in an application example of this application.

[0064] Figure 9 This is a flowchart of the process of the management end reviewing the monthly or annual report information submitted by students in the application example of this application. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail in conjunction with the embodiments and drawings. Here, the illustrative embodiments of this application and their descriptions are used to explain this application, but are not intended to limit this application.

[0066] It should also be noted here that in order to avoid obscuring the present application due to unnecessary details, the accompanying drawings only show structures and / or processing steps that are closely related to the scheme according to the present application, while other details that are not closely related to the present application are omitted.

[0067] It should be emphasized that the term "include / comprises" when used herein refers to the existence of features, elements, steps or components, but does not exclude the existence or addition of one or more other features, elements, steps or components.

[0068] It should also be noted that, unless otherwise specified, the term "connection" herein may refer not only to a direct connection but also to an indirect connection involving an intermediate.

[0069] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0070] It should be noted that the general process of the traditional paper review and assessment method is that the editor first conducts preliminary screening and formal review of the paper, and the reviewers conduct multiple rounds of review of the paper, checking for plagiarism and data authenticity, and finally checking for innovation, etc. before entering the publication process. However, when faced with the serious disconnect between the content of graduate students' papers and daily research, the following disadvantages may be exposed: (1) At the level of formal review, over-reliance on template requirements may cause students to sacrifice the coherence of content in order to meet format specifications. For example, some students force the temporarily assembled chapters to fit the framework in order to meet the journal format requirements, while traditional formal review only focuses on explicit elements such as figure numbers and reference formats, making it difficult to identify the gaps between data authenticity and research logic. (2) In terms of content review, the subjectivity of peer review makes it difficult to effectively identify the authenticity of the research process. When the paper is disconnected from daily research, the single-blind or double-blind review mechanism can reduce author identity bias, but it cannot trace the original scene of data generation. For example, some experimental data may be tampered with or fabricated to fit the conclusion, while traditional review mainly relies on expert experience and judgment, lacking dynamic verification of the research process. (3) If a graduate student fabricates the research process in a paper, the current review mechanism often only focuses on the formal statements at the time of submission, making it difficult to trace the actual behavior during the research cycle. For example, some students may over-modify data interpretations through editing services or even purchase ghostwriting services, but such behavior is difficult to detect under the traditional review framework, resulting in the long-term hidden risk of academic misconduct. (4) In terms of innovation assessment. When the content of a paper is disconnected from daily research, its innovation may come from literature patchwork rather than real research accumulation, but the current system relies on quantitative indicators such as citations and Altmetrics, making it difficult to distinguish between superficial innovation and substantive breakthroughs, resulting in the frequent occurrence of low-quality papers passing formal review. Altmetrics is a method for evaluating research influence, which aims to evaluate the influence of academic achievements by quantitatively analyzing the dissemination and citation of academic achievements on platforms such as social media, news media, and policy documents. It emerged to address the limitations of traditional academic evaluation indicators (such as impact factors) and to provide a more comprehensive, timely and diverse academic evaluation method.

[0071] Regarding methods for managing and controlling paper quality, there are many established and established approaches, as well as some emerging ones. For example, AI-powered collaborative review systems predict the risk of paper retraction by training a "paper quality profile" model and combining it with historical review data. This system has achieved high accuracy in practical applications. With the continuous development and improvement of the internet, blockchain technology is also being used to store and trace experimental data related to papers, ensuring that data cannot be tampered with by storing experimental data and revision records on-chain. Alternatively, smart contracts can automatically identify relationships between authors and funders to prevent conflicts of interest. Therefore, to accurately and comprehensively control the quality of submitted papers and achieve high-quality, high-quality results, it is essential to establish a paper submission review approach that combines process management with goal-based management. This integrated approach not only ensures the coherence and scientific integrity of papers from research to writing, but also improves the quality of papers from the very beginning, reducing the output of low-quality papers caused by chaotic research processes or unclear goals, and effectively promoting the development of high-quality academic research.

[0072] In the area of ​​paper quality control, emerging AI-powered collaborative review systems and blockchain technology have brought about numerous changes, but they also present certain drawbacks. AI-powered collaborative review systems rely on historical data to train their "paper quality profile" models. However, when encountering new research directions, there is little or no historical data available, leading to a high rate of misjudgment. While blockchain technology can ensure data immutability and store experimental data and revision records on-chain, it faces high storage costs and technical barriers, making it a costly and inefficient method for not all research teams and editorial departments.

[0073] Compared with these, the method of combining process management and target management proposed in this application has obvious advantages. From the perspective of process management, by tracing the research duration through the time data of the journal system, verifying the research location with the help of spatial information, and supervising the research process in all aspects, it is possible to avoid the temporary patchwork of research results. This is a deep control of the research process that AI and blockchain technology cannot achieve. In terms of target management, a comprehensive evaluation model covering quality and quantity dimensions is constructed, and papers are evaluated from multiple perspectives in combination with expert reviews. It can not only judge the current quality of the paper, but also pay attention to its consistency with daily research accumulation, and tap the value of long-term research precipitation. This comprehensive, in-depth and flexible method that can be adapted to all types of papers will not be limited by missing data or technical difficulties. It effectively guarantees the quality of papers from the source and provides solid support for the healthy development of academic research.

[0074] Based on this, in order to solve the problem that the existing AI paper quality portrait method relies on historical data to train the "paper quality portrait" model, if encountering an emerging research direction, there is too little or no historical data for the model to refer to, and the rationality and effectiveness of the paper review score cannot be guaranteed. In addition, in order to solve the problems of high equipment resource occupancy, high storage costs and technical thresholds faced by the blockchain method, the embodiments of the present application respectively provide a paper review method based on multidimensional data fusion, a paper review device based on multidimensional data fusion for executing the paper review method based on multidimensional data fusion, an electronic device, a computer-readable storage medium and a computer program product, which can improve the accuracy of the temporary patchwork judgment of the paper review, thereby effectively reducing the resource occupancy and storage cost of the equipment performing the paper review, and can improve the rationality and effectiveness of the paper review score.

[0075] The details are described in detail through the following examples.

[0076] Based on this, the embodiment of the present application provides a paper review method based on multidimensional data fusion that can be implemented by a paper review device based on multidimensional data fusion, see Figure 1 The paper review method based on multidimensional data fusion specifically includes the following contents:

[0077] Step 100: From a software system for storing research record data and check-in information periodically submitted by users, extract research record data that matches the current target paper as target record data, and based on the target record data and the paper text data corresponding to the target paper, obtain the time dimension score corresponding to the target paper for representing the time distribution of the paper research, the spatial variable for representing the spatial distribution of the paper research, and the text correlation coefficient for representing the similarity between the paper text data and the target record data.

[0078] The software system is used to store research record data and check-in information periodically submitted by users. The period for submitting research record data can be set according to actual application requirements, for example, weekly, monthly, or quarterly. If the period is monthly, the research record data is a monthly report. In one example of this application, the software system can adopt a graduate student monthly report system. The specific functions are detailed in subsequent application examples and will not be repeated here.

[0079] The target paper refers to the academic paper currently awaiting review.

[0080] The time dimension score is used to indicate the research time distribution of the paper, which can be represented by S; the score can be from 0 to 10, the higher the score, the more the target paper tends to be a real research result; the lower the score, the more the target paper tends to be a temporary patchwork.

[0081] The spatial variable is used to represent the spatial distribution of the research paper and can be represented by D; the spatial variable can be represented in a periodic coding manner.

[0082] The text relevance coefficient is used to represent the similarity between the paper text data and the target record data, which can be represented by k and can be specifically calculated using cosine similarity.

[0083] And, step 200: based on natural language processing technology, determine the content quality feature vector corresponding to the target paper for representing the content quality of the paper according to the paper text data, and determine the text quantity index corresponding to the target paper for representing the proportion of quantitative dimension information in the paper according to the paper text data.

[0084] In step 200, natural language processing (NLP) technology refers to a deep learning model built through large-scale pre-training and self-supervised learning technology, which aims to improve the computer's ability to understand and generate natural language.

[0085] It can be understood that the content quality feature vector can be represented by Q; the text quantity index can be represented by N, and the quantity dimension information can include information such as the amount of data corresponding to the target paper, the number of experiments, and the number of cited documents.

[0086] Step 300: Input the time dimension score, the text relevance coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper into a multidimensional data fusion paper review model based on the interactive attention mechanism, so that the paper review model outputs the paper review score corresponding to the target paper.

[0087] Interactive Attention is an advanced attention mechanism used to exchange information and enhance feature representation across modalities. This mechanism enables models to exchange information and enhance the learning of key features between different types of data streams, such as text and images, audio and video, when processing multimodal data. Interactive Attention is designed to address the limitation of traditional attention mechanisms, which may overlook cross-modal interactions when processing complex multimodal data.

[0088] The core advantage of interactive attention lies in its ability to deeply connect and collaborate across different information sources (i.e., the temporal score S, the text relevance coefficient k, the text quantity indicator N, the spatial variable D, and the content quality feature vector Q corresponding to the target paper). Through this connection, the model can capture richer and more useful information across different modalities, thereby improving overall task performance.

[0089] From the above description, it can be seen that the paper review method based on multidimensional data fusion provided by the embodiment of the present application extracts the research record data periodically submitted by users that matches the current target paper as target record data through a software system for storing research record data and check-in information periodically submitted by users, and obtains the time dimension score, spatial variable and text correlation coefficient corresponding to the target paper based on the target record data and the paper text data corresponding to the target paper, which can effectively trace the user's daily paper research time, research space distribution and similarity, and can effectively improve the accuracy of identifying whether the paper content is temporarily pieced together; and, through By automatically scoring the paper review score by using the time dimension score S, the text correlation coefficient k, the text quantity index N, the spatial variable D and the content quality feature vector Q corresponding to the target paper as multidimensional data, the rationality and effectiveness of the paper review score can be improved; that is, the present application can improve the accuracy of the temporary patchwork judgment of the paper review without adopting blockchain technology, thereby effectively reducing the resource occupancy rate and storage cost of the equipment performing the paper review and lowering the technical threshold; and compared with the method of using AI paper quality profiling, it can effectively improve the rationality and effectiveness of the paper review score and reduce the misjudgment rate.

[0090] In order to further improve the effectiveness and reliability of extracting research record data that matches the current target paper as target record data, in a paper review method based on multidimensional data fusion provided in an embodiment of the present application, see Figure 2 , step 100 in the paper review method based on multidimensional data fusion specifically includes the following contents:

[0091] Step 110: In a software system for storing research record data and check-in information periodically submitted by users, extract each research record data periodically submitted by the user who is the author of the target paper.

[0092] Step 120: Extract the paper text data corresponding to the target paper and the keywords corresponding to each of the research record data respectively, and calculate the frequency and inverse document frequency of each of the keywords in the paper text data or the research record data.

[0093] Step 130: Based on the frequencies and inverse document frequencies corresponding to the keywords, a word frequency inverse document frequency algorithm is used to search for research record data that matches the keywords of the target paper in each research record data as target record data.

[0094] The Term Frequency-Inverse Document Frequency (TF-IDF) algorithm is commonly used in information retrieval and text mining. Its core idea is to calculate the importance of a word in a document in order to rank and recommend documents in applications such as search engines.

[0095] In order to further improve the effectiveness and reliability of obtaining time dimension scores, in a paper review method based on multidimensional data fusion provided in an embodiment of the present application, see Figure 2 , step 100 in the paper review method based on multidimensional data fusion further specifically includes the following contents:

[0096] Step 140: Extract the corresponding submission time from the software system to obtain the submission time period corresponding to all the target record data; determine the time dimension score corresponding to the target paper for representing the research time distribution of the paper based on the submission time period and a preset experience threshold.

[0097] Specifically, the time dimension score S can be calculated according to formula (1):

[0098] Formula (1)

[0099] in, It refers to the scoring function. After solving the scoring function, the value of the scoring function is used as the time dimension score S; The submission time period corresponding to all the target record data, that is, the time period between the first time the target record data is detected in the software system and the last time the target record data is detected; is the empirical threshold, and this application takes sky, Day. In formula (1), when When (that is, completed instantly, extreme temporary patchwork), divide; when hour, divide; when hour, 10 points.

[0100] Step 150: Based on the target record data and the paper text data corresponding to the target paper, obtain the spatial variables corresponding to the target paper for representing the spatial distribution of paper research.

[0101] And, step 160: according to the target record data and the paper text data corresponding to the target paper, obtain the text correlation coefficient corresponding to the target paper for indicating the similarity between the paper text data and the target record data.

[0102] In order to further improve the effectiveness and reliability of obtaining spatial variables, in a paper review method based on multidimensional data fusion provided in an embodiment of the present application, see Figure 3 , step 150 in the paper review method based on multidimensional data fusion specifically includes the following contents:

[0103] Step 151: Obtain the check-in information corresponding to each target record data submitted by the user who is the author of the target paper in the software system, and determine the geographical location information corresponding to each check-in information according to the geographic information system.

[0104] Step 152: Periodically encode each of the geographic location information to obtain a spatial variable corresponding to the target paper for representing the spatial distribution of the paper research.

[0105] The core purpose of periodic coding is to convert periodic features into a form recognizable by the model through mathematical methods (such as sine or cosine functions) to capture repetitive patterns and multi-scale location information.

[0106] In order to further improve the effectiveness and reliability of obtaining the text relevance coefficient, in the paper review method based on multidimensional data fusion provided in the embodiment of the present application, see Figure 3 Step 160 of the paper review method based on multidimensional data fusion specifically includes the following contents:

[0107] Step 161: Generate summary text data corresponding to the target record data based on the large language model.

[0108] Step 162: Input the summary text data and the paper text data into the BERT model respectively, so that the BERT model outputs a summary feature vector corresponding to the summary text data and a paper feature vector corresponding to the paper text data respectively.

[0109] The BERT model uses the bidirectional self-attention mechanism of the Transformer encoder to capture left and right context information simultaneously in all layers to achieve richer semantic representation.

[0110] Step 163: Calculate the similarity between the abstract feature vector and the paper feature vector using cosine similarity to serve as a text correlation coefficient corresponding to the target paper for representing the similarity between the paper text data and the target record data.

[0111] In order to further improve the effectiveness and reliability of obtaining content quality feature vectors, in a paper review method based on multidimensional data fusion provided in an embodiment of the present application, see Figure 3 Step 200 of the paper review method based on multidimensional data fusion specifically includes the following contents:

[0112] Step 210: Preprocess the paper text data; wherein the preprocessing includes: removing noise, sentence segmentation and word segmentation.

[0113] Step 220: Based on the corresponding word segmenter of natural language processing technology, obtain the word embedding vector sequence corresponding to the preprocessed paper text data.

[0114] Step 230: Linearly map the hidden feature vectors corresponding to the word embedding vector sequence to obtain a content quality feature vector corresponding to the target paper for representing the content quality of the paper.

[0115] Specifically, the SciBERT model can be used for deep semantic encoding. The input paper text is normalized, including noise removal (such as extra spaces and punctuation), sentence segmentation, and word segmentation. SciBERT's built-in tokenizer is then used to convert the text into a subword sequence, ensuring that the input format meets the model's requirements. The SciBERT tokenizer generates a vocabulary sequence, which is then converted into a corresponding word embedding vector sequence.

[0116] The SciBERT model is a pre-trained model based on the BERT model, specifically for scientific text processing. It improves performance in natural language processing tasks in the scientific field by pre-training on a large amount of scientific literature.

[0117] In order to further improve the effectiveness and reliability of obtaining the text quantity index, in the paper review method based on multidimensional data fusion provided in the embodiment of the present application, see Figure 3 Step 200 of the paper review method based on multidimensional data fusion further specifically includes the following contents:

[0118] Step 240: Obtain the data volume, number of experiments, and number of cited documents corresponding to the paper text data.

[0119] Step 250: Standardize the data volume, number of experiments, and number of cited documents respectively to obtain a first standard value corresponding to the data volume, a second standard value corresponding to the number of experiments, and a third standard value corresponding to the number of cited documents.

[0120] Specifically, the first standard value corresponding to the data volume Z The calculation formula is shown in formula (2):

[0121] Formula (2)

[0122] in, Refers to the mean value corresponding to the data volume, Refers to the standard deviation corresponding to the data volume;

[0123] The second standard value corresponding to the number of experiments E The calculation formula is shown in formula (2):

[0124] Formula (3)

[0125] in, refers to the mean corresponding to the number of experiments, Refers to the standard deviation corresponding to the number of experiments;

[0126] Number of cited documents Corresponding third standard value The calculation formula is shown in formula (4):

[0127] Formula (4)

[0128] in, Refers to the mean value corresponding to the number of cited documents, Refers to the standard deviation corresponding to the number of cited documents.

[0129] Step 260: Calculate the original quantity index corresponding to the paper text data based on the first standard value, the second standard value, the second standard value and their respective corresponding weights.

[0130] Specifically, the original quantitative indicators corresponding to the paper text data The calculation formula is shown in formula (5):

[0131] Formula (5)

[0132] The weight of the data volume is , the weight of the number of experiments is , the weight of the number of cited documents is ,and .

[0133] Step 270: Perform a linear transformation on the original quantity index to convert it to the interval [0, 10], and obtain a text quantity index corresponding to the target paper, which is used to represent the proportion of quantity dimension information in the paper.

[0134] The calculation formula of the text quantity index N is shown in formula (6):

[0135] Formula (6)

[0136] Among them, the original quantity index The value range is ; The linear transformation formula converts the score to the interval [0,10] to obtain the value of N.

[0137] In order to further improve the effectiveness and reliability of the paper review based on multidimensional data fusion, in the paper review method based on multidimensional data fusion provided in the embodiment of the present application, see Figure 4 The multidimensional data fusion paper review model based on the interactive attention mechanism in the paper review method based on multidimensional data fusion specifically includes the following contents:

[0138] A scalar embedding layer is used to perform scalar embedding on the time dimension score S, the text relevance coefficient k, and the text quantity index N corresponding to the target paper to obtain a time feature vector S1 corresponding to the time dimension score, a text relevance feature vector k1 corresponding to the text relevance coefficient, and a text quantity feature vector N1 corresponding to the text quantity index;

[0139] A vector projection layer, configured to perform vector projection on the spatial variable D and the content quality feature vector Q corresponding to the target paper, so as to obtain a projected spatial variable D1 corresponding to the spatial variable and a projected content quality feature vector Q1 corresponding to the content quality feature vector;

[0140] An input sequence construction layer, configured to construct a corresponding input sequence according to the time feature vector S1, the text relevance feature vector k1, the text quantity feature vector N1, the projected spatial variable D1, and the projected content quality feature vector Q1;

[0141] A spatial and content cross attention layer is used to use the projected content quality feature vector Q1 in the input sequence as a query vector and the projected spatial variable D1 as a key-value pair vector to perform cross attention calculation to obtain a corresponding spatial and content fusion feature vector D_Q;

[0142] A quantity and quality cross attention layer is used to concatenate the projected content quality feature vector Q1 in the input sequence with the spatial and content fusion feature vector D_Q to obtain a corresponding first concatenation vector, which can be: Q+D_Q; use the first concatenation vector Q+D_Q as a query vector, and use the text quantity feature vector N1 in the input sequence as a key-value pair vector to perform cross attention calculation to obtain a corresponding quantity and quality fusion feature vector N_Q;

[0143] a vector concatenation layer, configured to concatenate the projected content quality feature vector Q1, the projected spatial variable D1, and the quantity and quality fusion feature vector N_Q in the input sequence to obtain a corresponding second concatenation vector;

[0144] A dynamic fusion layer, configured to dynamically fuse the second concatenated vector using the text relevance feature vector k1 in the input sequence as a modulation factor;

[0145] A three-dimensional attention routing layer is used to obtain a target output representation corresponding to the target paper based on the self-interaction result data between the time feature vector S1, the text relevance feature vector k1 and the text quantity feature vector N1, the self-interaction result data between the projected content quality feature vector Q1 and the projected spatial variable D1, and the second splicing vector after dynamic fusion;

[0146] A linear layer is used to map the target output representation to a paper review score corresponding to the target paper.

[0147] From the software level, this application also provides a paper review device based on multidimensional data fusion for executing all or part of the paper review method based on multidimensional data fusion, see Figure 5 The paper review device based on multidimensional data fusion specifically includes the following contents:

[0148] The first multidimensional data acquisition module 10 is used to extract research record data matching the current target paper from a software system for storing research record data and check-in information periodically submitted by users as target record data, and obtain the time dimension score S corresponding to the target paper for representing the time distribution of paper research, the spatial variable D for representing the spatial distribution of paper research, and the text correlation coefficient k for representing the similarity between the paper text data and the target record data based on the target record data and the paper text data corresponding to the target paper.

[0149] In addition, the second multidimensional data acquisition module 20 is used to determine the content quality feature vector Q corresponding to the target paper for representing the content quality of the paper based on the paper text data based on natural language processing technology, and to determine the text quantity index N corresponding to the target paper for representing the proportion of quantitative dimension information in the paper based on the paper text data.

[0150] The paper review scoring module 30 is used to input the time dimension score S, the text correlation coefficient k, the text quantity index N, the spatial variable D and the content quality feature vector Q corresponding to the target paper into a multidimensional data fusion paper review model based on the interactive attention mechanism, so that the paper review model outputs the paper review score corresponding to the target paper.

[0151] The embodiment of the paper review device based on multidimensional data fusion provided in this application can be specifically used to execute the processing flow of the embodiment of the paper review method based on multidimensional data fusion in the above embodiment. Its functions will not be repeated here, and reference can be made to the detailed description of the above embodiment of the paper review method based on multidimensional data fusion.

[0152] The part of the paper review based on multidimensional data fusion performed by the paper review device based on multidimensional data fusion can be completed in the server or client device. Specifically, it can be selected based on the processing power of the client device and the limitations of the user's usage scenario. This application is not limited to this. If all operations are completed in the client device, the client device may also include a processor for specific processing of the paper review based on multidimensional data fusion.

[0153] The aforementioned client device may include a communication module (i.e., a communication unit) capable of establishing a communication connection with a remote server to facilitate data transmission with the server. The server may include a server at the task scheduling center or, in other implementation scenarios, a server on an intermediate platform, such as a server on a third-party server platform that is communicatively linked to the task scheduling center server. The server may comprise a single computer device, a server cluster consisting of multiple servers, or a distributed server configuration.

[0154] The server and the client device may communicate using any suitable network protocol, including network protocols that have not yet been developed as of the filing date of this application. Examples of such network protocols include TCP / IP, UDP / IP, HTTP, and HTTPS. Furthermore, examples of such network protocols include RPC (Remote Procedure Call Protocol) and REST (Representational State Transfer) protocols, which are used on top of the aforementioned protocols.

[0155] From the above description, it can be seen that the paper review device based on multidimensional data fusion provided by the embodiment of the present application extracts the research record data periodically submitted by users that matches the current target paper as target record data through a software system for storing research record data and check-in information periodically submitted by users, and obtains the time dimension score, spatial variable and text correlation coefficient corresponding to the target paper according to the target record data and the paper text data corresponding to the target paper, which can effectively trace the user's daily paper research time, research space distribution and similarity, and can effectively improve the recognition accuracy of whether the paper content is temporarily pieced together; and, through By automatically scoring the paper review score by using the time dimension score S, the text correlation coefficient k, the text quantity index N, the spatial variable D and the content quality feature vector Q corresponding to the target paper as multidimensional data, the rationality and effectiveness of the paper review score can be improved; that is, the present application can improve the accuracy of the temporary patchwork judgment of the paper review without adopting blockchain technology, thereby effectively reducing the resource occupancy rate and storage cost of the equipment performing the paper review and lowering the technical threshold; and compared with the method of using AI paper quality profiling, it can effectively improve the rationality and effectiveness of the paper review score and reduce the misjudgment rate.

[0156] To further illustrate the above embodiment, this application also provides a specific application example of a paper review method based on multidimensional data fusion. In this application example, the software system is a graduate student monthly report system (which can be simply referred to as a monthly report system or campus management system, etc.) as an example. Correspondingly, the research record data can be expressed as a monthly report. Based on this, the application example of this application specifically includes the following content:

[0157] (1) Time data tracing based on the graduate monthly report system, and scoring function based on the time of the graduate monthly report system Get the time dimension score S:

[0158] Using the time dimension data such as the submission time and content update frequency of the graduate monthly report recorded by the graduate monthly report system, the term frequency-inverse document frequency (TF-IDF) method is used to extract keywords from the monthly reports and papers. First, the frequency of the keywords appearing in the text is calculated. :

[0159]

[0160] in, Indicates keywords In the documentation The number of times it appears in Represents the index variable, which is used to traverse all the different words that appear in document j; Indicates the number of occurrences of all different words; Indicates keywords In the documentation The frequency of occurrence in .

[0161] Then calculate the prevalence of the keyword, that is, the inverse document frequency :

[0162]

[0163] in, Indicates the number of all documents, Indicates that it contains keywords If there are fewer documents containing the i-th keyword, The larger the value is, the better the category differentiation ability of the entry is.

[0164] A high word frequency within a particular document, combined with a low word frequency across the entire document collection, can result in a high-weighted TF-IDF:

[0165]

[0166] in, Indicates keywords In the documentation TF - IDF value in; Indicates keywords In the documentation The frequency of occurrence in Represents inverse document frequency.

[0167] The TF-IDF method can be used to extract keywords from monthly reports and papers, compare relevant keywords, extract relevant monthly reports with a similarity of more than 80% with the paper keywords, and retrieve the monthly report submission time from the background.

[0168] Scoring function based on the time of the graduate monthly report system, depending on whether the content is temporarily assembled or accumulated over a long period of time The time dimension score S is calculated as shown in the above formula (1).

[0169] Based on the keyword matching results, analyze the temporal distribution of the relevant research content in the paper. Determine whether the research is a sudden, improvisation, or the result of long-term, continuous accumulation and gradual improvement. Assign a score of 0-10 to the paper accordingly.

[0170] (2) Integrate the spatial information verification of laboratories and other institutions to construct the spatial variable D:

[0171] Using the Geographic Information System (GIS) and the university's internal management system, combined with spatial information such as the Graduate School and laboratory's site usage records and equipment loan registrations, we verify whether the research work was conducted at the relevant locations of the paper submitter. Furthermore, for collaborative projects, we review the cooperation agreement and communication records to confirm the authenticity and level of participation in the collaboration, thereby demonstrating the close connection between the research work and the paper submitter. A Geographic Information System (GIS) is a computer system used to collect, store, manage, analyze, and display geospatial data.

[0172] The graduate monthly report system requires students to check in regularly. The IP location information of each check-in can be obtained. The latitude and longitude information (x, y) collected by the GIS system is used as input. Based on the geographic location information of the student's check-in, this application can use a periodic code to represent it:

[0173] Suppose there is a continuous space variable In order to capture its periodic information, this application predefines a set of frequency parameters , where each , , where Initial frequency.

[0174] but The periodic code vector It can be defined as:

[0175]

[0176] For two-dimensional space variables , we can periodically encode x and y separately, and then concatenate the two encoded vectors:

[0177]

[0178] This formal expression uses sine and cosine functions to map continuous spatial variables into a high-dimensional periodic feature space, which can preserve the periodic structure of the original spatial data and facilitate its use as part of the feature vector in subsequent automated scoring models.

[0179] (3) Use text analysis technology to evaluate content quality and construct content quality feature vectors :

[0180] Use Natural Language Processing (NLP) technology to conduct in-depth analysis of the paper text. Natural Language Processing (NLP) is an important branch of artificial intelligence and computer science that focuses on how to enable computers to understand and process human natural language.

[0181] The SciBERT model is used for deep semantic encoding. The input paper text is normalized, including noise removal (such as extra spaces and punctuation cleaning), sentence segmentation, and word segmentation. SciBERT's built-in word segmenter is then used to convert the text into a subword sequence, ensuring that the input format meets the model's requirements. Let the preprocessed text be T. After passing it through the SciBERT word segmenter, the resulting word sequence is converted into a corresponding word embedding vector sequence. After SciBERT model processing, the hidden layer representation corresponding to the output [CLS] tag is denoted as:

[0182]

[0183] in, represents the feature vector representing the classification head; Represents a real number vector with a dimension of 768, that is, this vector contains 768 real number elements

[0184] Then, through a linear mapping:

[0185]

[0186] Where W represents the weight matrix, which is used to transfer and transform information between different layers of the SciBERT model; It means W is a real matrix with shape of 256 rows and 768 columns.

[0187] The 768-dimensional vector is mapped to 256 dimensions to obtain the feature vector Q representing the content quality.

[0188] (4) Construct a comprehensive quality and quantity evaluation model and construct a text quantity index :

[0189] Based on the results of text analysis, we construct a comprehensive evaluation model, combining quantitative information such as the amount of data involved, the number of experiments, and the number of citations in the papers. By setting appropriate weights, we quantitatively evaluate the quality and quantity of the papers to comprehensively measure their performance at the data foundation level.

[0190] Assume that the amount of data involved in the paper is , the number of experiments is , the number of cited papers is First, each indicator is standardized to eliminate the dimensional effects between different indicators, as shown in the above formulas (2) to (4).

[0191] in, 、 、 are the mean of the amount of data, number of experiments, and number of cited literature, respectively. 、 、 are their standard deviations respectively. Assign weights to each indicator, and the weights are determined according to the characteristics and importance of the research field. If the research field is relatively new and the number of references is small, then the weight of the data volume should be appropriately set to , the weight of the number of experiments is , the weight of the number of cited documents is ,and .

[0192] Original number of papers It can be obtained by the above formula (5).

[0193] Assume that the original quality score calculated by the above formula (5) is The value range is ,in, and According to experience, the values ​​are set to -5 and 5 respectively, and the linear transformation formula is used to convert the score to the interval [0, 10], which is the text quantity index. Expressed as:

[0194]

[0195] (5) Compare the contents of the monthly reports and conduct correlation analysis to construct the text correlation coefficient k:

[0196] To quantify the relevance between papers and their corresponding monthly reports, this paper proposes an automated method based on large-scale model summary generation and BERT series model feature extraction. The specific steps are as follows:

[0197] In order to reduce redundant information and capture the core content of the monthly report, a large-scale pre-trained language model (such as GPT-4 or T5) is first used to train the original monthly report text. Generate summary and get summary text ,This step aims to extract key information reflecting the research focus in the ,monthly report and provide refined input for subsequent feature extraction.

[0198] Abstract and the full text of the paper Input into the pre-trained BERT series model and use its deep semantic encoding ability to extract the corresponding feature representation. Assume that the feature vectors obtained by the BERT model are:

[0199]

[0200]

[0201] in, Abstract text The eigenvector of The eigenvector of the full text T.

[0202] The cosine similarity is used to calculate the similarity between two feature vectors to measure the degree of relevance between the monthly report abstract and the paper content. The formula is:

[0203]

[0204] in, is the feature vector of the monthly report summary, is the feature vector of the paper text, That is, the text correlation coefficient, whose value range is [0,1]. The closer the value is to 1, the higher the correlation between the two contents.

[0205] (6) The final conclusion is drawn from the comprehensive multi-dimensional review results:

[0206] By integrating the time dimension, space dimension, data foundation and other information obtained in the previous five steps, this application can build an end-to-end model, namely a multi-dimensional data fusion paper review model based on the interactive attention mechanism. The system of this application has a large amount of data on the scores of papers by experts. This application uses this data as supervision to train the multi-dimensional data fusion paper review model based on the interactive attention mechanism of this application. The specific results of the multi-dimensional data fusion paper review model based on the interactive attention mechanism are as follows: Figure 6 shown.

[0207] The input of the multi-dimensional data fusion paper review model based on the interactive attention mechanism includes:

[0208] 1. Time dimension score S: scalar (0-10 points);

[0209] 2. Spatial variable D: a high-dimensional vector after periodic encoding (the dimension depends on the number of frequency parameters, and is 256 when n=64);

[0210] 3. Content quality feature vector Q: 256-dimensional semantic vector;

[0211] 4. Text quantity index N: scalar (0-10 points);

[0212] 5. Text relevance coefficient k: scalar (0-1).

[0213] In this model, the five input features, namely the temporal dimension score S, spatial variable D, content quality feature Q, quantity indicator N, and text relevance k, are first scalar-embedded or vector-projected, and then concatenated with the [CLS] tag to form a unified input sequence. Next, the model uses "spatial and content cross attention" to use Q1 as the query vector (Query) and D1 as the key-value pair vector (Key / Value) to learn the dependency between content quality and spatial variables, obtaining the fused D_Q. Subsequently, "quantity and quality cross attention" is used to further introduce the quantity indicator N_Q using Q+D_Q as the query and N as the key / value. The original D1, Q1, and N_Q are then concatenated and dynamically fused using the text relevance k as a modulation factor to balance the importance of spatiotemporal and quantitative features. Finally, the model uses "three-dimensional attention routing" to simultaneously consider the interactions between scalar features (S1, N1, k1), the interactions between vector features (D1, Q1), and the interactions across scalars and vectors, and comprehensively obtains the target output representation that can be used for downstream tasks. Finally, this application passes the [CLS] tag through a linear layer and maps it into a paper review score. The [CLS] tag is used to represent the semantic information of the entire text.

[0214] In addition, in order to further improve the application effectiveness and reliability of the above application examples, the use process of the graduate student monthly report system can be found in Figure 7 , among which, the process for students to submit monthly or annual report information is as follows Figure 8 As shown in the figure, the process of the management end reviewing the monthly or annual report information submitted by students is as follows Figure 9 shown.

[0215] Based on this, the application example of this application involves a method for reviewing paper submissions based on multidimensional data fusion based on NLP. It is a method for macro-quality control of submitted papers by combining AI algorithms, involving the fields of graduate student training and submission management, and adopts the following steps: (1) Based on the time data tracing of the graduate student monthly report system, a scoring function S(T) based on the time of the graduate student monthly report system is generated; (2) The spatial information verification of laboratories and other institutions is integrated to construct a spatial variable D; (3) Text analysis technology is used to evaluate content quality and a content quality evaluation index Q is constructed; (4) A comprehensive quality and quantity evaluation model is constructed to construct a text quantity index N; (5) Correlation analysis is performed on the monthly report content to construct a text correlation coefficient k; (6) An expert review and feedback mechanism is introduced to construct an expert evaluation index P; (7) The final conclusion is drawn based on the comprehensive multi-dimensional review results. This application proposes a method for graduate thesis quality management that combines process management with target management. It solves the problem that traditional thesis review methods focus more on the finished paper and pay insufficient attention to the research process. It uses natural language processing technology and text similarity algorithms to link the paper with the author's daily research accumulation, ensuring that the research is based on a solid process and is conducive to discovering truly valuable and high-quality papers.

[0216] In other words, the effects of the above application examples of this application are as follows:

[0217] (1) The currently used paper review methods mostly focus on the finished paper and pay insufficient attention to the research process. However, the new method can determine whether the research is a temporary patchwork or a long-term accumulation by tracing the time data of the journal system; with the help of institutional spatial information verification, the research location and cooperation situation can be verified. This all-round and multi-dimensional monitoring of the research process makes up for the defect of the traditional method that only focuses on the results, guarantees the quality of the paper from the source, and ensures that the research is based on a solid process rather than a hasty patchwork at the end.

[0218] (2) Common review methods rely heavily on subjective expert judgment to assess paper content, lacking precise quantification. The new method uses text analysis technology to assess basic quality and builds a comprehensive evaluation model based on quantitative dimensions such as data volume and number of experiments. It considers both the quality dimensions of the paper, such as vocabulary and logic, and the quantitative dimensions, achieving quantitative evaluation through reasonable weight setting. In comparison, this method is more scientific and objective, reduces evaluation bias caused by subjective factors, and can more accurately measure the performance of the paper at the data foundation level.

[0219] (3) Traditional review methods rarely link papers with the authors’ daily research accumulation. The new method compares daily accumulated content for correlation analysis and uses a text similarity algorithm to determine whether the paper integrates daily research ideas and results, ensuring the accumulation and coherence of the paper content. At the same time, expert review is introduced to evaluate from multiple perspectives such as academic value, and conclusions are drawn based on the results of multi-dimensional reviews. This makes the evaluation of papers more comprehensive and in-depth, focusing not only on current results but also on the accumulation of long-term research processes, which is conducive to discovering truly valuable and high-quality papers.

[0220] The present application also provides an electronic device, which may include a processor, a memory, a receiver, and a transmitter. The processor is configured to execute the paper review method based on multidimensional data fusion mentioned in the above embodiment. The processor and the memory may be connected via a bus or other means, with bus connection being used as an example. The receiver may be connected to the processor and the memory via a wired or wireless means.

[0221] The processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0222] The memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as the program instructions / modules corresponding to the paper review method based on multidimensional data fusion in the embodiments of the present application. The processor executes the various functional applications and data processing of the processor by running the non-transitory software programs, instructions, and modules stored in the memory, thereby implementing the paper review method based on multidimensional data fusion in the above method embodiments.

[0223] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include a memory remotely located relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0224] The one or more modules are stored in the memory, and when executed by the processor, perform the paper review method based on multidimensional data fusion in the embodiment.

[0225] In some embodiments of the present application, the user equipment may include a processor, a memory and a transceiver unit, and the transceiver unit may include a receiver and a transmitter. The processor, memory, receiver and transmitter may be connected through a bus system. The memory is used to store computer instructions, and the processor is used to execute the computer instructions stored in the memory to control the transceiver unit to send and receive signals.

[0226] As an implementation method, the functions of the receiver and transmitter in this application can be considered to be implemented through a transceiver circuit or a dedicated transceiver chip, and the processor can be considered to be implemented through a dedicated processing chip, a processing circuit or a general-purpose chip.

[0227] As another implementation method, it is possible to use a general-purpose computer to implement the server provided in the embodiments of the present application. That is, the program code for implementing the functions of the processor, receiver, and transmitter is stored in a memory, and the general-purpose processor implements the functions of the processor, receiver, and transmitter by executing the code in the memory.

[0228] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program implements the steps of the aforementioned paper review method based on multidimensional data fusion. The computer-readable storage medium can be a tangible storage medium, such as a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a floppy disk, a hard disk, a removable storage disk, a CD-ROM, or any other form of storage medium known in the art.

[0229] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the aforementioned paper review method based on multidimensional data fusion.

[0230] It should be understood by those skilled in the art that the various exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether it is implemented in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted on a transmission medium or communication link via a data signal carried in a carrier.

[0231] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0232] In this application, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or replace features of other embodiments.

[0233] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Those skilled in the art will appreciate that various modifications and variations of the present embodiment are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A paper review method based on multidimensional data fusion, characterized in that: include: Extracting research record data that matches the current target paper from a software system for storing research record data and punch-in information periodically submitted by users as target record data, and obtaining, based on the target record data and the paper text data corresponding to the target paper, a time dimension score corresponding to the target paper for representing the time distribution of paper research, a spatial variable for representing the spatial distribution of paper research, and a text correlation coefficient for representing the similarity between the paper text data and the target record data; Furthermore, based on natural language processing technology, a content quality feature vector corresponding to the target paper for representing the quality of the paper content is determined according to the paper text data, and a text quantity index corresponding to the target paper for representing the proportion of quantitative dimension information in the paper is determined according to the paper text data; The time dimension score, the text relevance coefficient, the text quantity index, the spatial variable and the content quality feature vector corresponding to the target paper are input into a multidimensional data fusion paper review model based on an interactive attention mechanism, so that the paper review model outputs a paper review score corresponding to the target paper.

2. The paper review method based on multidimensional data fusion according to claim 1 is characterized in that: Extracting the research record data that matches the current target paper as the target record data from the software system for storing the research record data and check-in information periodically submitted by the user includes: Extracting each research record data periodically submitted by the user who is the author of the target paper from a software system for storing research record data and check-in information periodically submitted by the user; Extracting the paper text data corresponding to the target paper and the keywords corresponding to each of the research record data respectively, and calculating the frequency and inverse document frequency of each of the keywords in the paper text data or the research record data; According to the frequencies and inverse document frequencies corresponding to the respective keywords, a word frequency inverse document frequency algorithm is used to search for research record data that matches the keywords in the target paper in the respective research record data to serve as target record data.

3. The paper review method based on multidimensional data fusion according to claim 1 is characterized in that: The method of obtaining, based on the target record data and the paper text data corresponding to the target paper, a time dimension score corresponding to the target paper for representing the time distribution of paper research, a spatial variable for representing the spatial distribution of paper research, and a text correlation coefficient for representing the similarity between the paper text data and the target record data, respectively, includes: Extracting the submission time corresponding to the target record data from the software system to obtain the submission time period corresponding to all the target record data; determining the time dimension score corresponding to the target paper for representing the research time distribution of the paper according to the submission time period and a preset empirical threshold; According to the target record data and the paper text data corresponding to the target paper, a spatial variable corresponding to the target paper is obtained for representing the spatial distribution of the paper research; Furthermore, based on the target record data and the paper text data corresponding to the target paper, a text correlation coefficient corresponding to the target paper is obtained, which is used to represent the similarity between the paper text data and the target record data.

4. The paper review method based on multidimensional data fusion according to claim 3 is characterized in that: The step of obtaining, based on the target record data and the paper text data corresponding to the target paper, a spatial variable corresponding to the target paper for representing the spatial distribution of paper research, includes: Obtaining the check-in information corresponding to each target record data submitted by the user who is the author of the target paper in the software system, and determining the geographical location information corresponding to each check-in information according to the geographic information system; The geographical location information is periodically encoded to obtain a spatial variable corresponding to the target paper for representing the spatial distribution of the paper research.

5. The paper review method based on multidimensional data fusion according to claim 3 is characterized in that: The step of obtaining, based on the target record data and the paper text data corresponding to the target paper, a text correlation coefficient corresponding to the target paper and used to represent the similarity between the paper text data and the target record data includes: Generate summary text data corresponding to the target record data based on the large language model; Inputting the summary text data and the paper text data into a BERT model respectively, so that the BERT model outputs a summary feature vector corresponding to the summary text data and a paper feature vector corresponding to the paper text data respectively; The cosine similarity is used to calculate the similarity between the abstract feature vector and the paper feature vector, so as to serve as a text correlation coefficient corresponding to the target paper and used to represent the similarity between the paper text data and the target record data.

6. The paper review method based on multidimensional data fusion according to claim 1 is characterized in that: The method of determining a content quality feature vector corresponding to the target paper and used to represent the content quality of the paper based on the paper text data based on the natural language processing technology includes: Preprocessing the paper text data; wherein the preprocessing includes: removing noise, sentence segmentation and word segmentation; Based on the corresponding word segmenter of natural language processing technology, obtain the word embedding vector sequence corresponding to the preprocessed paper text data; Linear mapping is performed on the hidden layer feature vectors corresponding to the word embedding vector sequence to obtain a content quality feature vector corresponding to the target paper for representing the content quality of the paper.

7. The paper review method based on multidimensional data fusion according to claim 1 is characterized in that: The determining, based on the paper text data, of a text quantity index corresponding to the target paper and used to represent the proportion of quantity dimension information in the paper includes: Obtain the data volume, number of experiments and number of cited literature corresponding to the paper text data; Standardizing the data volume, the number of experiments, and the number of cited documents, respectively, to obtain a first standard value corresponding to the data volume, a second standard value corresponding to the number of experiments, and a third standard value corresponding to the number of cited documents; Calculate the original quantity index corresponding to the paper text data according to the first standard value, the second standard value, the third standard value and their corresponding weights; The original quantity index is linearly transformed to convert it to the interval [0, 10] to obtain the text quantity index corresponding to the target paper, which is used to represent the proportion of quantity dimension information in the paper.

8. The paper review method based on multidimensional data fusion according to any one of claims 1 to 7, characterized in that: The multi-dimensional data fusion paper review model based on the interactive attention mechanism includes: A scalar embedding layer is used to perform scalar embedding on the time dimension score, the text relevance coefficient, and the text quantity index corresponding to the target paper to obtain a time feature vector corresponding to the time dimension score, a text relevance feature vector corresponding to the text relevance coefficient, and a text quantity feature vector corresponding to the text quantity index; A vector projection layer, configured to perform vector projection on the spatial variable and the content quality feature vector corresponding to the target paper to obtain a projected spatial variable corresponding to the spatial variable and a projected content quality feature vector corresponding to the content quality feature vector; An input sequence construction layer, configured to construct a corresponding input sequence according to the time feature vector, the text relevance feature vector, the text quantity feature vector, the projected spatial variable, and the projected content quality feature vector; a spatial and content cross attention layer, configured to use the projected content quality feature vector in the input sequence as a query vector and the projected spatial variable as a key-value pair vector to perform a cross attention calculation to obtain a corresponding spatial and content fusion feature vector; a quantity and quality cross attention layer, configured to concatenate the projected content quality feature vector in the input sequence with the spatial and content fusion feature vector to obtain a corresponding first concatenated vector; use the first concatenated vector as a query vector and the text quantity feature vector in the input sequence as a key-value pair vector to perform cross attention calculation to obtain a corresponding quantity and quality fusion feature vector; a vector splicing layer, configured to splice the projected content quality feature vector, the projected spatial variable, and the quantity and quality fusion feature vector in the input sequence to obtain a corresponding second splicing vector; a dynamic fusion layer, configured to dynamically fuse the second concatenated vector using the text relevance feature vector in the input sequence as a modulation factor; a three-dimensional attention routing layer, configured to obtain a target output representation corresponding to the target paper based on the self-interaction result data between the time feature vector, the text relevance feature vector, and the text quantity feature vector, the self-interaction result data between the projected content quality feature vector and the projected spatial variable, and the second splicing vector after dynamic fusion; A linear layer is used to map the target output representation to a paper review score corresponding to the target paper.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the paper review method based on multidimensional data fusion as described in any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for reviewing a paper based on multidimensional data fusion as described in any one of claims 1 to 8 is implemented.

Images