Structural analysis and scoring method for academic papers

Abstract

The invention discloses a structural analysis and scoring method for academic papers, and relates to the technical field of education evaluation. Comprising the following steps: collecting academic papers with academic authorization completed; acquiring attribute data and derivative data of the collected academic papers to form a data set A; performing defect checking and feature zooming on the data set A, and constructing a data training set B; performing multiple linear regression modeling on the data training set B to obtain an academic paper structure analysis and scoring model M; obtaining attribute data and derivative data of a to-be-tested academic paper to form a to-be-tested data set C; performing defect checking and feature zooming on the to-be-tested data set C, and constructing a to-be-tested evaluation data set D; and calculating the to-be-tested evaluation data set D by using the academic paper structure analysis and scoring model M to obtain a to-be-tested academic paper structure score, and giving reasonable suggestions of each structure. By analyzing and modeling the completed academic papers, structural scores of the academic papers to be tested and reasonable suggestions of each structure are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of educational evaluation, and in particular to a method for structural analysis and scoring of degree theses. Background Art

[0002] A dissertation is a research report or scientific paper written by the author to obtain a specific degree. It not only represents different academic levels, but is also one of the important sources of literature. Dissertations are usually not published in academic journals, but are mainly obtained through degree-granting institutions, specific collection institutions and private channels. Dissertations are the core link of higher education, and they undertake the important task of cultivating students' innovation ability from "0 to 1", the ability to propose and solve problems, critical thinking ability and collaborative communication ability. They are the most critical link in achieving the transformation from "learning knowledge" to "strengthening ability". In addition, dissertations are also an important observation point for higher education evaluation, and their quality reflects the overall quality of graduates to some extent. To this end, my country has specially issued a series of standards to regulate dissertations, such as the rules for writing dissertations (GB / T7713.1-2006), the format for writing scientific and technological reports, dissertations and academic papers (GB / T7713-1987), and the rules for citing references after the text (GB / T 7714-2005). In January 2014, the "Measures for the Random Inspection of Doctoral and Master's Degree Theses" (Degree

[2014] No. 5) was issued, and in January 2021, the Ministry of Education issued the "Measures for the Random Inspection of Undergraduate Graduation Theses (Designs) (Trial)" (Teaching Supervision

[2020] No. 5); the sampling inspection of undergraduate graduation theses focuses on the "qualification" inspection of the significance of the topic, writing arrangement, logical construction, professional ability and academic norms, and focuses on the basic academic norms and basic academic literacy of undergraduates. The intrinsic quality of a degree thesis includes: academic innovation, the significance of the thesis, the logic of the thesis, etc., and the external quality includes: formal structure, format standardization, etc. The evaluation of a degree thesis is a complex process with multiple dimensions and steps. It is usually conducted by a committee composed of instructors, peer scholars and several experts to conduct quality evaluation from the aspects of academic value, methodology and innovation.

[0003] Standardization, as an external basis for measuring the quality of degree theses, teachers will spend a lot of time checking the format of students' papers, resulting in the phenomenon of "focusing on format and neglecting content". In October 2010, "Computer Engineering and Design" published "Research and Application of Document Format Checking Technology Based on Java", which studied the technology of using Java to check the format of graduation thesis, realized the format detection and matching of graduation thesis format with standard documents, and generated a detailed format check report to improve the efficiency of graduation thesis format checking. Document CN110069785A discloses a paper standard control and analysis platform and system based on component conventions, which can automatically process the format of papers according to the paper format requirements of different colleges, disciplines, and majors using component conventions, solving the problems of low efficiency, high randomness, and high error rate of traditional manual editing. In 2016, the journal S of "Higher Education Exploration" published "Construction of Undergraduate Thesis Quality Evaluation System Based on Academic Norms". In response to the fact that academic irregularities were rarely considered in undergraduate quality evaluation in some evaluation studies in the past, academic norms indicators were introduced, and a set of undergraduate thesis evaluation methods was constructed, which consists of a three-level indicator system: 1. Target layer (highest level): refers to the predetermined goal of the problem; 2. Criteria layer (middle level): refers to the criteria that affect the achievement of the goal, with 5 indicators, including thesis topic selection and review, basic knowledge and theoretical application, ability level, comprehensive quality of thesis, and academic norms; 3. Measures layer (lowest level): refers to measures to promote the achievement of goals, with 19 secondary indicators, including compliance with professional training direction, appropriateness and operability of topic selection, basic knowledge, etc. In 2019, the second issue of "Journal of Harbin Institute of Technology" published "Evaluation of the Quality of Undergraduate Graduation Thesis and Its Improvement Path", which analyzed and processed the data of thesis quality of 297 graduates from the School of Economics of Anhui University of Finance and Economics from 2007 to 2017. It was inferred that the main reason for the low quality of undergraduate graduation thesis is that general education has not played its due role, and suggested: formulate scientific general education training goals and arrange general education courses reasonably; improve students' logical thinking ability through general education; improve students' innovation ability through general education. In March 2019, "Beijing Social Sciences" published "Normative Evaluation of Highly Cited Doctoral Dissertations in Higher Education in China", pointing out that there are prominent problems such as uncoordinated proportion of the main body structure of the thesis, lack of prominent problem awareness of literature review, poor awareness of research design and methods, confusion between the significance and problems of research, and lack of research ethics norms. Academic normative education should be strengthened, research method course setting should be further improved, training in educational research methods should be strengthened, writing norms for doctoral dissertations in higher education should be formulated, and the research ethics education mechanism for doctoral students should be improved.In 2019, the fifth issue of Heilongjiang Higher Education Research published "Identification of the Main Characteristics of Badly Reviewed Graduate Degree Theses", which selected the blind review opinions of 411 "badly reviewed" graduate degree theses from S University in 2017. The study found that the defects of the specific content of "badly reviewed" papers include the paper theme, logical structure, argument analysis, conclusions and suggestions, writing attitude, writing norms, and innovation. These characteristics are closely related and overlap with each other. In June 2018, the Journal of Editing published "Discussion on the Method of Reviewing and Processing Figures and Tables in Scientific and Technological Journals", pointing out that: Figures and tables play an indispensable role in papers, and have the advantages of large data (or things) capacity, high accuracy, and strong contrast. The purpose is to display the analysis results of data (or things) in a concise, intuitive and beautiful way. According to the expression needs of the paper content, the logical order of reviewing and processing figures and tables, the appropriateness of the selection of the types of figures and tables, the scientificity of expression, standardization and economy of layout.

[0004] The quality of thesis is closely related to the time and energy invested by students in graduation design, but the measurement of student investment is difficult to achieve due to the lack of effective data support. Document CN107239900A discloses an undergraduate graduation thesis quality evaluation method based on the extension cloud model, including the following steps: 1) Establishing an undergraduate graduation thesis quality evaluation index system; 2) Determining the division of grade standards and the grade cloud model: dividing the evaluation index into 5 grades and giving the corresponding cloud model; 3) Obtaining the evaluation value; 4) Determining the combined weight of the index: applying the order relationship analysis method and the entropy weight method to determine the subjective and objective weights of the index respectively, and determining the combined weight of the index according to the minimum discriminant information principle; 5) Determining the expert weight: applying the grey correlation analysis method to determine the expert weight; 6) Establishing the extension cloud correlation matrix; 7) Determining the evaluation grade, so that the quality evaluation of undergraduate graduation thesis is more reasonable, scientific and stable. Document CN108122180A discloses a real-time generation method for autonomous learning engagement based on online learning behavior. The method generates an engagement parameter Es from teaching video playback behavior, video viewing time and concurrent learning behavior data, automates and simplifies the acquisition of online learning engagement data, and makes it possible to obtain engagement data in real time and conduct process analysis, thereby completely changing the current analysis ideas and methods of online learning engagement and making data-supported online learning process guidance and analysis possible. Document CN109858769A provides an evaluation method for the quality of undergraduate graduation project guidance, which collects the percentage scores obtained in the required courses before the graduation project; uses the average score and the standard deviation of the scores of all students in each course to normalize the scores of the course to obtain the relative score; obtains the average score and the standard deviation of the relative scores of each student in each course; uses the average score and the standard deviation of the scores of all students in the graduation project to normalize the graduation project scores to obtain the relative score of the graduation project; compares the relative score of each student's graduation project with the average relative score and the standard deviation of the relative score obtained by the student in the required courses, and obtains the evaluation index of the quality of undergraduate graduation project guidance. CN109045664B discloses a diving scoring method, server and system based on deep learning, which uses a known diving video data set and a corresponding diving score value set to train a diving scoring model; inputs the diving video into the trained diving scoring model, and outputs the diving score value; it avoids artificial interference in the diving score value and improves the accuracy of the diving score value.CN110298038A A text scoring method and device, the method includes: performing word segmentation processing on the text to be scored to obtain the word segmentation of the text to be scored; determining the word meaning vector and the part-of-speech vector of the word segmentation to obtain the word vector composed of the word meaning vector and the part-of-speech vector; inputting the word vector into a pre-trained sequence encoder to obtain the output of the sequence encoder as the sequence encoding vector of the text to be scored; inputting the word vector into a pre-trained tree encoder to obtain the output of the tree encoder as the tree encoding vector of the text to be scored; fusing the sequence encoding vector and the tree encoding vector of the text to be scored to obtain the fused encoding vector of the text to be scored; based on the fused encoding vector, determining the credibility of the text to be scored as a specified type of text as the score of the text to be scored. The accuracy of scoring can be improved. CN110069785A discloses a paper specification control and analysis platform and system based on component conventions. According to the style conventions for components provided by colleges and majors, the input data of students (writers) is converted into XML components with relevant styles and compressed into text documents. At the same time, the student writing process is checked for duplicates in real time and the historical writing situation is recorded. The problem of low efficiency, high randomness and high error rate of traditional manual editing is solved. Based on the tracking and analysis of historical records, the plagiarism behavior of students (writers) in the paper writing process can be grasped in real time, and academic misconduct can be effectively controlled. The text analysis method based on deep learning can mine the potential relationship between the research trend of the paper field and the field entity, and grasp the overall progress of the paper work. Document CN110069768A provides an automatic scoring method for English argumentative papers based on chapter structure, which includes an automatic recognition module for chapter components and an automatic scoring module for chapter structure. The chapter component recognition results and the global structural features between paragraphs are considered comprehensively to improve the scoring effect of the chapter structure of argumentative essays to the greatest extent. In view of the current situation of lack of a complete argumentative essay structure system, document CN112214988A discloses an argumentative essay structure analysis method based on deep learning and rule combination, which can automatically analyze the argumentative essay structure without manual processing, speeding up the analysis of argumentative essay structure and saving labor costs. Document CN108595407A discloses an evaluation method and device based on the chapter structure of argumentative essays, which obtains multiple discourse elements of the argumentative essay to be evaluated through the paragraph type and sentence type of the argumentative essay to be evaluated; constructs the sequential features, plane features or hierarchical features of the article chapter structure through multiple discourse elements; obtains the evaluation results of the argumentative essay to be evaluated through a preset sequential feature model, plane feature model or hierarchical feature model according to the sequential features, plane features or hierarchical features of the article chapter structure.In the process of doctoral education and training, the dissertation is a comprehensive expression of the doctoral students' scientific research ability, innovation ability, ability to master and apply knowledge, and written expression ability. Its quality not only reflects the scientific research ability and academic level of the doctoral applicants, but also reflects the level of graduate education of the training unit to a certain extent. In order to find a method that can quantitatively and systematically analyze the influencing factors of the quality of doctoral dissertations, suggestions are provided for the quality of doctoral training. Document CN111027868A discloses a method for evaluating the influencing factors of dissertation quality based on a structural equation model. The author data of the completed dissertation is collected, the initial variables are calculated, the model fit index is calculated using the structural equation model, and the structural equation model is iteratively corrected to obtain the evaluation results of each variable on the quality of the dissertation.

[0005] CN110287319B invented a method for analyzing student teaching evaluation texts based on sentiment analysis technology, which analyzes and processes student teaching evaluation texts and can solve the defect of incorrect classification of advisory opinion sentences. In July 2021, the Journal of Xi'an Aeronautical University, Volume 39, Issue 4, published "Analysis of Research Hotspots and Trends in Undergraduate Thesis Quality Evaluation Based on Scientific Knowledge Graphs", using the relevant literature on undergraduate thesis quality evaluation included in CNKI from 2000 to 2020 as a sample, using knowledge graphs and bibliometric methods, sorted out the current status of undergraduate thesis quality evaluation research by domestic scholars, which can be divided into two stages in terms of time context: before 2005, the number of publications was relatively small, and the relevant research was mainly a summary of the author's work or management experience; after 2005, the number of publications increased rapidly, and the depth and breadth of research expanded accordingly. First, scholars use statistics, systems engineering and even computer science to conduct research. AHP (analytic hierarchy process), fuzzy comprehensive evaluation and Delphi method are keywords that appear frequently. Secondly, the application of evaluation covers an expanded range of fields. More and more scholars select and select evaluation indicators in a targeted manner based on the characteristics of the disciplines and majors they teach or manage, and build a more suitable and targeted evaluation indicator system. On this basis, they select mature evaluation methods and make further judgments. In the next few years, in terms of research depth, scholars will pay more attention to the quantification and subdivision of undergraduate thesis quality evaluation indicators, so that the methods and conclusions of quality evaluation will be more scientific and reasonable, and more applicable and operational. At the same time, the relevant theoretical knowledge of computer science and topological structure will be more deeply integrated into the relevant research on undergraduate thesis quality evaluation. On the other hand, in terms of research breadth, combining the 14th Five-Year Plan, building a high-quality education system, and standardizing undergraduate thesis quality evaluation methods are the overall research direction and trend of undergraduate thesis quality evaluation related research in the next few years.

[0006] For example, CNKI has a series of products including CNKI undergraduate thesis detection system, graduate thesis academic misconduct detection system, and CNKI undergraduate graduation design (thesis) management system. The detection of duplicate work by relevant superior departments and the management of degree thesis by functional departments of universities mainly refers to the search and duplicate checking services of third-party suppliers such as Wanfang. In June 2019, "Research on Higher Financial Education" published an article entitled "Duplicate Checking Rate, Supervisors and the Quality of Undergraduate Graduation Thesis", which pointed out that after the implementation of the duplication rate detection mechanism, students' main energy is to reduce the duplication rate and make revisions to pass the duplication rate detection, making the originally refined language verbose and the originally fluent sentences incomprehensible. Such modifications are meaningless and waste the energy of students and teachers. In recent years, although various universities have adopted measures such as undergraduate tutor system and writing academic papers, the implementation effect is not satisfactory. In September 2019, the 9th issue of "Computer Application Research" published "A Comprehensive Paper Plagiarism Detection and Evaluation Method". In response to the situation where the Internet's plagiarism detection system process is too extreme, a new comprehensive method is proposed. This method is used to detect anomalies and then require manual detection; its goal is to reduce misjudgments, so that papers that were not originally plagiarized but were judged to be plagiarized have the opportunity to apply for review; when the same paper has a large difference in the duplication rate detected on different plagiarism detection websites, a new comprehensive paper plagiarism detection method can be used to add the weight of human judgment to the final duplication result of the paper, reduce the factors that control the paper's duplication rate by the website, and make the plagiarism detection results not completely controlled by the website or humans. This website / manual dual hybrid detection makes up for the impact of website resource library problems on the plagiarism detection results, and improves the accuracy and credibility of the paper plagiarism detection results. In 2020, the second issue of Journal of Tianjin Normal University (Social Science Edition) published "Universities urgently need to deal with the side effects of plagiarism detection associated with knowledge integrity detection methods", pointing out that plagiarism, which is targeted by plagiarism detection, has not been eliminated, but has added side effects that are contrary to the fundamental purpose of universities to cultivate morality and educate people. In recent years, with the help of the Internet, students' information application ability has been significantly improved, and their ability to obtain resources has been significantly enhanced. The management system of graduation design papers in colleges and universities has become increasingly standardized and diverse, such as self-checking and other checks, but degree theses have not shown a corresponding trend of improvement. Instead, many problems such as logical disorder, pale text, and empty content have gradually increased, and the compilation of data, analysis and discussion, and citation literature are obvious.

[0007] Writing a dissertation is definitely a difficult and character-shaping process. In recent years, the rapid development of artificial intelligence technology has brought disruptive impacts to the academic community. AI tools such as ChatGPT (Vincent text), Midjourney (Vincent picture), Sora (Vincent video) and Suno (Vincent music) have attracted much attention for their outstanding ability to process complex text and visual information. They can not only perform tasks such as translation and proofreading more efficiently and accurately, but also create content efficiently and conveniently, generating high-quality articles, news reports and even novels. These technologies have changed the way information and knowledge are produced, and have become an important engine of social change, reshaping the interaction mode between humans and technology. The encounter between artificial intelligence technology and dissertations will have a profound impact on students and teachers, better understand and meet the teaching needs of both teachers and students, and free up more time and energy for both parties to focus on innovative research. The "Teacher Digital Literacy" (JY / T0646-2022), released on November 30, 2022, includes digital applications as one of the five first-level dimension requirements of the teacher digital literacy framework. Among them, digital applications are further divided into four second-level dimensions: digital teaching design, digital teaching implementation, digital academic evaluation, and digital collaborative education. Specifically, digital teaching implementation further requires: using digital technology resources to carry out individualized guidance, being able to use digital technology resources to discover student learning differences and carry out targeted guidance; digital academic evaluation further requires: ① Selecting and using evaluation data collection tools, being able to reasonably select and use digital tools to collect multimodal academic evaluation data; ② Applying data analysis models to conduct academic data analysis, being able to select and apply appropriate data analysis models to conduct academic data analysis; ③ Realizing the visualization and interpretation of academic data, being able to use digital tools to visualize the results of academic data analysis and provide reasonable interpretations.

[0008] In terms of research depth, scholars will pay more attention to the quantification and subdivision of undergraduate thesis quality evaluation indicators to make the quality evaluation methods and conclusions more scientific and reasonable, thereby enhancing their applicability and operability. At the same time, the relevant theoretical knowledge of computer science and topological structures will be more deeply integrated into the research on undergraduate thesis quality evaluation. On the other hand, in terms of research breadth, building a high-quality education system and standardizing undergraduate thesis quality evaluation methods will be the overall research direction and trend of undergraduate thesis quality evaluation related research in the next few years.

[0009] The intrinsic value of a dissertation is reflected in the rigor of its logical structure, the efficiency of information transmission, and the breadth and depth of its content. Whether it is possible to provide teachers and students with an efficient dissertation structure analysis assistant to analyze and score dissertations through their digital features (such as the total number of words and their distribution, the number and distribution of citations and tables, paragraph structure, etc.) and their implicit feature factors is a current problem.

[0010] Therefore, it is an urgent problem for technical personnel in this field to propose a method for structural analysis and scoring of degree theses to solve the difficulties existing in the prior art. Summary of the invention

[0011] In view of this, the present invention provides a method for structural analysis and scoring of degree theses, which can establish a structural analysis and evaluation model for degree theses through machine learning and analysis of degree theses, thereby realizing structural analysis and evaluation of degree theses and promoting improvement of the structural quality of degree theses.

[0012] In order to achieve the above object, the present invention adopts the following technical solution:

[0013] A method for structural analysis and scoring of a degree thesis comprises the following steps:

[0014] S1. Collect the dissertations that have completed the degree authorization;

[0015] S2, obtain the attribute data and derived data of the collected dissertations to form data set A;

[0016] S3, perform missing check and feature scaling on data set A to construct data training set B;

[0017] S4. Perform multivariate linear regression modeling on the data training set B to obtain the thesis structure analysis and scoring model M;

[0018] S5, obtaining the attribute data and derived data of the degree thesis to be tested, and forming the test data set C;

[0019] S6, check for missing data and perform feature scaling on the test data set C to construct the test evaluation data set D;

[0020] S7. Apply the thesis structure analysis and scoring model M to calculate the evaluation data set D to obtain the score of the thesis structure and give reasonable suggestions for each structure.

[0021] The above method optionally includes a dissertation data table Table_Dissertation, which is composed of field ID, field dissertation type, field enrollment time, field completion time, field degree awarding institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field dissertation grade, field third-party rating, field Chinese abstract rating, field English abstract rating, field acknowledgment rating, field reference citation rationality rating, and field figure rating;

[0022] The attribute data of the degree thesis corresponds to the fields of degree thesis type, field enrollment time, field completion time, field degree-granting institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field thesis grade, field third-party rating, which are stored in the degree thesis data table Table_Dissertation;

[0023] The derived data of the dissertation corresponds to the fields Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score stored in the dissertation data table Table_Dissertation.

[0024] The field word count and distribution, field references and distribution, field figures and distribution, and field tables and distribution are character strings composed of numbers, the symbol At (@), and the symbol period (.). The specific format is: number 1@number 2.number 3.number 4, where number 1 represents the number of words / references / figures / tables, @ represents the position, number 2 represents the first-level title (chapter), number 3 represents the second-level title (section), and number 4 represents the third-level title (article). Multiple character strings are separated by the symbol comma (,).

[0025] In the above method, optionally, in S2, the attribute data and derivative data of the collected dissertations are obtained to form a data set A, where:

[0026] The attribute data of degree theses collected include: type of degree thesis, enrollment date, completion date, degree-granting institution, professional field, author information, instructor, word count of Chinese abstract, word count of English abstract, number of chapters, word count and distribution, references and distribution, figures and distribution, tables and distribution, acknowledgments, thesis grade, and third-party rating;

[0027] The collected dissertation derivative data include: Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score;

[0028] The attribute data and derived data of the collected dissertations constitute the dissertation dataset A.

[0029] The above method optionally obtains the collected thesis-derived data, specifically, performs algorithmic analysis on the Chinese abstract, English abstract, acknowledgments, and references of the collected thesis, performs algorithmic analysis on each figure in the collected thesis to obtain a score, and obtains the thesis-derived data through comprehensive scores.

[0030] In the above method, optionally, the Chinese abstract score and the English abstract score are scores obtained by using algorithms to perform text analysis on the Chinese abstracts and English abstracts of the collected degree theses, respectively. The algorithms include: support vector machine, naive Bayes, K-means clustering and neural network;

[0031] The acknowledgment score is a score obtained by using an algorithm to perform text analysis on the acknowledgments of the collected dissertations. The algorithm uses: SnowNLP, a third-party Python library;

[0032] The rationality score of reference citation is to score the reference citation of the collected dissertations one by one by applying text analysis algorithm, and then calculate the average value after accumulating and summing up the scores. The text analysis algorithm includes: text similarity algorithm, semantic analysis algorithm, and graph algorithm.

[0033] The scoring of the images is achieved by applying an image analysis algorithm to score each image in the collected dissertations one by one, and then accumulating and summing up the scores to calculate the average value. The image analysis algorithm includes image quality analysis, image content understanding, and image correlation analysis.

[0034] Through the above technical solutions, it can be seen that compared with the prior art, the present invention provides a method for structural analysis and scoring of degree theses, which has the following beneficial effects: 1) Apply multiple algorithms and technologies to perform multi-dimensional analysis on the structure of degree theses, including text, images, references, etc., which is comprehensive and systematic; 2) Provide accurate feedback on the structure of the paper according to the analysis results, help the author identify and improve the shortcomings of the paper, thereby improving the quality of the paper structure; 3) Use automated analysis algorithms to significantly reduce the subjective bias of manual scoring and ensure the objectivity and consistency of the scoring results; 4) Greatly improve work efficiency, especially when dealing with a large number of degree theses, significantly reduce the time and labor intensity of manual review; 5) This method has good scalability and can be adjusted according to the needs of different disciplines, and is suitable for the quality assessment of various degree theses; 6) The present invention not only significantly improves the scientificity and efficiency of degree thesis review, but also provides strong support for the standardization and quality improvement of academic research. . BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0036] Figure 1 A flow chart of the structural analysis and scoring method for a degree thesis provided by the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] In this application, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, the elements defined by the sentence "comprise one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0039] Reference Figure 1 As shown, the present invention discloses a method for structural analysis and scoring of a degree thesis, comprising the following steps:

[0040] S1. Collect the dissertations that have completed the degree authorization;

[0041] S2, obtain the attribute data and derived data of the collected dissertations to form data set A;

[0042] S3, performing missing value checking (such as deleting missing values) and feature scaling (such as standardization, normalization, and maximum absolute value scaling) algorithm processing on data set A to construct data training set B;

[0043] S4. Perform a multivariate linear regression algorithm (such as stepwise regression method, Bayesian linear regression method) on the data training set B to obtain a thesis structure analysis and scoring model M;

[0044] S5, obtaining the attribute data and derived data of the degree thesis to be tested, and forming the test data set C;

[0045] S6, using the same missing check and feature scaling algorithm as step S3 to process the test data set C, and construct the test evaluation data set D;

[0046] S7. Apply the thesis structure analysis and scoring model M to calculate the evaluation data set D to obtain the score of the thesis structure and give reasonable suggestions for each structure.

[0047] Further, as shown in Table 1, there is a dissertation data table Table_Dissertation, which consists of field ID, field dissertation type, field enrollment time, field completion time, field degree awarding institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field dissertation grade, field third-party rating, field Chinese abstract rating, field English abstract rating, field acknowledgment rating, field reference citation rationality rating, field figure rating;

[0048] Furthermore, the attribute data of the degree thesis corresponds to the fields of degree thesis type, field enrollment time, field completion time, field degree awarding institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field thesis grade, field third-party rating, which are stored in the degree thesis data table Table_Dissertation;

[0049] The derived data of the dissertation corresponds to the fields Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score stored in the dissertation data table Table_Dissertation.

[0050] The field word count and distribution, field references and distribution, field figures and distribution, and field tables and distribution are character strings composed of numbers, the symbol At (@), and the symbol period (.). The specific format is: number 1@number 2.number 3.number 4, where number 1 represents the number of words / references / figures / tables, @ represents the position, number 2 represents the first-level title (chapter), number 3 represents the second-level title (section), and number 4 represents the third-level title (article). Multiple character strings are separated by the symbol comma (,).

[0051] Table 1 Dissertation data table Table_Dissertation

[0052]

[0053]

[0054] Furthermore, in S2, the attribute data and derived data of the collected dissertations are obtained to form a dataset A.

[0055] Furthermore, the attribute data of degree theses collected include: type of degree theses, enrollment date, completion date, degree-granting institution, professional field, author information, instructor, word count of Chinese abstract, word count of English abstract, number of chapters, word count and distribution, references and distribution, figures and distribution, tables and distribution, acknowledgments, thesis grade, and third-party rating;

[0056] The collected dissertation derivative data include: Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score;

[0057] The attribute data and derived data of the collected dissertations constitute the dissertation dataset A.

[0058] Furthermore, the derived data of the collected degree theses are obtained, specifically, algorithmic analysis is performed on the Chinese abstract, English abstract, acknowledgments, and references of the collected degree theses, and algorithmic analysis is performed on each figure in the collected degree theses to obtain scores and obtain the derived data of the degree thesis through comprehensive scores.

[0059] Furthermore, the Chinese abstract score and the English abstract score are obtained by using algorithms to perform text analysis on the Chinese abstracts and English abstracts of the collected dissertations, respectively. The algorithms use common text analysis and natural language processing algorithms and tools, including: support vector machines, naive Bayes, K-means clustering, and neural networks;

[0060] The acknowledgment score is obtained by using an algorithm to perform text analysis on the acknowledgments of the collected dissertations. The algorithm uses common text analysis and natural language processing algorithms and tools, such as the Python third-party library SnowNLP.

[0061] The rationality score of reference citation is to score the reference citation of the collected dissertations one by one by applying the text analysis algorithm, and then calculate the average value by accumulating and summing up the scores. The text analysis algorithm adopts common text analysis and natural language processing algorithms and tools, including: text similarity algorithm (cosine similarity, word frequency-inverse document frequency), semantic analysis algorithm (word vector model, BERT), and graph algorithm (citation network analysis PageRank);

[0062] The scoring of the images is achieved by applying an image analysis algorithm to score each image in the collected dissertations one by one, and then accumulating and summing up the scores to calculate the average value. The image analysis algorithm uses common image processing algorithms and tools, including image quality analysis (signal-to-noise ratio, contrast assessment, clarity assessment, image compression rate), image content understanding (edge ​​detection, YOLO), and image correlation analysis (ResNet, VGG).

[0063] Further, S5, the attribute data and derived data of the degree thesis to be tested are obtained to form the test data set C. Further, the attribute data of the degree thesis to be tested include: type of degree thesis, enrollment time, completion time, degree-granting institution, professional field, author information, instructor, word count of Chinese abstract, word count of English abstract, number of chapters, word count and distribution, references and distribution, figures and distribution, tables and distribution, acknowledgments, thesis grade, and third-party rating;

[0064] The derived data of the degree thesis to be tested include: Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score;

[0065] The attribute data and derived data of the thesis to be tested constitute the test data set C.

[0066] Further, the derived data of the degree thesis to be tested is obtained, specifically, the Chinese abstract, English abstract, acknowledgments, and references of the degree thesis to be tested are respectively subjected to algorithmic analysis, and each figure in the degree thesis to be tested is subjected to algorithmic analysis to obtain a score and calculate the comprehensive average score to obtain the derived data of the degree thesis to be tested;

[0067] The Chinese abstract score and the English abstract score are obtained by using algorithms such as support vector machine, naive Bayes, K-means clustering and neural network to analyze the Chinese abstract and English abstract of the degree thesis to be tested respectively.

[0068] The acknowledgment score is the score obtained by using an algorithm such as the Python third-party library SnowNLP to perform text analysis on the acknowledgments of the thesis to be tested;

[0069] The rationality score of reference citation is to score the reference citation of the collected dissertations one by one by applying text analysis algorithms, and then calculate the average value by accumulating and summing up the scores. The text analysis algorithms include: text similarity algorithm (cosine similarity, word frequency-inverse document frequency), semantic analysis algorithm (word vector model, BERT), and graph algorithm (citation network analysis PageRank).

[0070] The scoring of the images is achieved by applying image analysis algorithms to score each image in the collected dissertations one by one, and then accumulating and summing up the scores to calculate the average value. The image analysis algorithms include image quality analysis (signal-to-noise ratio, contrast assessment, clarity assessment, image compression rate), image content understanding (edge ​​detection, YOLO), and image correlation analysis (ResNet, VGG).

[0071] In a specific embodiment, taking 4 collected degree theses to be tested as an example,

[0072] Collect degree theses that have completed degree authorization (self-examination, external review and defense); extract the attribute data of these degree theses, analyze and obtain derivative data, and construct a degree thesis data set A; check for missing and feature scaling of degree thesis data set A to make the importance of all features consistent, and construct a degree thesis data training set B; perform multivariate linear regression modeling on the degree thesis data training set B to obtain the structural analysis and scoring model of degree thesis; carry out structural analysis and scoring of the degree thesis to be tested: extract the attribute data of the degree thesis to be tested, calculate and analyze to obtain derivative data, and form a degree thesis data set C to be tested; check for missing and feature scaling of the degree thesis data set C to be tested, and construct an evaluation data set D for the degree thesis to be tested; apply the structural analysis and scoring model of degree thesis to calculate the evaluation data set D for the degree thesis to be tested, obtain the structural score of the degree thesis to be tested, and give reasonable suggestions for each structure; apply text analysis algorithm to score the Chinese abstract part; apply text analysis algorithm to score the English abstract part; apply text sentiment analysis algorithm to score the acknowledgment part; apply text analysis model algorithm to score the rationality of reference citation; apply graphic analysis algorithm to score the quality of the figure. The attribute data of 4 degree theses to be tested and the derived data obtained by calculation are as follows:

[0073] 4 sets of attribute data of degree theses to be tested and derived data tables

[0074]

[0075]

[0076]

[0077] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0078] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for structural analysis and scoring of degree theses, characterized in that: The following steps are involved: S1. Collect the dissertations that have completed the degree authorization; S2, obtain the attribute data and derived data of the collected dissertations to form data set A; S3, perform missing check and feature scaling on data set A to construct data training set B; S4. Perform multivariate linear regression modeling on the data training set B to obtain the thesis structure analysis and scoring model M; S5, obtaining the attribute data and derived data of the degree thesis to be tested, and forming the test data set C; S6, performing missing check and feature scaling processing on the test data set C to construct the test evaluation data set D; S7. Apply the thesis structure analysis and scoring model M to calculate the evaluation data set D to obtain the score of the thesis structure and give reasonable suggestions for each structure.

2. A method for structural analysis and scoring of degree theses according to claim 1, characterized in that: There is a dissertation data table Table_Dissertation, which consists of field ID, field dissertation type, field enrollment time, field completion time, field degree awarding institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field dissertation grade, field third-party rating, field Chinese abstract rating, field English abstract rating, field acknowledgment rating, field reference citation rationality rating, field figure rating; The attribute data of the degree thesis corresponds to the fields of degree thesis type, field enrollment time, field completion time, field degree-granting institution, field professional field, field author information, field instructor, field Chinese abstract word count, field English abstract word count, field number of chapters, field word count and distribution, field references and distribution, field figure and distribution, field table and distribution, field acknowledgment, field thesis grade, field third-party rating, which are stored in the degree thesis data table Table_Dissertation; The derived data of the dissertation corresponds to the fields Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score stored in the dissertation data table Table_Dissertation. The field word count and distribution, field references and distribution, field figures and distribution, and field tables and distribution are character strings composed of numbers, the symbol At (@), and the symbol period (.). The specific format is: number 1@number 2.number 3.number 4, where number 1 represents the number of words / references / figures / tables, @ represents the position, number 2 represents the first-level title (chapter), number 3 represents the second-level title (section), and number 4 represents the third-level title (article). Multiple character strings are separated by the symbol comma (,).

3. A method for structural analysis and scoring of a degree thesis according to claim 1, characterized in that: The attribute data of degree thesis in S2 include: type of degree thesis, enrollment date, completion date, degree-granting institution, professional field, author information, instructor, word count of Chinese abstract, word count of English abstract, number of chapters, word count and distribution, references and distribution, figures and distribution, tables and distribution, acknowledgments, thesis grade, and third-party rating; The derived data of the dissertation include: Chinese abstract score, English abstract score, acknowledgment score, reference citation rationality score, and figure score; The attribute data and derived data of the dissertation constitute the dissertation dataset A.

4. A method for structural analysis and scoring of degree theses according to claim 3, characterized in that: The specific contents of the collected dissertation derivative data in S2 are as follows: Algorithmic analysis is performed on the Chinese abstracts, English abstracts, acknowledgments, and references of the collected dissertations. Algorithmic analysis is performed on each figure in the collected dissertations to obtain scores, and the scores are combined to obtain the dissertation-derived data.

5. A method for structural analysis and scoring of degree theses according to claim 4, characterized in that: The Chinese abstract score and English abstract score are obtained by using algorithms to analyze the Chinese abstracts and English abstracts of the collected dissertations respectively. The algorithms include support vector machine, naive Bayes, K-means clustering and neural network. The acknowledgment score is a score obtained by using an algorithm to perform text analysis on the acknowledgments of the collected dissertations. The algorithm uses: SnowNLP, a third-party Python library; The rationality score of reference citation is to score the reference citation of the collected dissertations one by one by applying text analysis algorithm, and then calculate the average value after accumulating and summing up the scores. The text analysis algorithm includes: text similarity algorithm, semantic analysis algorithm, and graph algorithm. The scoring of the images is achieved by applying an image analysis algorithm to score each image in the collected dissertations one by one, and then accumulating and summing up the scores to calculate the average value. The image analysis algorithm includes image quality analysis, image content understanding, and image correlation analysis.

Images