A method and system for automatically building courses for information professional experiments

Abstract

The present invention relates to a method and system for automatically building courses for information-related professional experiments, including the following steps: (1) initialization; A, crawling online information-related course textbook information, building a knowledge point keyword thesaurus and knowledge graph based on a course; B, encapsulating resources, recommending tags for resource packages, and performing tag processing; (2) building experimental courses according to user needs: C, pulling out the keyword thesaurus of the textbook or similar course textbooks according to the textbook book information provided by the user; D, matching the pulled out keyword thesaurus according to step C to construct the experimental course. The present invention provides the most advanced and comprehensive information resources based on the system, provides teachers with the most convenient automatic course building mode, and solves the problem of tedious lesson preparation for teachers; the present invention provides a comprehensive platform for autonomous learning and operation for students of information-related majors; the present invention enhances the interactivity between teachers and students, and maximizes the efficiency of experimental teaching.

Description

Technical Field

[0001] The present invention relates to a method and system for automatically building an information-related professional experimental course. Specifically, the present invention relates to using machine learning, natural language processing and other methods to classify information and construct a knowledge graph, as well as automatically building an information-related professional experimental course. Background Art

[0002] Against the backdrop of the rapid development of the Internet, with the continuous reform of the education system and the continuous improvement of the education level, the Internet-based online learning model has been supported by teachers, students and parents. The Internet teaching model has broken the traditional teaching methods. In recent years, online learning teaching methods have developed rapidly in today's big data era. Online learning platforms have become a positive factor in promoting and ensuring effective learning. At the same time, they also provide strong support and guarantee for building learning groups, enhancing learning cognitive abilities, carrying out learning activities, managing and sharing learning resources, ensuring teaching quality and promoting interaction between teachers and students. For information-related professional courses, online virtual experiment platforms have made great progress.

[0003] With the continuous improvement of desktop virtualization technology, more and more colleges and universities have begun to adopt virtual desktops. The rapid development of cloud computing and big data technology has led and promoted the information reform and construction of colleges and universities, changed the traditional mode of information teaching, and carried out a deeper combination of theory and practical operation, which has improved students' independent innovation awareness and innovation ability. The traditional computer experiment teaching equipment has poor sharing. Every time the software required for teaching is installed, hundreds of computers in the entire computer room must be operated simultaneously. In addition, the equipment in the computer room is also very different, the utilization rate is low, the management is complex, and the distribution of digital information resources is uneven. In addition, there is a lack of process management in the programming practice teaching mode. The teachers lack timely guidance and supervision of the students' operation practice process, which is not conducive to the personalized development of students. The above reasons have also accelerated the development of cloud computing technology in computer experiment teaching. The cloud platform is a new software application model. It builds a systematic experimental teaching cloud platform based on cloud computing technology, which solves the problems encountered in traditional experimental teaching. At the same time, it also realizes the transformation of experimental environment construction from single-task system to multi-terminal virtual simulation system, providing sufficient experimental and equipment space for college students; it provides a parallel and isolable experimental environment, so that it can achieve high efficiency under low-cost conditions; it improves the problem of repetitive labor of college teachers, and provides real-time and effective supervision and guidance of students in the process of experimental teaching; it also fully mobilizes students' interest in experimental courses, forming a professional experimental teaching platform for electronic information related majors. However, this kind of professional experimental teaching platform usually has the following problems: First, the experimental teaching platform will cause its resources to be rapidly iterated to a certain extent due to the update of information, which will result in a phenomenon that a server or container image contains multiple versions. This phenomenon, while providing multiple choices, also adds burden to the process of building experimental templates for users; secondly, the steps in configuring the experimental environment are relatively complicated. Most experimental cloud platforms do not have uniform operations when configuring the environment. It takes time to read the user manual, which is too complicated for initial users. In addition, a textbook involves many experiments. It is a relatively large amount of work for teachers to configure the experimental courses for a textbook; after formulating an experimental course, it is necessary to write operation documents or experimental reports based on the configured experimental course, which undoubtedly increases the workload of teachers. Through the description of the above problems, it can be seen that there is currently no mature and complete system to assist teachers of information professional experimental courses in creating experimental courses, which cannot meet the current teaching needs. Summary of the invention

[0004] In view of the problems in the teaching of information-related subjects described above, the present invention provides an automatic construction system for experimental courses of information-related professional courses, including generating knowledge graphs and experimental courses based on the textbooks. Among them, the professional knowledge graph of the professional course is generated for the existing information-related professional course textbook catalog, and the experimental courses corresponding to the textbooks are automatically constructed.

[0005] The present invention also provides a course automatic construction system for information professional experiments.

[0006] Terminology explanation:

[0007] Digitization: Digitization is the process of converting complex and ever-changing information into measurable numbers and data, and then using these numbers and data to build appropriate digital models, converting them into a series of binary codes, and introducing them into the computer for unified processing. This is the basic process of digitization.

[0008] The technical solution of the present invention is:

[0009] An automatic construction method for an information professional experimental course includes the following steps:

[0010] (1) Initialization;

[0011] A. Crawling online information course textbooks and books information, building a keyword thesaurus and knowledge graph based on a course’s knowledge points;

[0012] B. Encapsulate the resources, recommend tags to the resource packages through the knowledge point keyword thesaurus obtained in step A, and perform tag processing;

[0013] (2) Build experimental courses based on user needs:

[0014] C. Pull out the keyword database of the course knowledge points of the textbook according to the textbook information provided by the user;

[0015] D. Construct the experimental course according to the keyword vocabulary pulled out by matching in step C.

[0016] Preferably, in step A, crawling online information course textbook book information to construct a knowledge point keyword thesaurus and knowledge graph based on a course includes the following steps:

[0017] a. Perform web crawlers to obtain information on textbooks related to electronic information majors and form a database of textbooks related to electronic information majors. The textbooks related to electronic information majors include the titles, versions, publishers, authors, courses, catalog information and course knowledge maps of textbooks, which are divided by subject.

[0018] b. Perform character recognition on the catalog of teaching materials under the same course and implement digitization. On the basis of the digitized structure, limit the length of the catalog bar and set a threshold. The catalog bar longer than the threshold is defined as a long catalog bar, otherwise, it is defined as a short catalog bar, thus dividing the long catalog bar and the short catalog bar;

[0019] Perform part-of-speech tagging on the long directory according to the natural language processing method to obtain a part-of-speech array, and extract a batch of candidate nodes from the obtained part-of-speech array according to the part-of-speech rules;

[0020] The short-term directory uses part-of-speech rules to extract another batch of candidate nodes;

[0021] c. Identify the authenticity and reliability of the two batches of candidate nodes obtained in step b in Baidu Encyclopedia, complete filtering, and obtain a knowledge point keyword thesaurus based on a course;

[0022] d. For a course’s knowledge point keyword thesaurus, identify these candidate nodes based on the hierarchical structure of the directory in Baidu Encyclopedia to form a hierarchical relationship, and use the same-level structure of the directory to form a parallel relationship;

[0023] The primary skeleton of a knowledge graph based on a course is formed by the hierarchical structure and the parallel structure;

[0024] e. The parallel relationship that appears in the primary skeleton of the knowledge graph obtained in step d is further divided into upper and lower structures, that is, the parallel relationship is distinguished into strong and weak ones, the number of times the two parts of the nodes in the parallel relationship appear is counted, and the absolute frequency of occurrence, that is, the number of times the two parts of the nodes in the parallel relationship appear, is sorted. At the same time, a threshold is set. Nodes with an absolute frequency greater than the threshold are defined as strong parallel relationships, otherwise, they are defined as weak parallel relationships. This cycle is repeated, sorting and screening are performed, and the redefined nodes are supplemented to the primary skeleton according to the results of the distinction, and finally a complete knowledge graph based on a course is obtained.

[0025] Further preferably, in step b, the threshold is set to 9.

[0026] Further preferably, the long directory is tagged with parts of speech according to a natural language processing method to obtain a part-of-speech array, and a batch of candidate nodes are extracted from the obtained part-of-speech array according to part-of-speech rules. The specific implementation process is as follows:

[0027] Perform word segmentation and part-of-speech tagging on the long directory to obtain a part-of-speech array, and use the part-of-speech rules to extract a batch of candidate nodes from the obtained part-of-speech array;

[0028] ① Use the natural language processing tool FudanNLP to segment the long directory and mark the part of speech, split the long directory according to the comma and the parallel conjunctions with the part of speech as conjunctions, and split a long directory into a string array;

[0029] ② For each substring N[i] in each split string array N, form a part-of-speech array of <entry, part-of-speech> for each substring N[i];

[0030] ③ Merge adjacent elements in the part-of-speech array. In the merging process, different merging orders are used for strings at different positions in the string array. That is, when processing the part-of-speech array of the first substring N[0] in the string array N, consecutive adjectives and nouns are merged into one word from back to front. At the same time, the part-of-speech of the last word in the part-of-speech array of N[0] is used as the benchmark part-of-speech. The benchmark part-of-speech is used to improve the accuracy of merging the part-of-speech array of the next string in the string array N. When processing the part-of-speech array of the second substring N[1] and each subsequent substring, the following method is used: if the benchmark part-of-speech is a noun, then the part-of-speech array of each string after N[1] is matched from front to back until the last part-of-speech is a noun to form a word, otherwise, no result is returned; if the benchmark part-of-speech is not a noun or a marking symbol, then in the part-of-speech array of each string after N[1], when the first part-of-speech matches the benchmark part-of-speech, a word is formed, otherwise, no result is returned;

[0031] The final word is the candidate node.

[0032] Further preferably, in step e, the threshold is set to 20.

[0033] According to the preferred embodiment of the present invention, in step B, encapsulating resources, recommending tags to resource packages through the knowledge point keyword thesaurus obtained in step A, and performing tag processing means:

[0034] The configured experimental template, experimental environment, experimental report and operation documents are packaged. The experimental template includes the selected operating system and software configuration; the experimental environment includes virtual hardware facilities that match the experimental template, including CPU, memory, and hard disk specifications; for each configured experimental template and experimental environment, there are corresponding experimental reports and operation documents.

[0035] Label the resource package that contains the packaged virtual software package, experimental report and operation document. The label content contains various information about the software package, including the knowledge points involved in the experiment, memory size, targeted operation content, and estimated duration of the experiment. Among them, the main label of the resource package is recommended by searching the knowledge point keyword library obtained during the digital processing of the directory; the sub-label content is the specific information of the resource package.

[0036] Preferably, according to the present invention, in step C, a knowledge point keyword thesaurus of the textbook or similar course textbooks is pulled out based on the textbook book information provided by the user, specifically: after the user enters the textbook book name, one or more textbook book information is retrieved in the electronic information textbook information library, including the textbook book name, textbook version, publisher, author name, corresponding course, catalog information, and course knowledge graph information. The user selects the textbook book information item he needs, and the textbook provided by the user is matched with the course knowledge graph.

[0037] According to the preferred embodiment of the present invention, in step D, the experimental course is constructed according to the course knowledge graph pulled out by matching in step C, which specifically includes the following steps:

[0038] f. Through step b and step c, the candidate nodes are obtained by digital processing of the keyword word library pulled out by step C, and these candidate nodes are proofread and filtered in Baidu Encyclopedia, and the cleaned knowledge point keywords are formed into a knowledge point keyword word library for this course;

[0039] These knowledge point keyword lexicon also constitute the nodes of the knowledge graph of this course;

[0040] Associate the nodes of the knowledge graph of this course with the main label of the resource package obtained in step B;

[0041] g. After selecting the resource package, it will be combined with the knowledge graph built based on the course to form an experimental course.

[0042] A course automatic construction system for information professional experiments, used to implement the above-mentioned course automatic construction method for information professional experiments, including a knowledge graph construction unit, a label processing unit, a keyword vocabulary matching unit and an experimental course construction unit;

[0043] The knowledge graph construction unit is used to implement step A; the label processing unit is used to implement step B; the keyword vocabulary matching unit is used to implement step C; and the experimental course construction unit is used to implement step D.

[0044] The beneficial effects of the present invention are:

[0045] 1. The present invention faces the problem of complicated steps in formulating experimental courses for information courses. The present invention breaks the limitations of the current traditional information subject teaching mode due to factors such as time, location, hardware, etc., and the limitation that it cannot meet the current teaching needs;

[0046] 2. The present invention provides the most advanced and comprehensive information resources based on the system, and provides teachers with the most convenient automatic course building mode, solving the problem of tedious lesson preparation for teachers;

[0047] 3. The present invention provides a comprehensive platform for autonomous learning and operation for students majoring in information technology;

[0048] 4. The present invention enhances the interaction between teachers and students and maximizes the efficiency of experimental teaching. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the method for automatically building experimental courses for information majors according to the present invention;

[0050] Figure 2 Constructing a process diagram for the knowledge graph of the present invention;

[0051] Figure 3 Constructing a flow chart for the experimental course of the present invention;

[0052] Figure 4 This is a sample diagram of crawled electronic information related professional textbook information;

[0053] Figure 5 This is a sample diagram of the operating system and software configuration selected in the experiment template;

[0054] Figure 6 This is a sample diagram of the CPU, memory, and hard disk specifications included in the experimental environment. DETAILED DESCRIPTION

[0055] The present invention will be further defined below in conjunction with the accompanying drawings and embodiments, but is not limited thereto.

[0056] Example 1

[0057] An automatic construction method for experimental courses in information majors, such as Figure 1 As shown, the steps include:

[0058] (1) Initialization;

[0059] A. Crawling online information course textbooks and book information, building a keyword thesaurus and knowledge graph based on the knowledge points of a course; using web crawler technology to crawl nearly 20,000 textbook book information about electronic information majors on a certain book marketing platform, and storing the information in the database. Taking Python language programming as an example, the information is listed as textbook name, version, publisher, author, course and catalog information, as shown in the figure below.

[0060] B. Encapsulate the resources, recommend tags to the resource packages through the knowledge point keyword thesaurus obtained in step A, and perform tag processing;

[0061] (2) Build experimental courses based on user needs:

[0062] C. Pull out the keyword database of the course knowledge points of the textbook according to the textbook information provided by the user;

[0063] D. Construct the experimental course according to the keyword vocabulary pulled out by matching in step C.

[0064] Example 2

[0065] The automatic construction method for information professional experimental courses described in Example 1 is different in that:

[0066] In step A, crawling online information course textbook book information, building a knowledge point keyword thesaurus and knowledge graph based on a course, including the following steps:

[0067] a. We crawled Dangdang.com and obtained nearly 20,000 pieces of information on textbooks for electronic information majors on a certain book marketing platform to form a database of textbooks for electronic information majors. The textbook information for electronic information majors includes the title, version, publisher, author, course, catalog information, and course knowledge graph of textbooks, such as Figure 4 As shown, it is divided by subject;

[0068] b. Figure 2 As shown, character recognition is performed on the textbook catalog under the same course, and the catalog is digitized. On the basis of the digitized structure, the length of the catalog bar is limited, and a threshold is set to 9. A catalog bar greater than the threshold is defined as a long catalog bar, otherwise, it is defined as a short catalog bar, thereby dividing the long catalog bar and the short catalog bar;

[0069] Perform part-of-speech tagging on the long directory according to the natural language processing method to obtain a part-of-speech array, and extract a batch of candidate nodes from the obtained part-of-speech array according to the part-of-speech rules;

[0070] The short-term directory uses part-of-speech rules to extract another batch of candidate nodes;

[0071] c. Identify the authenticity and reliability of the two batches of candidate nodes obtained in step b in Baidu Encyclopedia, complete filtering, and obtain a knowledge point keyword thesaurus based on a course; use Baidu Encyclopedia entries to build an index, and when processing candidate nodes in the catalog, conduct Baidu Encyclopedia identification, and those that pass the identification will be included, while nodes that are not identified and verb nodes will not be included, so as to achieve the effect of filtering candidate nodes. The candidate nodes obtained after filtering form a keyword thesaurus, thereby obtaining a knowledge point keyword thesaurus based on a course;

[0072] d. For a course’s knowledge point keyword thesaurus, identify these candidate nodes based on the hierarchical structure of the directory in Baidu Encyclopedia to form a hierarchical relationship, and use the same-level structure of the directory to form a parallel relationship;

[0073] According to the catalog of each category of course materials, integrate the keyword thesaurus for each course's knowledge points. After the thesaurus is integrated, the primary framework of the knowledge graph based on a course is formed according to the identification of the superior and subordinate structures. The catalog content of the book has a good organizational structure, a relatively clear hierarchy, and the knowledge units and chapters are listed according to certain order rules. In other words, in books, especially in course knowledge materials, there is a certain correlation between the superior catalog and the subordinate catalog in many aspects, such as the relationship between the theme and the sub-theme. Therefore, for the catalog itself, it can be used as a high-quality corpus for superior and inferior extraction. By calculating the correlation between nodes and processing the nodes of the parallel relationship as superior and inferior node relationships through the book structure, they are added to the knowledge graph as an increment, and the information of each array saved at the same time can be used for the extraction of the next superior and inferior relationship nodes. The primary skeleton of the knowledge graph based on a course is formed by the superior and inferior structure and the parallel structure.

[0074] The technology for forming an indication map based on a book or a course is now relatively mature. The catalog of a course book itself has a good hierarchical relationship between superiors and subordinates, but in the hierarchical relationship, there may be a problem of many-to-many nodes for the hierarchical relationship, so it is necessary to analyze the hierarchical relationship words to obtain a hierarchical relationship with stronger relevance and discrimination. For each pair of superordinate chapter entity words a and subordinate entity words b in the knowledge point keyword thesaurus formed in step c, extract classification features. The classification features of the hierarchical words include two categories: superordinate and subordinate pattern features, and directory structure features.

[0075] The hyponym pattern feature refers to whether a and b share a prefix or suffix. If two words satisfy the prefix-suffix relationship, they generally satisfy the hyponym relationship. In general, for (a, b), if a and b share a suffix, and b is longer than a, b will generally be a hyponym of a. For example, (queue, chain queue), (operator, arithmetic operator), etc.

[0076] The directory structure feature refers to the number of different parent chapters in which each word appears. Generally speaking, for a specific sub-directory, in the entire knowledge point keyword database, its parent directories should not be very many or the parent directories should not have many different topics. For example, the word "computer" can have parent directories such as "computer science", "electrical appliances" and "natural sciences". However, the word "answer" may appear in the sub-directory of any directory.

[0077] After the classification is completed according to the features extracted in the previous step, each word and all its hyponyms are sorted out. The classification generated by the hypernyms and hyponyms of the knowledge point keyword vocabulary is used. For each book, the hypernym and all its hyponyms are grouped together. For the same hypernym in all books, the similarity of its hyponyms is counted, and the hypernyms with similarities close to infinity are merged to form the framework of a primary indication map.

[0078] e. The parallel relationships in the primary skeleton of the knowledge graph obtained in step d are further divided into upper and lower structures, that is, the parallel relationships are distinguished into strong and weak ones, the number of times the two parts of the nodes in the parallel relationship appear is counted, and the absolute frequency of occurrence, that is, the number of times the two parts of the nodes in the parallel relationship appear, is sorted. At the same time, the threshold is set to 20, and the nodes with an absolute frequency greater than the threshold are defined as strong parallel relationships, otherwise, they are defined as weak parallel relationships. This cycle is repeated, sorting and screening, and the redefined nodes are supplemented to the primary skeleton according to the results of the distinction, and finally a complete knowledge graph based on a course is obtained. The knowledge graph of each course is incorporated into the electronic information professional textbook information database, which becomes the basis for subsequent matching resource packages with textbook information input by users.

[0079] According to the natural language processing method, the long directory is tagged with parts of speech to obtain a part-of-speech array. A batch of candidate nodes are extracted from the obtained part-of-speech array according to the part-of-speech rules. The specific implementation process is as follows:

[0080] Use the Chinese word segmentation system to segment and tag the long directory to obtain a part-of-speech array, and use the part-of-speech rules to extract a batch of candidate nodes from the obtained part-of-speech array;

[0081] ① Use the natural language processing tool FudanNLP to segment the long directory and mark the part of speech, split the long directory according to the comma and the parallel conjunctions with the part of speech as conjunctions, and split a long directory into a string array;

[0082] ② For each substring N[i] in each split string array N, form a part-of-speech array of <entry, part-of-speech> for each substring N[i];

[0083] ③ Merge adjacent elements in the part-of-speech array. In the merging process, different merging orders are used for strings at different positions in the string array. That is, when processing the part-of-speech array of the first substring N[0] in the string array N, consecutive adjectives and nouns are merged into one word from back to front. At the same time, the part-of-speech of the last word in the part-of-speech array of N[0] is used as the benchmark part-of-speech. The benchmark part-of-speech is used to improve the accuracy of merging the part-of-speech array of the next string in the string array N. When processing the part-of-speech array of the second substring N[1] and each subsequent substring, the following method is used: if the benchmark part-of-speech is a noun, then the part-of-speech array of each string after N[1] is matched from front to back until the last part-of-speech is a noun to form a word, otherwise, no result is returned; if the benchmark part-of-speech is not a noun or a marking symbol, then in the part-of-speech array of each string after N[1], when the first part-of-speech matches the benchmark part-of-speech, a word is formed, otherwise, no result is returned;

[0084] The final word is the candidate node.

[0085] In step B, the resources are packaged, and the knowledge point keyword thesaurus obtained in step A is used to recommend tags for the resource package, and tag processing is performed, which means:

[0086] The configured experimental templates, experimental environments, experimental reports and operation documents are packaged. The experimental templates include the selected operating system and software configuration; the experimental environment includes virtual hardware facilities that match the experimental templates, including CPU, memory, and hard disk specifications; for each configured experimental template and experimental environment, there are corresponding experimental reports and operation documents; for the packaged resource packages, there are detailed instructions for each module, such as the targeted operation content, the knowledge points involved, the memory size, and the estimated operation time. This information will later exist as a sub-label of the software package.

[0087] The resource package containing the packaged virtual software package, experimental report and operation document is labeled. The label content contains various information of the software package, including the knowledge points involved in the experiment, memory size, targeted operation content, and expected duration of the experiment. Among them, the main label of the resource package is recommended by searching the knowledge point keyword library obtained during the digital processing of the directory. When the resource package is packaged and packaged, the main label needs to be retrieved in the knowledge point keyword library when it is packaged. For example, if the experimental content is the relevant part of Python statements, when the resource is packaged and labeled, enter keywords such as "Python statement" in the knowledge point keyword library, and the label recommendations such as "Python loop statement", "Python jump statement", and "Python while loop statement" in the knowledge point keyword library will be retrieved. At that time, the appropriate vocabulary elements are selected and introduced into the resource package as the main label. The information of these labels is used as the main basis for subsequent matching with the course knowledge graph nodes to form the experimental course. The content of the secondary label is the specific information of the resource package. For example, the targeted operation content, the knowledge points involved, the memory size, and the expected operation time.

[0088] In step C, the knowledge point keyword thesaurus of the textbook or similar course textbooks is pulled out based on the textbook information provided by the user. Specifically, after the user enters the textbook name, one or more textbook information is retrieved from the electronic information textbook information library, including the textbook name, textbook version, publisher, author name, course, catalog information, and course knowledge graph information. The user selects the textbook information item he needs, and the textbook provided by the user is matched with the course knowledge graph.

[0089] In step D, the experimental course is constructed according to the course knowledge graph pulled out in step C, such as Figure 3 As shown, the specific steps are as follows:

[0090] f. Through step b and step c, the keyword thesaurus pulled out by step C is digitized to obtain candidate nodes, and these candidate nodes are proofread and filtered in Baidu Encyclopedia, and the cleaned knowledge point keywords are formed into a knowledge point keyword thesaurus for this course; it serves as an important basis for the tag recommendation required by the packaged resource package.

[0091] These knowledge point keyword lexicon also constitute the nodes of the knowledge graph of this course;

[0092] Associate the nodes of the knowledge graph of this course with the main labels of the resource packages obtained in step B; since the nodes of the knowledge graph of this course and the main labels of the resource packages are both from the knowledge point keyword library of the course, that is, they belong to the same attribute, so when the teacher inputs the textbook information and the corresponding course knowledge graph is pulled, the corresponding resource package will be recommended when the relevant node of the knowledge graph is selected. After getting the experimental course resource package recommendation, the teacher selects the resource package that best suits his or her wishes based on the matched resource packages;

[0093] Import physical machines, virtual machines, containers, clusters and other resources into the system, and create resource packages for each knowledge point that requires on-the-job operation in the experimental course. The resource packages include experimental templates, experimental environments, experimental reports and operation documents. The experimental templates contain selected operating systems and software configurations, such as Figure 5 shown.

[0094] The experimental environment contains virtual hardware facilities that match the experimental template, including CPU, memory, and hard disk specifications, such as Figure 6 shown.

[0095] g. After selecting the resource package, it will be combined with the knowledge graph built based on the course to form an experimental course.

[0096] Teachers can also customize or modify the content of resource packages, such as modifying experimental templates and experimental environments, or experimental reports and operation documents. This supports automatic generation of knowledge graphs and experimental procedures that match the selected textbooks for teachers, and allows teachers to fully control the experimental environment templates and save customized operations after customization is completed.

[0097] After formulating the experimental course, the teacher can upload the course to the system. Students can conduct autonomous and independent experimental operations, and the experimental environment is relatively stable, the experimental data is relatively safe, and the import and export of experimental operation reports are supported. At the same time, it supports teachers to provide online guidance and online approval for students' experimental operations, and students can also view their own grades in the system.

[0098] Example 3

[0099] A course automatic construction system for information professional experiments, used to implement the course automatic construction method for information professional experiments described in Example 1 or 2, including a knowledge graph construction unit, a label processing unit, a keyword vocabulary matching unit and an experimental course construction unit; the knowledge graph construction unit is used to implement step A; the label processing unit is used to implement step B; the keyword vocabulary matching unit is used to implement step C; the experimental course construction unit is used to implement step D.

Claims

1. A method for automatically building experimental courses for information majors, characterized in that: The steps include: (1) Initialization; A. Crawling online information course textbooks and books information, building a keyword thesaurus and knowledge graph based on a course’s knowledge points; B. Encapsulate the resources, recommend tags to the resource packages through the knowledge point keyword thesaurus obtained in step A, and perform tag processing; (2) Build experimental courses based on user needs: C. Pull out the keyword database of the course knowledge points of the textbook according to the textbook information provided by the user; D. Construct the experimental course according to the keyword vocabulary extracted by matching in step C; In step B, the resources are packaged, and the knowledge point keyword thesaurus obtained in step A is used to recommend tags for the resource package, and tag processing is performed, which means: Package the configured experimental template, experimental environment, experimental report and operation documents. The experimental template includes the selected operating system and software configuration; the experimental environment includes virtual hardware facilities that match the experimental template, including CPU, memory, and hard disk specifications; for each configured experimental template and experimental environment, there are corresponding experimental reports and operation documents; Label the resource package that contains the packaged virtual software package, the experimental report and the operation document. The label content contains various information about the software package, including the knowledge points involved in the experiment, the memory size, the operation content targeted, and the estimated duration of the experiment. Among them, the main label of the resource package is recommended by searching the knowledge point keyword library obtained during the catalog digitization process; the sub-label content is the specific information of the resource package; In step D, the experimental course is constructed according to the course knowledge graph pulled out by matching in step C, which specifically includes the following steps: f. Through step b and step c, the candidate nodes are obtained by digital processing of the keyword word library pulled out by step C, and these candidate nodes are proofread and filtered in Baidu Encyclopedia, and the cleaned knowledge point keywords are formed into a knowledge point keyword word library for this course; These knowledge point keyword lexicon also constitute the nodes of the knowledge graph of this course; Associate the nodes of the knowledge graph of this course with the main label of the resource package obtained in step B; g. After selecting the resource package, it is combined with the knowledge graph built based on the course to form an experimental course.

2. According to claim 1, a method for automatically building an experimental course for information majors is characterized in that: In step A, crawling online information course textbooks and book information, building a keyword vocabulary and knowledge graph based on a course knowledge point, including the following steps: a. Perform web crawlers to obtain information on textbooks related to electronic information majors and form a database of textbooks related to electronic information majors. The textbooks related to electronic information majors include the titles, versions, publishers, authors, courses, catalog information and course knowledge maps of textbooks, which are divided by subject. b. Perform character recognition on the catalog of teaching materials under the same course and implement digitization. On the basis of the digitized structure, limit the length of the catalog bar and set a threshold. The catalog bar longer than the threshold is defined as a long catalog bar, otherwise, it is defined as a short catalog bar, thus dividing the long catalog bar and the short catalog bar; Perform part-of-speech tagging on the long directory according to the natural language processing method to obtain a part-of-speech array, and extract a batch of candidate nodes from the obtained part-of-speech array according to the part-of-speech rules; The short-term directory uses part-of-speech rules to extract another batch of candidate nodes; c. Identify the authenticity and reliability of the two batches of candidate nodes obtained in step b in Baidu Encyclopedia, complete filtering, and obtain a knowledge point keyword thesaurus based on a course; d. For a course’s knowledge point keyword thesaurus, identify these candidate nodes based on the hierarchical structure of the directory in Baidu Encyclopedia to form a hierarchical relationship, and use the same-level structure of the directory to form a parallel relationship; The primary skeleton of a knowledge graph based on a course is formed by the hierarchical structure and the parallel structure; e. The parallel relationship that appears in the primary skeleton of the knowledge graph obtained in step d is further divided into upper and lower structures, that is, the parallel relationship is distinguished into strong and weak ones, the number of times the two parts of the nodes in the parallel relationship appear is counted, and the absolute frequency of occurrence, that is, the number of times the two parts of the nodes in the parallel relationship appear, is sorted. At the same time, a threshold is set. Nodes with an absolute frequency greater than the threshold are defined as strong parallel relationships, otherwise, they are defined as weak parallel relationships. This cycle is repeated, sorting and screening are performed, and the redefined nodes are supplemented to the primary skeleton according to the results of the distinction, and finally a complete knowledge graph based on a course is obtained.

3. The automatic construction method for information professional experimental courses according to claim 2 is characterized in that: In step b, the threshold is set to 9.

4. The automatic construction method for information professional experimental courses according to claim 1 is characterized in that: According to the natural language processing method, the long directory is tagged with parts of speech to obtain a part-of-speech array. A batch of candidate nodes are extracted from the obtained part-of-speech array according to the part-of-speech rules. The specific implementation process is as follows: Perform word segmentation and part-of-speech tagging on the long directory to obtain a part-of-speech array, and use the part-of-speech rules to extract a batch of candidate nodes from the obtained part-of-speech array; ① Use the natural language processing tool FudanNLP to segment the long directory and mark the part of speech, split the long directory according to the comma and the parallel conjunctions with the part of speech as conjunctions, and split a long directory into a string array; ② For each substring N[i] in each split string array N, form a part-of-speech array of <entry, part-of-speech> for each substring N[i]; ③ Merge adjacent elements in the part-of-speech array. In the merging process, different merging orders are used for strings at different positions in the string array. That is, when processing the part-of-speech array of the first substring N[0] in the string array N, consecutive adjectives and nouns are merged into one word from back to front. At the same time, the part-of-speech of the last word in the part-of-speech array of N[0] is used as the benchmark part-of-speech. The benchmark part-of-speech is used to improve the accuracy of merging the part-of-speech array of the next string in the string array N. When processing the part-of-speech array of the second substring N[1] and each subsequent substring, the following method is used: if the benchmark part-of-speech is a noun, then the part-of-speech array of each string after N[1] is matched from front to back until the last part-of-speech is a noun to form a word, otherwise, no result is returned; if the benchmark part-of-speech is not a noun or a marking symbol, then in the part-of-speech array of each string after N[1], when the first part-of-speech matches the benchmark part-of-speech, a word is formed, otherwise, no result is returned; The final word is the candidate node.

5. The automatic construction method for information professional experimental courses according to claim 1 is characterized in that: In step e, the threshold is set to 20.

6. The automatic construction method for information professional experimental courses according to claim 1 is characterized in that: In step C, the knowledge point keyword thesaurus of the textbook or similar course textbooks is pulled out based on the textbook information provided by the user. Specifically, after the user enters the textbook name, one or more textbook information is retrieved from the electronic information textbook information library, including the textbook name, textbook version, publisher, author name, course, catalog information, and course knowledge graph information. The user selects the textbook information item he needs, and the textbook provided by the user is matched with the course knowledge graph.

7. A course automatic construction system for information professional experiments, characterized in that: A method for automatically building a course for information-related professional experiments according to any one of claims 1 to 6, comprising a knowledge graph construction unit, a label processing unit, a keyword vocabulary matching unit, and an experimental course construction unit; The knowledge graph construction unit is used to implement step A; the label processing unit is used to implement step B; the keyword vocabulary matching unit is used to implement step C; and the experimental course construction unit is used to implement step D.

Images