Knowledge graph construction method fusing large-scale language model

Abstract

The invention provides a knowledge graph construction method fused with a large-scale language model, and relates to the technical field of education, and the method is characterized in that the method comprises the following steps: 1, collecting Internet information through a text information capturing module and an information connection module, and labeling keywords; step 2, inputting a large amount of text information contents obtained in the step 1 into a text information classification module through wireless data, wherein the text information classification module is matched with a large language model to identify and evaluate phrase contents and keyword contents; 3, after character recognition judgment is completed in the step 2, information is uploaded to a storage module according to the classification condition; 4, performing information retrieval and management on the information sorted and stored in the step 3 by a knowledge point retrieval and combination module, and preparing a corresponding framework and a map at the same time; 5, after the whole framework is built, calling can be carried out through the Internet. The method has the advantages that knowledge points can be conveniently obtained and sorted, and the overall learning efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of educational technology, and in particular to a knowledge graph construction method integrating a large-scale language model. Background Art

[0002] With the development of information technology, more and more industries have begun to use language models to organize text materials in detail, so as to facilitate personnel to consult and check, and improve the efficiency of data acquisition and learning. However, in the technologies currently used in teaching and education, although knowledge points can be updated and improved quickly, the whole set of methods cannot be effectively combined with the established methods, resulting in the need to re-edit and produce new content, resulting in low overall information update efficiency and inability to meet existing usage needs. Summary of the invention

[0003] The purpose of the present invention is to provide a knowledge graph construction method integrating a large-scale language model, which solves the technical problem of low efficiency of information acquisition and change adjustment in the prior art, and achieves the technical effect of improving the efficiency of information verification and change adjustment.

[0004] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0005] A method for constructing a knowledge graph integrating a large-scale language model comprises the following steps: Step 1, collecting Internet information through a text information capture module and an information connection module, and combining a preset resource acquisition website to perform large-scale information collection, wherein the information collection will mark keywords;

[0006] Step 2: The large amount of text information obtained in step 1 will be input into the text information classification module through wireless data. The text information classification module will cooperate with the large language model to identify and evaluate the paragraph content and keyword content, and will identify and judge the specific meanings of the professional vocabulary and restricted vocabulary marked in the language paragraph, so as to obtain detailed and specific text content;

[0007] Step 3: After the text recognition and judgment in step 2 is completed, the information will be uploaded to the storage module according to the classification. The storage module will recognize the content entered at this time, and determine whether the storage module contains the same type or the same content, and screen and store it according to the actual content;

[0008] In step 4 and step 3, the information stored in the storage can be retrieved and managed by the knowledge point retrieval module. The knowledge point retrieval module can interact with the manual terminal and obtain the corresponding knowledge point framework. At this time, the knowledge point retrieval module will retrieve the corresponding related knowledge points according to the framework and insert them into the framework in sequence.

[0009] Step 5. After the overall framework is built, it can be accessed through the Internet.

[0010] As an improvement, when the text information is retrieved, the text information capture module in step 1 will evaluate it based on the knowledge category and actual scientific research results to determine whether the content has corresponding scientific basis or theoretical support. The text information capture module will also mark some irrelevant and repeated text in the paragraph to facilitate subsequent deletion and refinement.

[0011] As an improvement, the text information classification module in step 2 will back up the text content after recognition and processing, and can upload this information to the terminal for manual processing. The manual processing can update the large language model, and can identify and judge the newly generated vocabulary and enter it into the large language model to achieve more accurate information recognition and judgment.

[0012] As an improvement, the storage module in step 3 includes local information storage and cloud storage. When the text information enters the storage module, the filtering program will read and identify the corresponding keywords, and then preliminarily screen out the stored text data content based on the keywords for judgment. When the text content is completely consistent, the temporary record file will be directly deleted. When the text content similarity is above 50%, the corresponding paragraph details will be further obtained for judgment. When the semantics are similar, the temporary storage file will be deleted to reduce the cluttered information in the storage module.

[0013] As an improvement, the knowledge point retrieval combination module in step 4 can additionally beautify the information in the framework through the graph construction module, and link the stored information as notes by identifying text content and related knowledge points, so as to facilitate information retrieval of various contents.

[0014] As an improvement, the completed content in step 5 can be stored in conjunction with other information for comparison, and the specific description details can be manually adjusted and changed, and the corresponding modified content can be recorded.

[0015] The beneficial effects of the present invention are as follows: by setting up a text information capture module, a large amount of text information can be quickly acquired, and text paragraphs can be identified through keyword annotation, while redundant text can be deleted; by setting up a text information classification module, the text information can be further reviewed, and the storage module can be used to filter and fragment the text content, so that it can be quickly combined and matched to achieve a complete graph knowledge point summary. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flowchart of a method for constructing a knowledge graph that integrates a large-scale language model in the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] It should be noted that the terms "first" and "second" in this application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0019] like Figure 1As shown, a knowledge graph construction method integrating a large-scale language model includes the following steps: Step 1, collecting Internet information through a text information capture module and an information connection module, and combining a preset resource acquisition website to perform large-scale information collection, and the information collection will mark keywords; Step 2, a large amount of text information content obtained in Step 1 will be input into a text information classification module through wireless data, and the text information classification module will cooperate with the large language model to identify and evaluate the content of the paragraph and the keyword content, and will identify and judge the specific meanings of the marked professional vocabulary, limiting vocabulary, etc. in the language paragraph, so as to obtain detailed and specific text content; Step Step 3: After completing the text recognition judgment in step 2, the information will be uploaded to the storage module according to the classification situation. The storage module will identify the content entered at this time, and determine whether the storage module contains the same type or the same content, and screen and store it according to the actual content; Step 4: The information stored in step 3 can be retrieved and managed by the knowledge point retrieval combination module. The knowledge point retrieval combination module can interact with the manual terminal and obtain the corresponding knowledge point framework. At this time, the knowledge point retrieval combination module will retrieve the corresponding related knowledge points according to the framework situation and insert them into the framework in sequence; Step 5: After the overall framework is built, it can be retrieved through the Internet. After the overall framework is completed, other relevant information can also be further obtained through the terminal to expand the overall framework and realize the unified and rapid acquisition of multiple groups of information.

[0020] In the step 1, the text information capture module will evaluate the text information according to the knowledge category and actual scientific research results when accessing the text information, and judge whether the content has corresponding scientific basis or theoretical support. The text information capture module will also mark some irrelevant and repeated texts in the paragraph to facilitate subsequent deletion and refinement. In the step 2, the text information classification module will back up the text content after recognition and processing, and upload this information to the terminal for manual processing. The manual processing can update the large language model, and can identify and judge the newly generated vocabulary, and enter it into the large language model to achieve more accurate information recognition and judgment. In the step 3, the storage module includes local information storage and cloud storage. When the text information enters the storage module, the filtering program will read and identify the corresponding keywords, and then preliminarily screen the stored text data content according to the keywords for judgment. When the text content is completely consistent, the temporary record file will be directly deleted. When the text content similarity is more than 50%, the corresponding paragraph details will be further obtained for judgment. When the semantics are similar, the temporary storage file will be deleted to reduce the messy information in the storage module. The knowledge point retrieval combination module in step 4 can additionally beautify and typeset the information in the framework through the graph construction module, and link the stored information as notes by identifying the text content and the associated knowledge points, so as to facilitate the information consultation of various contents. The completed content in step 5 can be stored in conjunction with other information for comparison, and the specific description details can be adjusted and changed manually, and the corresponding modified content can be recorded. Each final draft after adjustment will be stored separately, and the overall modification data will be recorded to facilitate the subsequent review and verification of the content.

[0021] When in use, the operator can input the paper publication website and patent publication website in advance, so that the text information crawling module can obtain text content based on such authoritative websites. At this time, the text content will be marked according to the keywords and corresponding key positions. The keywords include professional terms, numbers and units, and the key positions include title content and font change positions, etc.; after the information is identified and marked, some redundant text will be deleted, and only the core solution or technical part will be retained for recording. Subsequently, such content will be evaluated based on the paragraph content and keyword semantics to determine the overall content classification, such as biological content, chemical content, etc., and then the subcategories will be screened again in the corresponding classification library according to the keywords for storage.

[0022] When retrieving, the operator can preset the category and related content of the required knowledge points, and then export the knowledge points directly from the storage module in accordance with the framework, and construct a graph in accordance with the framework structure. Subsequently, additional sub-category knowledge points can be expanded based on the graph to facilitate understanding and learning.

[0023] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for constructing a knowledge graph integrating a large-scale language model, characterized in that: The following steps are involved: Step 1: Collect Internet information through a text information capture module and an information connection module, and collect large-scale information in combination with a preset resource acquisition website, wherein the information collection will mark keywords; Step 2: The large amount of text information obtained in step 1 will be input into the text information classification module through wireless data. The text information classification module will cooperate with the large language model to identify and evaluate the paragraph content and keyword content, and will identify and judge the specific meanings of the professional vocabulary and restricted vocabulary marked in the language paragraph, so as to obtain detailed and specific text content; Step 3: After the text recognition and judgment in step 2 is completed, the information will be uploaded to the storage module according to the classification. The storage module will recognize the content entered at this time, and determine whether the storage module contains the same type or the same content, and screen and store it according to the actual content; In step 4 and step 3, the information stored in the storage can be retrieved and managed by the knowledge point retrieval module. The knowledge point retrieval module can interact with the manual terminal and obtain the corresponding knowledge point framework. At this time, the knowledge point retrieval module will retrieve the corresponding related knowledge points according to the framework and insert them into the framework in sequence. Step 5. After the overall framework is built, it can be accessed through the Internet.

2. According to the method for constructing a knowledge graph integrating a large-scale language model according to claim 1, it is characterized in that: When the text information is retrieved in step 1, the text information capture module will evaluate it according to the knowledge category and actual scientific research results to determine whether the content has corresponding scientific basis or theoretical support. The text information capture module will also mark some irrelevant and repeated text in the paragraph to facilitate subsequent deletion and refinement.

3. The method for constructing a knowledge graph integrating a large-scale language model according to claim 1, characterized in that: In step 2, the text information classification module will back up the text content after recognition and processing, and can also upload this information to the terminal for manual processing. The manual processing can update the large language model, and can also identify and judge the newly generated words and enter them into the large language model to achieve more accurate information recognition and judgment.

4. The method for constructing a knowledge graph integrating a large-scale language model according to claim 1, characterized in that: The storage module in step 3 includes local information storage and cloud storage. When the text information enters the storage module, the filtering program will read and identify the corresponding keywords, and then preliminarily screen out the stored text data content based on the keywords for judgment. When the text content is completely consistent, the temporary record file will be directly deleted. When the text content similarity is above 50%, the corresponding paragraph details will be further obtained for judgment. When the semantics are similar, the temporary storage file will be deleted to reduce the cluttered information in the storage module.

5. The method for constructing a knowledge graph integrating a large-scale language model according to claim 1, characterized in that: The knowledge point retrieval combination module in step 4 can additionally beautify the information in the framework through the graph construction module, and link the stored information as notes by identifying text content and related knowledge points, so as to facilitate information retrieval of various contents.

6. The method for constructing a knowledge graph integrating a large-scale language model according to claim 1, characterized in that: The completed content in step 5 can be stored in conjunction with other information for comparison, and the specific description details can be manually adjusted and modified, and the corresponding modified content can be recorded.

Images