Large model enhanced cross-community knowledge graph construction method and system

Abstract

The invention relates to the technical field of large language models and knowledge maps, in particular to a large model enhanced cross-community knowledge map construction method and system. The method comprises the following steps of: extracting a knowledge triple from a Gialub community data set, and constructing a Gialub community knowledge triple of a first attribute triple and a first relationship triple; a knowledge triple is extracted from the StackOverflow community data set, and a StackOverflow community knowledge triple of a second attribute triple and a second relation triple is constructed; the method comprises the following steps: extracting concept hypernyms and hyponyms from a classification system of a Wikipedia data set, and constructing a Wikipedia community knowledge triple of a relationship between the hypernyms and the hyponyms; and a large model is adopted, the Githhub community knowledge triad and the StackOverflow community knowledge triad are matched with the Wikipedia community knowledge triad, alignment and fusion of the Githhub community knowledge and the StackOverflow community knowledge are completed, and a cross-community knowledge graph is obtained.

Description

Technical Field

[0001] The present invention relates to the field of large language models and knowledge graph technology, and in particular to a large model-enhanced cross-community knowledge graph construction method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the rapid development of the Internet, the diversity and complexity of data sources are increasing. Traditional knowledge graphs are gradually unable to meet the needs of cross-domain information integration, intelligent recommendation and deep reasoning. Therefore, the concept of cross-community knowledge graphs came into being, involving the sharing and integration of knowledge between different communities, industries and fields. This development process not only promotes the seamless interconnection of knowledge, but also provides support for intelligent applications in multiple fields.

[0004] The construction of cross-community knowledge graphs is of great significance, mainly reflected in the following aspects: (1) With the advancement of science and technology, the boundaries between different fields are becoming increasingly blurred, and the demand for cross-field research and application is growing. For example, knowledge in multiple fields such as medicine, finance, and technology needs to be integrated to achieve comprehensive decision support and risk control; (2) The diverse needs of users in information acquisition and application mean that a single community knowledge graph is difficult to fully meet the different expectations of users; (3) In today's booming development of big data and artificial intelligence, cross-community knowledge graphs can not only provide rich knowledge background, but also support better training of intelligent algorithms, significantly improving the performance of the system in complex scenarios. In this context, building a cross-community knowledge graph has become an important means to improve the efficiency of information utilization and realize intelligent services.

[0005] Although the construction of cross-community knowledge graphs has broad prospects, it still faces many challenges in practical applications. First, the knowledge representation methods and standards of different communities may be highly inconsistent, resulting in compatibility issues when integrating information. Secondly, the privacy protection and security requirements within each community also create obstacles to knowledge sharing, and some sensitive data may not be effectively used. In addition, the diversity of data sources makes it difficult to ensure the accuracy and completeness of knowledge. False or outdated information not only affects the quality of knowledge graph construction, but may also lead to decision-making errors in subsequent intelligent applications. Summary of the invention

[0006] In order to solve the technical problems faced in the process of building a cross-community knowledge graph, such as the highly inconsistent knowledge representation methods and standards, and the difficulty of integrating knowledge from different sources, the present invention provides a large model-enhanced cross-community knowledge graph construction method and system, a large language model (referred to as a large model, such as ChatGPT, Wenxinyiyan, etc.), which has powerful information understanding and abstraction capabilities, can process a variety of texts and structured data, and can help knowledge between different communities to perform effective semantic mapping and integration. In addition, the powerful reasoning ability of the large model can filter out high-quality information in retrieval, thereby reducing the negative impact caused by data quality problems. Combining these technologies, it can not only improve the data integration capabilities of the cross-community knowledge graph, but also realize flexible intelligent analysis and application, and promote collaborative innovation in various fields. Therefore, with the help of large models, the construction of cross-community knowledge graphs will become more efficient and accurate, providing strong support for the multi-level needs of the future smart society.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] The first aspect of the present invention provides a large-model enhanced cross-community knowledge graph construction method.

[0009] A large model-enhanced cross-community knowledge graph construction method, including:

[0010] Get the Github community dataset, StackOverflow community dataset, and Wikipedia dataset;

[0011] Extract knowledge triples from the Github community dataset and construct Github community knowledge triples of first attribute triples and first relationship triples; extract knowledge triples from the StackOverflow community dataset and construct StackOverflow community knowledge triples of second attribute triples and second relationship triples; extract concept hypernyms and hyponyms from the classification system of the Wikipedia dataset and construct Wikipedia community knowledge triples of the relationship between hypernyms and hyponyms;

[0012] Using a large model, we match the Github community knowledge triples and StackOverflow community knowledge triples with the Wikipedia community knowledge triples, align and integrate the Github community knowledge and StackOverflow community knowledge, and obtain a cross-community knowledge graph.

[0013] Furthermore, the first attribute triplet includes attribute triplet corresponding to the user entity, code repository entity, Issue entity and subject entity, and the first relationship triplet includes relationship triplet among the user entity, code repository entity, Issue entity and subject entity.

[0014] Furthermore, the process of obtaining the user entity, code repository entity, issue entity, and subject entity includes:

[0015] By constructing the URL of the user's GitHub personal page, sending a request and obtaining the HTML content of the page; parsing the HTML content, obtaining the user information, and constructing the user entity;

[0016] Access the user's code repository list page, use a request to obtain the content of the page, parse it, obtain the code repository information, and build a code repository entity;

[0017] For each code repository, construct the URL of its issue page, send a request to obtain the content of the page, parse the issue information, and construct the Issue entity;

[0018] By constructing the URL of the theme page of the code repository, sending a request to obtain the theme information related to the repository, and constructing the theme entity.

[0019] Furthermore, the second attribute triples include attribute triples corresponding to the user entity, question entity, reply entity and tag entity, and the second relationship triples include relationship triples among the user entity, question entity, reply entity and tag entity.

[0020] Furthermore, the Wikipedia community knowledge triples include the relationship between concept hypernyms and hyponyms in the classification system.

[0021] Furthermore, the large model is used to match the Github community knowledge triples and the StackOverflow community knowledge triples with the Wikipedia community knowledge triples, thereby completing the alignment and fusion of the Github community knowledge and the StackOverflow community knowledge; the method includes:

[0022] Based on the Github community knowledge triples, a large model is used to construct a first prompt word containing contextual semantic information; based on the first prompt word, a concept hypernym / concept hyponym matching the first prompt word is selected from the Wikipedia community knowledge triples;

[0023] Based on the StackOverflow community knowledge triples, a large model is used to construct a second prompt word containing contextual semantic information; based on the second prompt word, a concept hypernym / concept hyponym matching the second prompt word is selected from the Wikipedia community knowledge triples;

[0024] Based on the concept hypernyms / concept hyponyms that match the first prompt word and the concept hypernyms / concept hyponyms that match the second prompt word, a large model is used to obtain the alignment matching results, thereby realizing the alignment and fusion of Github community knowledge and StackOverflow community knowledge.

[0025] The second aspect of the present invention provides a large-model enhanced cross-community knowledge graph construction system.

[0026] A large-model-enhanced cross-community knowledge graph construction system, including:

[0027] The data source acquisition module is used to obtain the Github community dataset, StackOverflow community dataset, and Wikipedia dataset;

[0028] The knowledge triple extraction module is used to extract knowledge triples from the Github community dataset and construct Github community knowledge triples of first attribute triples and first relationship triples; extract knowledge triples from the StackOverflow community dataset and construct StackOverflow community knowledge triples of second attribute triples and second relationship triples; extract concept hypernyms and hyponyms from the classification system of the Wikipedia dataset and construct Wikipedia community knowledge triples of the relationship between hypernyms and hyponyms;

[0029] The large-model enhanced entity alignment and fusion module is used to adopt the large model to match the Github community knowledge triples and the StackOverflow community knowledge triples with the Wikipedia community knowledge triples, complete the alignment and fusion of the Github community knowledge and the StackOverflow community knowledge, and obtain a cross-community knowledge graph.

[0030] A third aspect of the present invention provides a computer-readable storage medium.

[0031] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the large-model-enhanced cross-community knowledge graph construction method as described in the first aspect above.

[0032] A fourth aspect of the present invention provides a computer device.

[0033] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for constructing a cross-community knowledge graph enhanced by a large model as described in the first aspect above are implemented.

[0034] A fifth aspect of the present invention provides a computer program product or a computer program.

[0035] The present invention provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the large model-enhanced cross-community knowledge graph construction method as described in the first aspect above.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention provides a large-model enhanced cross-community knowledge graph construction method and system. First, the information such as users (User), code warehouses (Project), problem discussions (Issue), topics (Topic) in the Github community is deeply analyzed to extract attribute triples and relationship triples between entities; similarly, attribute triples related to users (User), questions and answers (Question-Answer), and tags (Tag) in the StackOverflow community and relationship triples between entities are extracted; in addition, based on the Wikipedia encyclopedia, the concepts of encyclopedia entries and their hierarchical relationships between hyponyms and hyponyms are extracted to construct a concept classification system involving multiple fields, provide a classification framework for the knowledge graph, and improve the hierarchical system of knowledge triples in the cross-community knowledge graph on the basis of connecting cross-community knowledge triples; finally, a large-model prompt word is constructed based on the knowledge triples of the Github and StackOverflow communities, and then it is input into the large model to obtain the alignment matching result output by the large model, and finally the cross-community knowledge fusion is realized to obtain the final cross-community knowledge graph. Taking into account the difficulty of aligning and fusing knowledge from different communities, the present invention introduces a large model-enhanced entity alignment and fusion strategy, and utilizes the powerful semantic understanding ability of the large model to improve the accuracy of alignment and fusion.

[0038] Based on two different types of knowledge communities, Github and StacOverflow, this paper proposes a cross-community knowledge graph construction method. On the basis of cross-community knowledge extraction, the taxonomy system in Wikipedia is used to connect cross-community knowledge triples to improve the hierarchy of the knowledge system in the cross-community knowledge graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0040] Figure 1 It is a flowchart of the large model enhanced cross-community knowledge graph construction method shown in the present invention;

[0041] Figure 2 is an example diagram of the Wikipedia classification system shown in the present invention;

[0042] Figure 3 is an example (partial) diagram of a cross-community knowledge graph shown in the present invention;

[0043] Figure 4 It is a structural diagram of the large model enhanced cross-community knowledge graph construction system shown in the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0046] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0047] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and systems according to various embodiments of the present disclosure. It should be noted that each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of a code may include one or more executable instructions for implementing the logical functions specified in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the flowchart and / or block diagram, and the combination of boxes in the flowchart and / or block diagram can be implemented using a dedicated hardware-based system that performs a specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0048] Embodiment 1

[0049] like Figure 1 As shown, this embodiment provides a large model enhanced cross-community knowledge graph construction method. This embodiment uses the method applied to the server as an example. It can be understood that the method can also be applied to terminals, and can also be applied to terminals, servers and systems, and implemented through the interaction between terminals and servers. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network servers, cloud communications, middleware services, domain name services, security services CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected by wired or wireless communication, and this application is not limited here. This embodiment relies on the Shandong Provincial Key Laboratory of Artificial Intelligence Application for People's Livelihood Services and the Shandong Provincial General Artificial Intelligence Future Industry Laboratory, and the method includes the following steps:

[0050] Step 1: Extract knowledge triples from the Github community. By constructing the URL of the user's GitHub personal page, send a request and obtain the HTML content of the page; access the user's code repository list page, use a request to obtain the content of the page and then parse it; for each code repository, construct the URL of its issue page and send a request to obtain the content of the page; by constructing the URL of the code repository's topic page, send a request to obtain topic information related to the repository; according to the relationship between each type of knowledge entity, construct it into a triple form.

[0051] Specifically, as shown in Tables 1 and 2, knowledge triples are extracted from the Github community.

[0052] Table 1 Github community knowledge triple entity

[0053]

[0054] Table 2 Github community knowledge triple relationship

[0055]

[0056] Step 1.1: By constructing the URL of the user's GitHub personal page, send a request and obtain the HTML content of the page. After parsing the HTML, extract the user's basic information, such as user ID, user name, and personal profile. When parsing, you can find the corresponding HTML tags to obtain the required data and organize it into the form of user attribute triples, as shown below:

[0057] <User,UserID,U0001>

[0058] <User,UserName,Jack>

[0059] <User,Description,“He is…”>

[0060] Step 1.2: Access the user's code repository list page, use a request to obtain the content of the page, and then parse it. Extract detailed information about the repository, including: repository ID (usually obtained through the URL), repository name, repository introduction, README file content, number of stars, number of forks, and other statistical information, and organize them into the form of code repository (Project) attribute triples, as shown below:

[0061] <Project,ProjectID,P0001>

[0062] <Project,Title,Django>

[0063] <Project,Introduction,“Django is a Python web framework…”>

[0064] Step 1.3: For each code repository, construct the URL of its issue page and send a request to obtain the content of the page. When parsing the issue information, extract the title, status (open or closed), and content of each issue from the HTML and organize them into the form of an Issue attribute triple, as shown below:

[0065] <Issue,IssueID,I0001>

[0066] <Issue,Title,“Please fix OutOfMemoryException”>

[0067] <Issue,State,Closed>

[0068] Step 1.4: Send a request to obtain topic information related to the repository by constructing the URL of the repository's topic page. Extract the name and description of each topic and organize them into a triple of topic attributes, as shown below:

[0069] <Topic,TopicID,T0001>

[0070] <Topic,TopicName,Deep Learning>

[0071] <Topic,TopicIntro,“Deep learning is an AI function…”>

[0072] Step 1.5: As shown in Table 2, according to the relationship between each type of knowledge entity, organize them into the form of relationship triples, as shown in the following example:

[0073] <User,Relation,Project>

[0074] <User,Relation,Issue>

[0075] <Project,Relation,Issue>

[0076] <Project,Relation,Topic>

[0077] The above attribute and relationship triples together constitute the knowledge triples from Github.

[0078] Step 2: Extract knowledge triples from the StackOverflow community. Construct the URL of the Stack Overflow personal page, send a request and obtain the HTML content of the page; visit the user's question page, construct the URL and send a request to obtain the content of the page; for each question, visit the detailed page of the question and obtain relevant information about the reply; extract label information on the question page; construct it into triples based on the relationship between each type of knowledge entity.

[0079] As shown in Tables 3 and 4, knowledge triples are extracted from the StackOverflow community.

[0080] Table 3 StackOverflow community knowledge triple entity

[0081]

[0082] Table 4 StackOverflow community knowledge triple relationship

[0083]

[0084] Step 2.1: First, construct the URL of the user's StackOverflow personal page. Then, send a request and get the HTML content of the page. After parsing the HTML, the following information can be extracted: User ID: usually found in the URL of the user's page; Username: extract the user's name from the page; Credit: the user's credit score will be displayed on the page; Number of visits: the number of times the user's page has been visited; Number of likes: the total number of likes the user has received; Number of dislikes: the total number of dislikes the user has received; Last login time: usually found in the user information section and organized into the form of a user attribute triple, as shown below:

[0085] <User,UserID,U0001>

[0086] <User,DisplayName,John>

[0087] <User,Reputation,2>

[0088] Step 2.2: Get the question information raised by the user and visit the user's question page. Construct a URL and send a request to get the content of the page. After parsing the HTML, extract the relevant information of each question, including: Question ID: usually found in the URL of the question; Title: the title of the question; Score: the score of the question; Question content: the detailed content of the question; Views: the number of times the question has been viewed; Replies: the number of replies under the question; Comments: the number of comments on the question; Favorites: the number of times the question has been favorited, and organize it into the form of a triple of question attributes, as shown below:

[0089] <Question,QuestionID,Q0001>

[0090] <Question,Title,“How to toggle the visibility…”>

[0091] <Question,Score,90>

[0092] Step 2.3: For each question, you can further extract the reply information under it. Visit the detailed page of the question to obtain the relevant information of the reply, including: reply ID: usually found in the URL of the reply; score: the score of the reply; reply content: the detailed content of the reply; number of comments: the number of comments under the reply, and organize it into the form of a triplet of answer attributes, as shown below:

[0093] <Answer,AnswerID,A0001>

[0094] <Answer,Score,90>

[0095] <Answer,Body,“You can open the View-> Appearance…”>

[0096] Step 2.4: Extract tags on the page of the question. Each tag usually contains the following information: tag ID; Name: the name of the tag; WikiUrl: the Wikipedia page link of the tag, and is organized into the form of a tag attribute triple, as shown below:

[0097] <Tag,TagID,T0001>

[0098] <Tag,TagName,[javascript]>

[0099] <Tag,WikiUrl,“… / tagged / javascript…”>

[0100] Step 2.5: As shown in Table 4, according to the relationship between each type of knowledge entity, organize them into the form of relationship triples, as shown in the following example:

[0101] <User,Relation,Question>

[0102] <User,Relation,Answer>

[0103] <Question,Relation,Answer>

[0104] <Question,Relation,Tag>

[0105] The above attribute and relationship triples together constitute the knowledge triples from StackOverflow.

[0106] Step 3: Extract concept hypernyms and hyponyms from the Wikipedia classification system. For each concept, construct a page URL; send a request to the target URL to obtain the HTML content of the page and parse it; in the parsed HTML document, find specific tags and class names to extract the required hypernym and hyponym information; construct them into triples based on the relationship between hypernyms and hyponyms.

[0107] like Figure 2 As shown, concept hypernyms and hyponyms are extracted from the Wikipedia classification system.

[0108] Step 3.1: For each concept, construct a page URL format: https: / / zh.wikipedia.org / wiki / Concept name;

[0109] Step 3.2: Send an HTTP GET request to the target URL to obtain the HTML content of the page and parse it to find the part containing hypernym and hyponym information. Hypernyms often refer to broader classifications, while hyponyms refer to more specific categories. According to the relationship between hypernyms and hyponyms, they are organized into the form of relational triples, as shown below:

[0110] <Machine Learning,Relation,AI Technology>

[0111] <Natural Language Process,Relation,Transformer>

[0112] <Computer Vision,Relation,Convolutional Neural Network>

[0113] Step 4: Use the big model to align and fuse knowledge triples across communities. Based on the knowledge triples extracted from Github in step 1, use the big model to match the "topic" entity with the conceptual hypernym or hyponym extracted from the Wikipedia classification system; based on the knowledge triples extracted from StackOverflow in step 2, use the big model to match the "label" entity with the conceptual hypernym or hyponym; through the above process, the alignment and fusion of knowledge across Github and StackOverflow communities can be achieved to form the final cross-community knowledge graph.

[0114] Using large models for entity alignment and fusion of cross-community knowledge triples.

[0115] Step 4.1: Based on the knowledge triples extracted from Github in step 1, use the big model to build prompt words containing contextual semantic information, match the "topic" entity with the concept hypernym or hyponym extracted from the Wikipedia classification system, and organize them into the form of relation triples. The specific process is as follows:

[0116] Given a Github topic knowledge triple:

[0117] <Topic,TopicID,T0001>

[0118] <Topic,TopicName,Deep Learning>

[0119] <Topic,TopicIntro,“Deep learning is an AI function…”>

[0120] Tips for building a large model:

[0121] The name of the "topic" entity in Github is "Deep Learning", the introduction is "Deep learning is an AI function...", and the list of hyponyms of the concept in Wikipedia is<Machine Learning,Relation,AITechnology> ,<Natural Language Process,Relation,Transformer> ,<Computer Vision,Relation,Convolutional Neural Network> , please select the concept that best matches the subject entity from the Wikipedia concept candidate set [DeepLearning, Natural Language Process, Computer Vision].

[0122] Large model output: According to semantic reasoning and matching, the concept that best matches the subject entity is "MachineLeaning".

[0123] Step 4.2: Based on the knowledge triples extracted from StackOverflow in step 2, use the big model to build prompt words containing contextual semantic information, match the "label" entities with the concept hypernyms or hyponyms, and organize them into the form of relation triples. The specific implementation process is similar to step 4.1.

[0124] Step 4.3: Through the above process, we can achieve the alignment and integration of knowledge across Github and StackOverflow communities to form the final knowledge graph. Some of the contents are as follows: Figure 3 shown.

[0125] Embodiment 2

[0126] like Figure 4 As shown, this embodiment provides a large model enhanced cross-community knowledge graph construction system, including:

[0127] The data source configuration module is used to define and manage various data sources required by the system, including community websites, databases, and open data sets.

[0128] In some embodiments, the data source configuration module is also used to allow users to select different types of data sources according to actual needs, and configure their access permissions and data formats to ensure the efficiency and accuracy of subsequent data collection.

[0129] In some embodiments, the data source configuration module is also used to provide a preview function of the data source to help users understand the data structure and content in different sources.

[0130] The crawler parameter setting module allows users to flexibly set crawler parameters according to the characteristics of the target data source, such as crawling frequency, depth, and scope. Users can specify specific information types or content, such as news, forum posts, or social media dynamics.

[0131] In some embodiments, the crawler parameter setting module is also used to integrate crawler optimization strategies to ensure that the efficiency and success rate of the crawler are maximized while complying with the usage specifications of the data source.

[0132] The knowledge triple extraction module is used to identify and extract structured information from crawled data to form knowledge triples. This module uses natural language processing and machine learning technology to customize corresponding extraction strategies and algorithms based on the characteristics of different communities, identify community-specific information, and improve the accuracy and completeness of knowledge extraction.

[0133] The large model-enhanced entity alignment and fusion module is used to use the large model to perform entity alignment and data fusion across communities. By performing semantic similarity analysis and context understanding on entities from different sources, the module can accurately identify the same or similar entities in different communities and fuse them to eliminate redundant and inconsistent information.

[0134] The knowledge graph storage module is used to store the sorted and integrated knowledge triples in a graph database (such as the common graph database Neo4j), ensuring that the knowledge graph can support complex queries and expansions. At the same time, the module also provides data indexing, version control, and permission management functions, so that the knowledge graph can not only quickly respond to user query requests, but also be easily updated and maintained.

[0135] Embodiment 3

[0136] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the large-model enhanced cross-community knowledge graph construction method as described in the above-mentioned embodiment 1 are implemented.

[0137] Embodiment 4

[0138] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the large-model-enhanced cross-community knowledge graph construction method as described in the first embodiment above are implemented.

[0139] Embodiment 5

[0140] This embodiment provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the large model-enhanced cross-community knowledge graph construction method described in the first embodiment.

[0141] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0142] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0143] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0144] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0145] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0146] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A large model-enhanced cross-community knowledge graph construction method, characterized in that: include: Get the Github community dataset, StackOverflow community dataset, and Wikipedia dataset; Extract knowledge triples from the Github community dataset and construct Github community knowledge triples of first attribute triples and first relationship triples; extract knowledge triples from the StackOverflow community dataset and construct StackOverflow community knowledge triples of second attribute triples and second relationship triples; Extract concept hypernyms and hyponyms from the classification system of the Wikipedia dataset and construct Wikipedia community knowledge triples of the relationship between hypernyms and hyponyms; Using a large model, we match the Github community knowledge triples and StackOverflow community knowledge triples with the Wikipedia community knowledge triples, align and integrate the Github community knowledge and StackOverflow community knowledge, and obtain a cross-community knowledge graph.

2. The large model-enhanced cross-community knowledge graph construction method according to claim 1 is characterized in that: The first attribute triplet includes attribute triplet corresponding to the user entity, the code repository entity, the Issue entity and the subject entity, and the first relationship triplet includes the relationship triplet between the user entity, the code repository entity, the Issue entity and the subject entity.

3. The large model-enhanced cross-community knowledge graph construction method according to claim 2 is characterized in that: The process of obtaining user entities, code repository entities, issue entities, and subject entities includes: By constructing the URL of the user's GitHub personal page, sending a request and obtaining the HTML content of the page; parsing the HTML content, obtaining the user information, and constructing the user entity; Access the user's code repository list page, use a request to obtain the content of the page, parse it, obtain the code repository information, and build a code repository entity; For each code repository, construct the URL of its issue page, send a request to obtain the content of the page, parse the issue information, and construct the Issue entity; By constructing the URL of the theme page of the code repository, sending a request to obtain the theme information related to the repository, and constructing the theme entity.

4. The large model-enhanced cross-community knowledge graph construction method according to claim 1 is characterized in that: The second attribute triples include attribute triples corresponding to the user entity, the question entity, the reply entity and the tag entity, and the second relationship triples include relationship triples among the user entity, the question entity, the reply entity and the tag entity.

5. The large model-enhanced cross-community knowledge graph construction method according to claim 1 is characterized in that: The Wikipedia community knowledge triples include the relationship between the hypernyms and hyponyms of concepts in the classification system.

6. The large model-enhanced cross-community knowledge graph construction method according to claim 1 is characterized in that: The method adopts a large model to match the Github community knowledge triples and the StackOverflow community knowledge triples with the Wikipedia community knowledge triples, thereby completing the alignment and integration of the Github community knowledge and the StackOverflow community knowledge. The method includes: Based on the Github community knowledge triples, a large model is used to construct a first prompt word containing contextual semantic information; based on the first prompt word, a concept hypernym / concept hyponym matching the first prompt word is selected from the Wikipedia community knowledge triples; Based on the StackOverflow community knowledge triples, a large model is used to construct a second prompt word containing contextual semantic information; based on the second prompt word, a concept hypernym / concept hyponym matching the second prompt word is selected from the Wikipedia community knowledge triples; Based on the concept hypernyms / concept hyponyms that match the first prompt word and the concept hypernyms / concept hyponyms that match the second prompt word, a large model is used to obtain the alignment matching results, thereby realizing the alignment and fusion of Github community knowledge and StackOverflow community knowledge.

7. A large model-enhanced cross-community knowledge graph construction system, characterized in that: include: The data source acquisition module is used to obtain the Github community dataset, StackOverflow community dataset, and Wikipedia dataset; The knowledge triple extraction module is used to extract knowledge triples from the Github community data set and construct Github community knowledge triples of first attribute triples and first relationship triples; extract knowledge triples from the StackOverflow community data set and construct StackOverflow community knowledge triples of second attribute triples and second relationship triples; Extract concept hypernyms and hyponyms from the classification system of the Wikipedia dataset and construct Wikipedia community knowledge triples of the relationship between hypernyms and hyponyms; The large-model enhanced entity alignment and fusion module is used to adopt the large model to match the Github community knowledge triples and the StackOverflow community knowledge triples with the Wikipedia community knowledge triples, complete the alignment and fusion of the Github community knowledge and the StackOverflow community knowledge, and obtain a cross-community knowledge graph.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the large-model enhanced cross-community knowledge graph construction method as described in any one of claims 1-6 are implemented.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the large model enhanced cross-community knowledge graph construction method as described in any one of claims 1-6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, which, when executed by a processor, implements the steps in the large-model enhanced cross-community knowledge graph construction method as described in any one of claims 1-6.

Images