Academic industry knowledge graph fusion method and system based on large model

Abstract

The invention discloses an academic industry knowledge graph fusion method and system based on a large model, and is applied to the technical field of artificial intelligence, and the method comprises the steps: constructing an academic achievement knowledge graph based on academic achievement text data, and constructing an industrial knowledge graph based on industrial text data; converting the processed academic achievement attribute information by adopting a semantic vector model of a large model; adding labels to the academic achievement attributes represented by the vectors to serve as a training set, constructing and training a multi-label random forest model based on the training set, and classifying the academic achievements in the academic achievement knowledge graph to obtain an industrial field classification result of the academic achievements; according to the industrial field classification result of the academic achievement, carrying out industrial field division on scientific research talents to generate industrial labels; and calculating a correlation degree between the industrial label and the industrial knowledge graph, and completing fusion of the academic achievement knowledge graph and the industrial knowledge graph based on a correlation degree result.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and system for integrating academic and industrial knowledge graphs based on a large model. Background Art

[0002] Knowledge graph is a knowledge management technology with great application prospects. Currently, some platforms have launched technology services based on technology knowledge graphs and explored the combined application of knowledge graphs and technology services. However, there is almost no research on the use of large models to extract entities to build knowledge graphs, and the use of semantic vector models and multi-label random forest algorithms based on large models to achieve the division of scientific and technological achievements and talent fields.

[0003] Text classification refers to the process of automatically determining the text category based on the content of the text under a certain classification system. At present, the commonly used text classification technology is to classify the text by using the similarity between the content of the text to be classified and the existing text in each category, including KNN algorithm, Bayesian algorithm, support vector machine algorithm and concept reasoning network algorithm based on semantic network. However, its classification accuracy is limited by the feature extraction of the text, and with the popularity of large models, the semantic vector model derived from it can better understand the text semantic information and convert the text semantic information into fixed-dimensional features for subsequent calculations. At present, there is little research on the classification of scientific and technological texts using the semantic vector model of the large model combined with traditional classification methods, and there is a lack of clear classification standards and classification methods.

[0004] Knowledge graph fusion refers to the process of merging the same entity, attribute or relationship from different knowledge sources to form a complete, consistent and high-quality knowledge graph. Knowledge graph fusion can be achieved based on the semantic information or structural information of the knowledge graph using methods such as entity alignment and representation learning.

[0005] At present, there are some problems with the classification of industrial fields to which scientific and technological achievements belong. First, with the rapid development and cross-integration of science and technology, many scientific and technological achievements involve multiple industrial fields and are difficult to simply classify into a specific category. This leads to ambiguity and uncertainty in classification, which affects the accurate evaluation and effective utilization of scientific and technological achievements. Secondly, different institutions or experts may have differences in the division of industrial fields, and the lack of unified standards and norms makes it difficult for the classification results to be widely accepted and recognized. In terms of the integration of academic achievement knowledge graphs and industrial knowledge graphs, there is currently a lack of relevant technical exploration, and there is a separation between academic achievements, scientific research talents and specific technical fields and industrial fields, which makes it difficult to select talents or retrieve academic achievement information.

[0006] In order to overcome these shortcomings, this application proposes a method and system for integrating academic and industrial knowledge graphs based on a large model. Summary of the invention

[0007] The purpose of this application is to provide a method and system for the fusion of academic and industrial knowledge graphs based on a large model, aiming to solve the problems in the prior art of unclear division of academic achievements into industrial fields and complex knowledge graph fusion process.

[0008] To achieve the above objectives, this application provides the following technical solutions: This application provides a method for integrating academic and industrial knowledge graphs based on a large model, including: Acquire academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data; Performing text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and converting the processed academic achievement attribute information using a semantic vector model of a large model to obtain academic achievement attributes represented by vectors; Add labels to the academic achievement attributes represented by the vector as a training set, build and train a multi-label random forest model based on the training set, classify the academic achievements in the academic achievement knowledge graph, and obtain the industrial field classification results of the academic achievements; Based on the industry classification results of the academic achievements, scientific research talents are divided into industry fields and industry labels are generated; The correlation between the industry label and the industry knowledge graph is calculated, and the fusion of the academic achievement knowledge graph and the industry knowledge graph is completed based on the correlation result.

[0009] Furthermore, in the steps of obtaining academic achievement text data and industry text data, constructing an academic achievement knowledge graph based on the academic achievement text data, and constructing an industry knowledge graph based on the industry text data, the following steps are specifically included: Preprocessing and annotating the academic achievement text data, identifying the preprocessed academic achievement text data by fine-tuning the large model, extracting the preprocessed academic achievement text data by using prompt word engineering, and combining to obtain academic achievement entities; constructing an academic achievement knowledge graph based on the academic achievement entities; The industrial text data is preprocessed and annotated, the preprocessed industrial text data is recognized by fine-tuning the large model, the preprocessed industrial text data is extracted using prompt word engineering, and the industrial field entities are obtained by combining them; an industrial knowledge graph is constructed based on the industrial field entities.

[0010] Furthermore, in the step of performing text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, converting the processed academic achievement attribute information using the semantic vector model of the large model to obtain the academic achievement attributes represented by the vector, the following steps are specifically included: Perform text cleaning on the academic achievement attribute information, and use regular expressions to remove preset irrelevant characters; use the NLTK library to perform stem extraction on the academic achievement attribute information to obtain word roots; Through the semantic vector model of the large model, the preprocessed academic achievement attribute information is converted into the corresponding vector representation to obtain the academic achievement attributes represented by the vector.

[0011] Furthermore, in the step of adding labels to the academic achievement attributes represented by the vector as a training set, constructing and training a multi-label random forest model based on the training set, classifying the academic achievements in the academic achievement knowledge graph, and obtaining the industrial field classification results of the academic achievements, the following steps are specifically included: A multi-label random forest model is built based on the training set. Several decision trees are constructed during the training process, and a multi-label learning strategy is used for classification. After the training is completed, the trained multi-label random forest model is used to classify and predict the academic achievements in the academic achievement knowledge graph to obtain the industrial field classification results of the academic achievements.

[0012] Furthermore, the industrial knowledge graph divides the industrial field into four layers, and the academic achievement knowledge graph includes papers published by scientific research talents and patents invented.

[0013] This application provides an academic industry knowledge graph fusion system based on a large model, including: Graph construction module: obtain academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data; Feature extraction module: performs text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and uses the semantic vector model of the large model to convert the processed academic achievement attribute information to obtain the academic achievement attributes represented by vectors; Training module: adding labels to the academic achievement attributes represented by the vector as a training set, constructing and training a multi-label random forest model based on the training set, classifying the academic achievements in the academic achievement knowledge graph, and obtaining the industrial field classification results of the academic achievements; Division module: Based on the industry classification results of the academic achievements, scientific research talents are divided into industry fields and industry labels are generated; Graph fusion module: calculates the correlation between the industry label and the industry knowledge graph, and completes the fusion of the academic achievement knowledge graph and the industry knowledge graph based on the correlation result.

[0014] The present application provides a device, which includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing a method for fusion of academic and industrial knowledge graphs based on a large model; the processor is used to execute the program instructions stored in the memory to implement a method for fusion of academic and industrial knowledge graphs based on a large model.

[0015] The present application provides a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute a method for integrating academic and industrial knowledge graphs based on a large model.

[0016] This application provides a method and system for integrating academic and industrial knowledge graphs based on a large model, which has the following beneficial effects: (1) The large model can more accurately capture the contextual information in the text, thereby improving the accuracy of extraction. At the same time, the large model has stronger generalization ability and can effectively cope with diverse text content and structure, reducing the dependence on a large amount of annotated data, so that it can maintain good performance even when data is scarce. In addition, the large model can better handle unknown entities, improving the robustness of the system. (2) Existing methods for classifying academic achievements into industrial fields often rely on manual annotation or classification based on the institutions or journals to which the achievements belong. There is a lack of clear standards, and there is a lack of clear definitions and definitions of subordinate relationships for various industrial fields. This application provides a clear definition for field classification by integrating the academic achievement knowledge graph and the industrial knowledge graph. It uses a large-scale semantic vector model to analyze academic achievement text data, and combines a multi-label random forest algorithm to achieve automated academic achievement industry field classification, making classification more accurate and efficient. (3) By dividing academic achievements and scientific research talents into industrial fields, it is conducive to the realization of subsequent applications such as talent selection and industrial development planning, and enriches the information of the knowledge graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the process of a large-model-based academic-industry knowledge graph fusion method according to Example 1 of the present application; Figure 2 A schematic diagram of the structure of a method for integrating academic and industrial knowledge graphs based on a large model according to Example 1 of the present application; Figure 3 This is a schematic diagram of the structure of the academic achievement knowledge graph of Example 1 of this application; Figure 4This is a schematic diagram of the structure of the industry knowledge graph of Example 1 of the present application; Figure 5 This is a schematic diagram of the structure of an academic-industry knowledge graph fusion system based on a large model according to Example 2 of the present application; Figure 6 This is a schematic diagram of the device structure of Example 3 of the present application; Figure 7 This is a schematic diagram of the storage medium structure of Example 4 of the present application. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0019] 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 making creative work are within the scope of protection of this application.

[0020] Example 1 See also Figure 1 , is a flow chart of a method for integrating academic and industrial knowledge graphs based on a large model in Example 1 of the present application; the steps include: S1: Acquire academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data.

[0021] In this embodiment, the academic achievement text data is preprocessed and annotated, the preprocessed academic achievement text data is recognized by fine-tuning the large model, the preprocessed academic achievement text data is extracted using prompt word engineering, and the academic achievement entities are obtained by combination; and the academic achievement knowledge graph is constructed based on the academic achievement entities.

[0022] Preprocess and annotate the industrial text data, and identify the preprocessed industrial text data by fine-tuning the large model. Select a pre-trained model, such as BERT, GPT, RoBERTa, etc., based on the characteristics of the industrial text data and task requirements. Establish training sets, validation sets, and test sets based on industrial text data. Set fine-tuning parameters such as learning rate, batch size, training rounds, etc. according to task characteristics and model features. Further train the pre-trained model and optimize the pre-trained model by adjusting model weights and parameters. Use the validation set to evaluate the fine-tuned model, and adjust the model structure and parameters based on the evaluation results until satisfactory performance is achieved.

[0023] The preprocessed industrial text data is extracted using the prompt word project, and then combined to obtain industrial domain entities. Specifically, it includes: selecting the type, range, and prompt words of the extracted text information. Introducing variable elements into the prompt words to improve the reusability and flexibility of the prompt words. The designed prompt words are input into the fine-tuned model together with the preprocessed industrial text data. The model performs reasoning and extraction based on the prompt words and text data, generates industrial domain entities, and constructs an industrial knowledge graph based on the industrial domain entities.

[0024] S2: Perform text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and use the semantic vector model of the large model to convert the processed academic achievement attribute information to obtain the academic achievement attributes represented by vectors.

[0025] In this embodiment, the academic achievement attribute information is cleaned, and regular expressions are used to remove preset irrelevant characters, such as non-alphanumeric characters, punctuation marks, special symbols, etc., and to delete redundant spaces, line breaks and tabs to ensure the uniformity and neatness of the text format. The NLTK library is used to extract the academic achievement attribute information, and the prefix and suffix of the word are removed to obtain the root. It is helpful to normalize words of different forms to the same form.

[0026] Through the semantic vector model of a large model, such as the BGE model. Convert the preprocessed academic achievement attribute information into the corresponding vector representation to obtain the academic achievement attributes represented by the vector. Specifically, it includes: preprocessing the academic achievement attribute information, including data cleaning, format unification, removal of irrelevant information, etc., to ensure the quality and consistency of the input data. Build the software and hardware environment, and ensure the availability of computing resources; use the corresponding tools and libraries, such as the Transformers library to load the BGE model and its related word segmenters. Encode the preprocessed academic achievement attribute information through the word segmenter and convert it to the input format of the model; input the encoded text into the BGE model to calculate the corresponding semantic vector.

[0027] S3: Label the academic achievement attributes represented by the vector as a training set, build and train a multi-label random forest model based on the training set, classify the academic achievements in the academic achievement knowledge graph, and obtain the industrial field classification results of the academic achievements.

[0028] In this embodiment, a multi-label random forest model is constructed based on the training set, and the parameters of the random forest are set, such as the number of trees, the maximum depth, the minimum number of samples, etc.; and several decision trees are constructed on the training set. Each tree is trained by randomly extracting samples and features from the training set, and at the nodes of each decision tree, partitioning is performed according to the selected features until a preset stop condition is reached.

[0029] For multi-label classification tasks, a variety of strategies are used to handle them. The multi-label problem is converted into multiple binary classification problems, with each label corresponding to a binary classifier; or a label correlation-based method is used to consider the relationship between labels. In a random forest, each tree can adopt the same or different multi-label learning strategies.

[0030] On each decision tree, training is performed based on the training set to obtain the division rules of each node and the category labels of the leaf nodes. The prediction results of all decision trees are integrated to obtain the final prediction results. The trained multi-label random forest model is evaluated using the validation set or test set to calculate indicators such as accuracy, recall, and F1 score. The model parameters are adjusted according to the evaluation results to improve the model performance. After the training is completed, the academic achievements in the academic achievement knowledge graph are classified and predicted through the trained multi-label random forest model to obtain the industrial field classification results of the academic achievements.

[0031] S4: Based on the industrial field classification results of the academic achievements, scientific research talents are divided into industrial fields and industrial labels are generated.

[0032] In this embodiment, based on the industrial field classification results of the academic achievements published by scientific research talents, assuming that the talent has N academic achievements in total, and their achievements involve m industrial fields in total, if the number of achievements in a certain field is greater than N / m, a label of the industrial field is added to the talent.

[0033] S5: Calculate the correlation between the industry label and the industry knowledge graph, and complete the fusion of the academic achievement knowledge graph and the industry knowledge graph based on the correlation result.

[0034] In this embodiment, the correlation between the industry label and the industry knowledge graph is calculated by using a matching method. That is, by matching the label of the industry knowledge graph with the industry label, the correlation between the industry label and the label of the industry knowledge graph is calculated according to the position and number of the labels in the matched industry knowledge graph. The correlation here represents the degree of mastery of scientific research talents in the technical field.

[0035] Assuming that there are n third-level labels in the industry knowledge graph, and there are m labels associated with a scientific research talent, where m / n is greater than 1 / 2, the industry label of the previous level will be linked to the scientific research talent, otherwise only the industry label of the same level will be linked to it. At this time, the correlation between the industry label of the previous level and the talent is m / n, and the correlation between the industry label of the same level and the talent is 1. When m / n is equal to 1, or other correlations are 1, it means that the technical field has a direct correspondence with the scientific research talent or the scientific research talent has fully mastered all relevant technical fields in the field. Through the correlation calculation, a higher-level industry label is linked to the scientific research talent.

[0036] For example, the industry labels of a scientific research talent include "sentiment analysis", "machine translation", "knowledge graph", and "image segmentation". In the industry knowledge graph, there is a secondary classification label "natural language processing", and its sub-labels are "sentiment analysis", "machine translation", "knowledge graph", "named entity recognition", and "text classification"; there is a secondary classification label "computer vision", and its sub-labels are 5. Through matching, the industry label occupies more than half of the "natural language processing" sub-labels, so the "natural language processing" label is linked to the scientific research talent to form an associated node, and the correlation between the scientific research talent and "natural language processing" is 3 / 5. On the other hand, "image segmentation" belongs to the third-level classification and belongs to "computer vision". The correlation with "computer vision" is 1 / 5, which does not exceed the prescribed 1 / 2. It is considered that the correlation between the scientific research talent and "computer vision" is not enough, so only "image segmentation" is linked to the scientific research talent, and its correlation is 1. Through the calculation of the correlation, a higher-level industry label is established for the scientific research talent, and it is believed that the higher-level industry label can better highlight the contribution of scientific research talents in the industry. By establishing connections between academic achievements, industry labels, and scientific research talents, links are established between the two knowledge graphs to complete the knowledge graph fusion.

[0037] See also Figure 2 , which is a structural diagram of a method for integrating academic and industrial knowledge graphs based on a large model in Example 1 of the present application. Specifically, it includes: first, preprocessing and annotating the original documents obtained, that is, the academic achievement text data and the industrial text data; improving the accuracy of entity recognition and extraction through a large language model (LLM), obtaining industrial field entities and academic achievement entities, and constructing an industrial knowledge graph based on industrial field entities; then, using LLM to convert academic achievement attributes into vector representations, and using a multi-label random forest algorithm to divide academic achievements into industrial fields, thereby improving the accuracy of classification; finally, dividing industrial labels according to the industrial field classification results of academic achievements, and then calculating the correlation between industrial labels and industrial knowledge graphs, thereby realizing the fusion of knowledge graphs.

[0038] In summary, the entity extraction technology based on the large model in this embodiment 1 significantly improves the accuracy of knowledge map construction. First, the large model is used to pre-process and annotate academic achievement text data and industrial text data, and the accuracy of entity recognition is improved by fine-tuning the large model, and the academic achievement knowledge map and industrial knowledge map are constructed in combination with the prompt word project. Subsequently, the semantic vector model of the large model is used to convert the academic achievement attributes into vector representations, and the multi-label random forest algorithm is used to divide the academic achievements into industrial fields, thereby improving the accuracy of classification. Finally, the industrial labels are divided according to the industrial field classification results of the academic achievements, and the correlation between the industrial labels and the industrial knowledge map is calculated, realizing the fusion of the knowledge map. It provides a clear industrial field division for academic achievements and scientific research talents, which is conducive to the subsequent application of talent selection, industrial development planning, etc., and greatly enriches the information content of the knowledge map.

[0039] Example 2 See also Figure 5 , is a structural diagram of an academic industry knowledge graph fusion system based on a large model in Example 2 of the present application; the specific contents include: Graph construction module: obtain academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data; Feature extraction module: performs text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and uses the semantic vector model of the large model to convert the processed academic achievement attribute information to obtain the academic achievement attributes represented by vectors; Training module: adding labels to the academic achievement attributes represented by the vector as a training set, constructing and training a multi-label random forest model based on the training set, classifying the academic achievements in the academic achievement knowledge graph, and obtaining the industrial field classification results of the academic achievements; Division module: Based on the industry classification results of the academic achievements, scientific research talents are divided into industry fields and industry labels are generated; Graph fusion module: calculates the correlation between the industry label and the industry knowledge graph, and completes the fusion of the academic achievement knowledge graph and the industry knowledge graph based on the correlation calculation result.

[0040] Example 3 See also Figure 6 , is a schematic diagram of the device structure of Embodiment 3 of the present application. The device 50 includes a processor 51 and a memory 52 coupled to the processor 51 .

[0041] The memory 52 stores program instructions for implementing the above-mentioned academic-industry knowledge graph fusion method based on a large model.

[0042] The processor 51 is used to execute program instructions stored in the memory 52 to realize an academic-industry knowledge graph fusion based on a large model.

[0043] The processor 51 may also be referred to as a CPU (Central Processing Unit).

[0044] The processor 51 may be an integrated circuit chip with signal processing capability. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0045] Example 4 See also Figure 7 , which is a schematic diagram of the structure of the storage medium of Example 4 of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods, wherein the program file 61 can be stored in the above storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, or a computer, server, mobile phone, tablet and other devices.

[0046] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0047] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0048] Although the embodiments of the present application have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the appended claims and their equivalents.

[0049] Of course, the present invention may have many other implementations. Based on this implementation, other implementations obtained by ordinary technicians in this field without any creative work are all within the scope of protection of the present invention.

Claims

1. A method for integrating academic and industrial knowledge graphs based on a large model, characterized in that: include: Acquire academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data; Performing text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and converting the processed academic achievement attribute information using a semantic vector model of a large model to obtain academic achievement attributes represented by vectors; Add labels to the academic achievement attributes represented by the vector as a training set, build and train a multi-label random forest model based on the training set, classify the academic achievements in the academic achievement knowledge graph, and obtain the industrial field classification results of the academic achievements; Based on the industry classification results of the academic achievements, scientific research talents are divided into industry fields and industry labels are generated; Calculate the correlation between the industry label and the industry knowledge graph, and complete the fusion of the academic achievement knowledge graph and the industry knowledge graph based on the correlation result; specifically include the following steps: By matching the labels of the industry knowledge graph with the industry labels, the correlation between the industry labels and the labels of the industry knowledge graph is calculated according to the positions and quantities of the labels in the matched industry knowledge graph.

2. According to the large model-based academic and industrial knowledge graph fusion method of claim 1, it is characterized by: The steps of obtaining academic achievement text data and industry text data, constructing an academic achievement knowledge graph based on the academic achievement text data, and constructing an industry knowledge graph based on the industry text data specifically include the following steps: Preprocessing and annotating the academic achievement text data, identifying the preprocessed academic achievement text data by fine-tuning the large model, extracting the preprocessed academic achievement text data by using prompt word engineering, and combining to obtain academic achievement entities; constructing an academic achievement knowledge graph based on the academic achievement entities; The industrial text data is preprocessed and annotated, the preprocessed industrial text data is recognized by fine-tuning the large model, the preprocessed industrial text data is extracted using prompt word engineering, and the industrial field entities are obtained by combining them; an industrial knowledge graph is constructed based on the industrial field entities.

3. According to the large model-based academic and industrial knowledge graph fusion method of claim 1, it is characterized by: The step of performing text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, converting the processed academic achievement attribute information using a semantic vector model of a large model, and obtaining the academic achievement attributes represented by a vector specifically includes the following steps: Perform text cleaning on the academic achievement attribute information, and use regular expressions to remove preset irrelevant characters; use the NLTK library to perform stem extraction on the academic achievement attribute information to obtain word roots; Through the semantic vector model of the large model, the preprocessed academic achievement attribute information is converted into the corresponding vector representation to obtain the academic achievement attributes represented by the vector.

4. According to the large model-based academic and industrial knowledge graph fusion method of claim 1, it is characterized by: The step of adding labels to the academic achievement attributes represented by the vector as a training set, constructing and training a multi-label random forest model based on the training set, classifying the academic achievements in the academic achievement knowledge graph, and obtaining the industrial field classification results of the academic achievements specifically includes the following steps: A multi-label random forest model is built based on the training set. Several decision trees are constructed during the training process, and a multi-label learning strategy is used for classification. After the training is completed, the trained multi-label random forest model is used to classify and predict the academic achievements in the academic achievement knowledge graph to obtain the industrial field classification results of the academic achievements.

5. According to the large model-based academic and industrial knowledge graph fusion method of claim 1, it is characterized by: The industrial knowledge graph divides the industrial field into four layers, and the academic achievement knowledge graph includes papers published by scientific research talents and patents invented.

6. A system for the large-model-based academic-industry knowledge graph fusion method according to claim 1, characterized in that: include: Graph construction module: obtain academic achievement text data and industry text data, construct an academic achievement knowledge graph based on the academic achievement text data, and construct an industry knowledge graph based on the industry text data; Feature extraction module: performs text cleaning on the academic achievement attribute information in the academic achievement knowledge graph, and uses the semantic vector model of the large model to convert the processed academic achievement attribute information to obtain the academic achievement attributes represented by vectors; Training module: adding labels to the academic achievement attributes represented by the vector as a training set, constructing and training a multi-label random forest model based on the training set, classifying the academic achievements in the academic achievement knowledge graph, and obtaining the industrial field classification results of the academic achievements; Division module: Based on the industry classification results of the academic achievements, scientific research talents are divided into industry fields and industry labels are generated; Graph fusion module: calculates the correlation between the industry label and the industry knowledge graph, and completes the fusion of the academic achievement knowledge graph and the industry knowledge graph based on the correlation result.

7. A device, characterized in that: The device includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing a method for fusion of academic and industrial knowledge graphs based on a large model as described in any one of claims 1-5; the processor is used to execute the program instructions stored in the memory to implement a method for fusion of academic and industrial knowledge graphs based on a large model.

8. A storage medium, characterized in that: The device stores program instructions executable by a processor, and the program instructions are used to execute a large-model-based academic-industry knowledge graph fusion method as described in any one of claims 1 to 5.

Images