Industrial knowledge generation type decision-making method based on multi-granularity semantics and large model assistance

By employing a multi-granular semantic and large-model-assisted industrial knowledge-generating decision-making method, the problems of inaccurate knowledge matching and one-sided decision-making schemes in industrial decision-making are solved. This method enables efficient and automated multimodal data processing and decision-making, thereby improving the production efficiency and decision-making quality in industrial environments.

CN118966342AActive Publication Date: 2024-11-15CHONGQING UNIV

Patent Information

Application Number
CN202411100117.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-15
Estimated Expiration
2044-08-12

AI Technical Summary

Technical Problem

Existing industrial decision-making methods are not precise enough when dealing with complex business needs and knowledge matching, especially when dealing with a large number of proper nouns and domain-specific terms. They are unable to fully capture and match multimodal data, resulting in low information utilization efficiency, one-sided decision results, low degree of automation, and difficulty in coping with complex and dynamic production environments.

Method used

We adopt an industrial knowledge-generating decision-making method based on multi-granularity semantics and large models. We construct high-dimensional semantic features through the BERT large language model, use graph neural networks (GNN) and graph convolutional networks (GCN) to fuse multi-granularity semantic features, and combine SPARQL queries and fine-tuning of the large language model to achieve deep understanding and automated decision-making of multimodal data.

Benefits of technology

It improves the accuracy and automation of industrial decision-making, enhances the ability to process complex semantic relationships, realizes comprehensive analysis and real-time performance of global information, and improves the continuity and efficiency of the decision-making process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118966342A_ABST
    Figure CN118966342A_ABST
Patent Text Reader

Abstract

The invention discloses an industrial knowledge generation type decision-making method based on multi-granularity semantics and large model assistance. The method comprises the following steps of 1, encoding a business problem into a high-dimensional problem semantic feature; 2, processing the multi-modal industrial knowledge graph by utilizing entity links; based on TOP-K path expansion, obtaining different sub-graphs; 3, capturing different granularity features in the sub-graphs by adopting a graph neural network GNN to obtain industrial knowledge multi-granularity semantic features; 4, performing unified fusion on the industrial knowledge multi-granularity semantic features and the problem semantic features by using a GCN (Graph Convolutional Network); 5, constructing a triple sampling process based on SPARQL writing query, and extracting related triple data; 6, rewriting the structured triple data into a free format text, and constructing a problem-decision pair data set; and 7, performing fine adjustment on the large language model based on the problem-decision pair data set to obtain a knowledge text enhanced industrial knowledge generation type decision model.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial intelligent decision-making, and specifically is an industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance. Background Art

[0002] Industrial environments often involve complex operating processes and diverse equipment, resources, and personnel. Effective decision-making can manage these complexities, ensure coordinated operation of all links, improve production efficiency, and reduce errors. By optimizing production processes and resource allocation, decision-making not only improves production efficiency and product quality, but also helps companies maintain cost advantages in the fierce market competition. In addition, industrial environments often face uncertainties and risks, such as market changes, supply chain disruptions, equipment failures, etc. Scientific decision-making methods can identify and evaluate these risks in advance, and formulate corresponding response strategies to reduce the negative impact of uncertainty.

[0003] However, existing decision-making methods for the industrial field are often not precise enough when dealing with complex business needs and knowledge matching, especially when involving a large number of proper nouns and domain-specific terms. Traditional methods have limited understanding and processing capabilities of semantic relationships, making it difficult to fully capture and match business needs at different levels. Industrial decision-making usually requires the integration of information from multiple data sources, such as sensor data, text reports, image data, etc. Existing methods are difficult to effectively integrate these heterogeneous data, resulting in inefficient information utilization. Although the application of knowledge graphs has improved, there are still challenges in the deep fusion and semantic understanding of multimodal data. In addition, traditional decision-making methods often only focus on data or features in one aspect, lack comprehensive analysis of global information, resulting in one-sided decision results and difficulty in dealing with complex situations in actual business. For example, equipment fault diagnosis may only rely on historical fault data, but fail to consider the real-time status and environmental factors of equipment operation. At the same time, the decision-making process relies on manual experience and manual operations, has a low degree of automation, and is easily affected by human factors. Especially in complex and dynamic production environments, the real-time and accuracy of the decision-making process are difficult to guarantee, affecting the continuity and efficiency of production. Therefore, it is particularly important to improve the accuracy and automation of industrial decision-making methods, especially when dealing with complex and dynamic production environments.

[0004] The multi-granularity semantic information modeling method aims to model sentences at different granularities, such as entity level, sentence level, and topic level. After modeling, the multi-granularity semantic representations need to be merged to form the final semantic representation. These multi-granularity semantic information can be extracted through graph neural networks (GNNs). GNNs can model the dependencies between semantic vectors of different granularities and realize the modeling of multi-granularity semantic information, thereby solving the problem of semantic sparsity and low matching accuracy caused by short texts and the inclusion of domain-specific vocabulary in text matching tasks. This method can improve the translation quality and text matching accuracy, and enhance the model's ability to handle complex semantic relationships.

[0005] As an important technology in the field of natural language processing (NLP), knowledge graphs effectively improve the performance of tasks such as information retrieval and question-answering systems by representing structured knowledge as graph structures. In a knowledge graph, nodes represent entities or concepts, and edges represent the relationships between entities. The construction and use of knowledge graphs require solving key issues such as entity recognition, relationship extraction, and knowledge fusion. Large language models (LLMs), such as GPT and BERT, have demonstrated strong generalization capabilities in a variety of NLP tasks through the pre-training-fine-tuning paradigm. These models are also significantly used in building knowledge graphs. For example, entity recognition and relationship extraction through pre-trained models can automatically extract knowledge from massive amounts of text. Summary of the invention

[0006] In view of this, in order to solve the problems of inaccurate knowledge matching and one-sided decision-making schemes in industrial decision-making, the purpose of the present invention is to provide an industrial knowledge generation decision-making method based on multi-granularity semantics and large models. By assisting large language models and multi-level semantic modeling, multimodal data can be effectively integrated, which can improve the accuracy and degree of automation of decisions, and enhance the accuracy and full cycle of decision-making schemes.

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

[0008] An industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance includes the following steps:

[0009] Step 1: Based on the high-dimensional feature representation method of industrial business problems, a business problem semantic feature embedding model based on the BERT large language model is constructed to encode business problems into high-dimensional problem semantic features;

[0010] Step 2: Use entity linking to perform multi-layer feature extraction and global fusion processing on the multimodal industrial knowledge graph; based on TOP-K path expansion, extend a path from each subject entity and derive the corresponding tree, merge the trees of different subject entities to form a unified subgraph, and obtain different subgraphs;

[0011] Step 3: Use graph neural network (GNN) to capture different granularity features in subgraphs and obtain multi-granularity semantic features of industrial knowledge;

[0012] Step 4: Use the graph convolutional network (GCN) to unify the multi-granularity semantic features of industrial knowledge and the semantic features of questions, so that the multi-granularity semantic features of industrial knowledge and the semantic features of questions are integrated in the same feature space to form a semantic feature fusion space of full-cycle industrial knowledge.

[0013] Step 5: Construct a triple sampling process based on SPARQL query writing, define business problems and convert them into corresponding SPARQL queries, and extract relevant triple data from the multimodal industrial knowledge graph and semantic feature fusion space;

[0014] Step 6: Rewrite the structured triple data into free-format text to construct a problem-decision pair dataset based on industrial business;

[0015] Step 7: Fine-tune the large language model based on the question-decision pair dataset to obtain an industrial knowledge generation decision model enhanced by knowledge text.

[0016] Furthermore, in step 2, based on TOP-K path expansion, the method steps for extending a path from each subject entity are as follows:

[0017] 21) Encode question q into a vector representation, calculate the relationship distribution through the linear classification layer, and select the top K relationships with the highest probability as the one-hop relationship path;

[0018] 22) Connect question q and the relationship path of the previous step as input, encode them into vector representation and calculate the relationship distribution; sort by the scores of the predicted relationship paths to obtain different subgraphs.

[0019] Further, the step 21) further comprises the following steps:

[0020] 211) Given a problem q, use PLM to encode the problem q and obtain the vector representation q v :

[0021] q v =PLM(q)

[0022] Among them: PLM is the pre-trained language model;

[0023] 212) Let the vector represent q v Input linear classification layer to predict potential hop counts h1,h2,...,h H The probability distribution D′ h :

[0024]

[0025] Where: Linear is the linear transformation function; is a given problem representation q v The number of hops h c Probability:

[0026]

[0027] 213) Select the hop count h with the highest probability as the prediction result:

[0028]

[0029] Where: H is the total number of potential hops.

[0030] Furthermore, during training, the ground truth distribution D h Represented as a one-hot vector, where the true hop count h gold The probability of is 1, and the probability of other hops is 0:

[0031]

[0032] Use cross entropy loss L CE , predicted probability distribution D′ h The difference between the ground truth distribution D h Being punished for being different:

[0033]

[0034] in: is the probability of the actual hop count distribution of the model during training.

[0035] Further, the step 22) comprises the following steps:

[0036] 221) Let the vector represent q v Input the linear classification layer to calculate the probability distribution D′ of R relations in the knowledge graph r,1 :

[0037]

[0038] in: is a given problem representation q v The relationship c Probability:

[0039]

[0040] Select the top K relationships with the highest probability as the first-hop relationship path p1; R represents the total number of possible relationships in the knowledge graph;

[0041] 222) In the t-th hop relationship, the t-1-th hop relationship path p t―1 The relationship path p in t―1,i It can be expressed as:

[0042] p t―1,i =r i,1 |r i,2 |...|r i,t―1 ,i=1,2,...,K t―1

[0043] The question q and the relationship path p t―1,i Use "|" to connect as input sequence Q t :

[0044] Q t =q|r i,1 |r i,2 |...|r i,t―1

[0045] Q t Encoded by PLM as vector table Q t,v , calculate the relationship distribution D′ in KG through the linear classification layer r,t :

[0046] Q t,v =PLM(Q t )

[0047]

[0048] in: Is a given input sequence representation Q t,v Time Relationship c Probability:

[0049]

[0050] Keep the first K relations with the highest probability as the t-1th hop relation path p t―1,i The t-th hop relationship of

[0051] 223) After h-step prediction, we can get K h relationship path; relationship path p t,i The score of is the product of the probabilities of all relationships in the path:

[0052]

[0053] Among them: Score(p t,i ) represents the relationship path p t,i score;

[0054] Different subgraphs are obtained by sorting the predicted relationship paths according to their scores.

[0055] Furthermore, in step seven, the method steps for fine-tuning the large language model are as follows:

[0056] 71) Select a pre-trained model: First, select a pre-trained model suitable for the target task;

[0057] 72) Prepare fine-tuning data: Prepare fine-tuning datasets related to the target task;

[0058] 73) Designing prompt information: Designing additional prompt information to guide model learning during fine-tuning; the prompt information includes labels, features or constraints related to the task to enhance the model's understanding and learning of the task;

[0059] 74) Constructing a prompt learning task: combining the prompt information with the fine-tuning data to construct a prompt learning task;

[0060] 75) Fine-tune the model: Fine-tune the pre-trained model using fine-tuning data and prompt learning tasks; during the fine-tuning process, the prompt information is input together with the model input to guide model learning; the model parameters are updated through the back-propagation algorithm and optimizer to gradually adapt the model to the target task;

[0061] 76) Tuning and evaluation: During the fine-tuning process, adjust the fine-tuning strategy based on the performance feedback of the validation set; use the test set to evaluate the performance of the fine-tuned model, and further optimize and improve the model.

[0062] 77) Deployment and Application: Deploy the fine-tuned model to real applications and apply it to solve specific tasks or problems. Continuously monitor the performance of the model and adjust and optimize it according to actual conditions.

[0063] The beneficial effects of the present invention are:

[0064] The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance of the present invention has shown significant advantages in industrial decision-making and knowledge application. Specifically, it has the following technical effects:

[0065] (1) By adopting large language models such as BERT, it is possible to deeply understand and embed high-dimensional semantic features of business problems, thereby ensuring accurate understanding and matching of complex business needs. This invention not only improves the accuracy of text matching and multimodal industrial knowledge processing, but also enhances the adaptability of the model in processing complex semantic relationships and specific domain terms.

[0066] (2) Using graph neural networks (GNNs) to model multi-granularity semantic information can effectively capture the dependencies and global context between features of different granularities, improving the efficiency of building and applying industrial knowledge graphs. Through the GNN subgraph fusion method, a seamless transition from local detailed information to global semantic understanding is achieved, making knowledge integration in the industrial decision-making process more complete and accurate.

[0067] The application of the triple sampling process and rewriting module based on SPARQL queries enables structured knowledge graph data to be converted into free-format text, and then the large language model is trained and optimized to generate high-quality decision solutions. This not only improves the degree of automation in the decision-making process, but also enhances the adaptability and practicality of the model to actual business scenarios.

[0068] In summary, the present invention is based on a multi-granularity semantic and large-model assisted industrial knowledge generation decision-making method, which integrates advanced natural language processing technology and graph neural network algorithms, and significantly improves the full cycle of industrial knowledge processing and decision-making quality. It not only helps to fully understand and match business knowledge needs, but also can effectively deal with problems such as inaccurate knowledge matching and one-sided decision-making plans in the industrial field, thereby building a more intelligent and efficient industrial knowledge application system. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to make the purpose, technical solution and beneficial effects of the present invention clearer, the present invention provides the following drawings for illustration:

[0070] Figure 1 It is a flow chart of the industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance of the present invention;

[0071] Figure 2 This is the schematic diagram of the triple sampling process;

[0072] Figure 3 An example diagram for rewriting structured triple data into free-form text;

[0073] Figure 4 An example diagram of an industrial knowledge generation decision model;

[0074] Figure 5 Flowchart for fine-tuning large models. DETAILED DESCRIPTION

[0075] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0076] This embodiment aims to solve the problems of inaccurate knowledge matching and one-sided decision solutions in industrial decision making by constructing a Retrieve-Rewrite-Answer framework. The framework implements the feature interaction of different granularities of full-cycle business knowledge by constructing modules such as subgraph retrieval, question-decision pair, and knowledge text enhanced decision. Figure 1 As shown, this embodiment is based on a multi-granularity semantics and large model-assisted industrial knowledge generation decision-making method, and includes the following steps.

[0077] Step 1: Based on the high-dimensional feature representation method of industrial business problems, a business problem semantic feature embedding model based on the BERT large language model is constructed to encode business problems into high-dimensional problem semantic features. Specifically, in this embodiment, by studying the characteristics of multimodal data (such as text, images, sensor data, etc.), the BERT large language model is used to transform business problems q into high-dimensional semantic feature embedding to ensure accurate understanding of business needs.

[0078] Step 2: Study the accurate graph embedding method for multimodal industrial knowledge graphs to achieve accurate graph embedding. Specifically, use entity linking to perform multi-layer feature extraction and global fusion processing on the multimodal industrial knowledge graph; based on TOP-K path expansion, extend a path from each subject entity and derive the corresponding tree, merge the trees of different subject entities to form a unified subgraph, and obtain different subgraphs.

[0079] Specifically, in this embodiment, an entity linking method based on multi-layer feature extraction and global fusion is adopted to model jump prediction as a classification task based on a pre-trained language model. Based on TOP-K path expansion, a path is extended from each subject entity and the corresponding tree is derived, and then the trees of different subject entities are merged to form a unified subgraph. Specifically, the question q is encoded as a vector representation, and the relationship distribution is calculated through a linear classification layer, and the top K relationships with the highest probability are selected as one-hop relationship paths. In the subsequent steps, the question q and the relationship path of the previous step are connected as input, encoded as a vector representation, and then the relationship distribution is calculated. Parameter optimization is performed through cross entropy loss, and the scores of the predicted relationship paths are sorted to achieve the acquisition of different subgraphs, which are used as relevant knowledge to enhance the efficient decision-making and collective intelligence application of industrial knowledge by large language models.

[0080] Specifically, in this embodiment, based on TOP-K path expansion, the method steps for extending a path from each subject entity are as follows:

[0081] 21) The purpose of jump prediction is to predict the number of jumps of the problem, which is used to predict the next relationship path. Jump prediction is modeled as a classification task based on PLM. The problem q is encoded as a vector representation, the relationship distribution is calculated through the linear classification layer, and the top K relationships with the highest probability are selected as one-hop relationship paths. Specifically, it includes the following steps.

[0082] 211) Given a problem q, use PLM to encode the problem q and obtain the vector representation q v :

[0083] q v =PLM(q)

[0084] Among them: PLM is the pre-trained language model;

[0085] 212) Let the vector represent q v Input linear classification layer to predict potential hop counts h1,h2,...,h H The probability distribution D′ h :

[0086]

[0087] Where: Linear is the linear transformation function; is a given problem representation q v The number of hops h c Probability:

[0088]

[0089] 213) Select the hop count h with the highest probability as the prediction result:

[0090]

[0091] Where: H is the total number of potential hops; For the model to represent the given problem vector q v When the predicted hop count h c probability.

[0092] During training, the ground truth distribution D h Represented as a one-hot vector, where the true hop count h gold The probability of is 1, and the probability of other hops is 0:

[0093]

[0094] Use cross entropy loss L CE , predicted probability distribution D′ h The difference between the ground truth distribution D h Being punished for being different:

[0095]

[0096] in: is the probability of the actual hop count distribution of the model during training.

[0097] 22) Connect question q and the relationship path of the previous step as input, encode them into vector representation and calculate the relationship distribution; sort the predicted relationship paths according to their scores to achieve the acquisition of different subgraphs. That is, given question q and the predicted number of hops h, perform h-step predictions, each step corresponding to a hop relationship. In step t, the t-th hop relationship is predicted based on the predicted t-1 hop relationship path and question q through PLM as a classification task. That is, for each predicted relationship path, the candidate relationship for the next step is sampled. Specifically, the following steps are included.

[0098] 221) Let the vector represent q v Input the linear classification layer to calculate the probability distribution D′ of R relations in the knowledge graph r,1 :

[0099]

[0100] in: is a given problem representation q v The relationship c Probability:

[0101]

[0102] The top K relations with the highest probability are selected as the first-hop relation path p1; R is represented as the total number of possible relations in the knowledge graph, which are the target categories when the model makes classification predictions in multi-hop reasoning tasks.

[0103] 222) In the t-th hop relationship, the t-1-th hop relationship path p t―1 The relationship path p in t―1,i It can be expressed as:

[0104] p t―1,i =r i,1 |r i,2 |...|r i,t―1 ,i=1,2,...,K t―1

[0105] The question q and the relationship path p t―1,i Use "|" to connect as input sequence Q t :

[0106] Q t =q|r i,1 |r i,2 |...|ri,t―1

[0107] Q t Encoded by PLM as vector table Q t,v , calculate the relationship distribution D′ in KG through the linear classification layer r,t :

[0108] Q t,v =PLM(Q t )

[0109]

[0110] in: Is a given input sequence representation Q t,v Time Relationship c Probability:

[0111]

[0112] Keep the first K relations with the highest probability as the t-1th hop relation path p t―1,i The t-th hop relationship of .

[0113] 223) After h-step prediction, we can get K h relationship path; relationship path p t,i The score of is the product of the probabilities of all relationships in the path:

[0114]

[0115] Among them: Score(p t,i ) represents the relationship path p t,i score.

[0116] Different subgraphs are obtained by sorting the predicted relationship paths according to their scores.

[0117] Step 3: Use graph neural network (GNN) to capture features of different granularities in the subgraph, obtain multi-granularity semantic features of industrial knowledge, form a multi-granularity feature space, and realize the deep integration and semantic understanding of business needs and industrial knowledge graphs.

[0118] Specifically, in this embodiment, a subgraph fusion method based on a graph neural network GNN is adopted. Based on obtaining several subgraphs with different granularity features with semantic associations, the node and edge structure in each subgraph can capture local features at the corresponding granularity level while maintaining the association with the global graph. These subgraphs are subjected to feature extraction through the GNN model, and rich graph structure information is extracted at different granularities. Through the multi-head attention mechanism, fine-grained and coarse-grained features are fused together to generate a unified feature representation, forming a multi-granularity feature space, and realizing the retention of local detailed information and the capture of global context information.

[0119] Step 4: Study the unified modeling method of multi-granularity features of industrial knowledge and semantic features of problems, and use graph convolutional network (GCN) to unify the multi-granularity semantic features of industrial knowledge and semantic features of problems, so that the multi-granularity semantic features of industrial knowledge and semantic features of problems can be integrated in the same feature space to form a semantic feature fusion space for full-cycle industrial knowledge, supporting the entire process from knowledge generation, updating to application.

[0120] Step 5: Construct a triple sampling process based on SPARQL query writing, define business problems and convert them into corresponding SPARQL queries, and extract relevant triple data from the multimodal industrial knowledge graph and semantic feature fusion space, such as Figure 2 shown.

[0121] Specifically, in this embodiment, the definition of SPARQL query is: perform SPARQL query to obtain triples that meet the conditions. These triples are usually composed of a subject, a predicate, and an object, representing entities and their relationships in the knowledge graph. SPARQL is a standard language for querying RDF data. By writing a specific query, triples that meet business needs are selected from the knowledge graph.

[0122] Step 6: Rewrite the structured triple data into free-format text to build a dataset based on industrial business problem-decision pairs.

[0123] Specifically, in this embodiment, a rewriting module is used to convert structured triples into free-form text, such as Figure 3 As shown. Train the BERT large language model based on the question-related image-text pairs, given a graph and the corresponding free-form text; convert the triples in the graph into triple-form text by connecting the subject, predicate and object; convert the triple text into the graph-text conversion prompt p1 through the template T1; input the prompt p1 into the large model for fine-tuning, adopt a forced strategy in training, and use the cross-entropy loss function to update the parameters, and finally obtain the free-form text. Integrate it into a paragraph as knowledge related to the question to enhance the performance of the language model, making the industrial knowledge graph data more suitable for natural language processing.

[0124] In this embodiment, the template T1 is defined as: "Your task is to convert a knowledge graph into one or more sentences. The knowledge graph is: {text x in triple form}. The sentence is: ". Meanwhile, the prompt p1 and the free-form text y are used as the input and output of the rewriting module, respectively.

[0125] Step 7: Fine-tune the large language model based on the problem-decision pair dataset to obtain a knowledge text-enhanced industrial knowledge generative decision model, achieve efficient decision-making and collective intelligence application of multimodal industrial knowledge, and improve the application efficiency and decision-making quality of industrial knowledge. Import real data, compare and evaluate it with the model prediction results to determine whether it is correct.

[0126] like Figure 4 As shown, in this embodiment, enhanced reasoning assisted by knowledge text knowledge is performed through a large model. The generated knowledge y is integrated with the question q, and a template T2 is designed: "The following are facts that may be relevant to answering the question: {free-form text y} question: {question q} answer:". Template T2 is used to map the free-form text y and question q to KG to enhance the prompt p2, and the prompt p2 is input into the Meta LIama3 large model that has been fine-tuned to accurately generate a decision plan to solve the problem. Real data is imported, compared and evaluated with the model prediction results to determine whether it is correct. A standard is established to evaluate the accuracy of the answer, that is, if the answer contains at least one answer entity, the question is considered to be answered correctly.

[0127] In this embodiment, template T2 is defined as: "The following are facts that may be relevant to answering the question: {free-form text y} question: {question q} answer:". Template T2 is used to map the free-form text y and question q to the KG-enhanced prompt P2, and then the prompt P2 is input into the fine-tuned Meta LIama3 large model to accurately generate a decision plan to solve the problem.

[0128] Large model fine-tuning refers to the process of further adjusting model parameters on a specific task to improve performance based on a pre-trained model. This method has been widely used in fields such as natural language processing and computer vision, and has achieved remarkable results. Figure 5 As shown, in this embodiment, the method steps for fine-tuning the large language model are:

[0129] 71) Select a pre-trained model: First, select a pre-trained model suitable for the target task; common pre-trained models include BERT, GPT, RoBERTa, etc. Select a suitable pre-trained model according to the nature and requirements of the task.

[0130] 72) Prepare fine-tuning data: Prepare fine-tuning datasets related to the target task; including collecting, cleaning, and preprocessing data to ensure that the data quality and format meet the input requirements of the model.

[0131] 73) Design hint information: Design additional hint information to guide model learning during fine-tuning; the hint information includes task-related labels, features, or constraints to enhance the model’s understanding and learning of the task.

[0132] 74) Construct a prompt learning task: Combine the prompt information with the fine-tuning data to construct a prompt learning task; the prompt learning task can be a supervised learning task in which the prompt information serves as an auxiliary label, or a reinforcement learning task in which the prompt information serves as a reward signal.

[0133] 75) Fine-tune the model: Use fine-tuning data and prompt learning tasks to fine-tune the pre-trained model; during the fine-tuning process, the prompt information is input together with the model input to guide model learning; the model parameters are updated through the back-propagation algorithm and optimizer to gradually adapt the model to the target task.

[0134] 76) Tuning and evaluation: During the fine-tuning process, adjust the fine-tuning strategy based on the performance feedback of the validation set; use the test set to evaluate the performance of the fine-tuned model, and further optimize and improve the model.

[0135] 77) Deployment and Application: Deploy the fine-tuned model to real applications and apply it to solve specific tasks or problems. Continuously monitor the performance of the model and adjust and optimize it according to actual conditions.

[0136] Large model fine-tuning can utilize additional cue information to improve the performance of the model on specific tasks and is an effective model optimization strategy.

[0137] This embodiment uses a total of two different large language models for different applications. Specifically, the BERT large language model is used to convert the business question q into a high-dimensional semantic feature embedding process, and the BERT large language model is used to convert the industrial knowledge graph into text. The LIama3 large language model is used for text-enhanced reasoning answers.

[0138] Specifically, the BERT large language model performs well in question semantic embedding because its bidirectional encoder mechanism can capture contextual information within a sentence, thereby generating high-quality semantic representations. In the conversion of knowledge graphs to text, BERT can understand complex relationships and entities through pre-training and fine-tuning, making text generation more natural and accurate, significantly improving the effectiveness of automated knowledge expression and question-answering systems.

[0139] The Meta LIama3 large language model has significant advantages in text-enhanced reasoning answers. Its optimized architecture and large-scale training data enable it to have stronger contextual understanding and reasoning capabilities. The Meta LIama3 large language model can generate more accurate and coherent answers, handle complex questions and provide detailed explanations, thereby improving user experience, and is widely used in fields such as intelligent customer service, education and research.

[0140] The above-described embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or changes made by those skilled in the art based on the present invention are within the protection scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. An industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance, characterized by: The steps include: Step 1: Based on the high-dimensional feature representation method of industrial business problems, a business problem semantic feature embedding model based on the BERT large language model is constructed to encode business problems into high-dimensional problem semantic features; Step 2: Use entity linking to perform multi-layer feature extraction and global fusion processing on the multimodal industrial knowledge graph; Based on TOP-K path expansion, a path is extended from each subject entity and the corresponding tree is derived, and the trees of different subject entities are merged to form a unified subgraph, thus obtaining different subgraphs; Step 3: Use graph neural network (GNN) to capture different granularity features in subgraphs and obtain multi-granularity semantic features of industrial knowledge; Step 4: Use the graph convolutional network (GCN) to unify and fuse the multi-granularity semantic features of industrial knowledge and the semantic features of questions, so that the multi-granularity semantic features of industrial knowledge and the semantic features of questions are fused in the same feature space to form a semantic feature fusion space of full-cycle industrial knowledge. Step 5: Construct a triple sampling process based on SPARQL query writing, define business problems and convert them into corresponding SPARQL queries, and extract relevant triple data from the multimodal industrial knowledge graph and semantic feature fusion space; Step 6: Rewrite the structured triple data into free-format text to construct a problem-decision pair dataset based on industrial business; Step 7: Fine-tune the large language model based on the question-decision pair dataset to obtain an industrial knowledge generation decision model enhanced by knowledge text.

2. The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance according to claim 1 is characterized by: In the step 2, based on the TOP-K path expansion, the method steps for extending a path from each subject entity are as follows: 21) Encode question q into a vector representation, calculate the relationship distribution through the linear classification layer, and select the top K relationships with the highest probability as the one-hop relationship path; 22) Take the question q and the relation path of the previous step as input, encode them into vector representation and calculate the relation distribution; Different subgraphs are obtained by sorting the predicted relationship paths according to their scores.

3. The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance according to claim 2 is characterized by: The step 21) further comprises the following steps: 211) Given a problem q, use PLM to encode the problem q and obtain the vector representation q v : q v =PLM(q) Among them: PLM is the pre-trained language model; 212) Let the vector represent q v Input linear classification layer to predict potential hop counts h1,h2,...,h H The probability distribution D′ h : Where: Linear is the linear transformation function; is a given problem representation q v The number of hops h c Probability: 213) Select the hop count h with the highest probability as the prediction result: Where: H is the total number of potential hops.

4. The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance according to claim 3 is characterized by: During training, the ground truth distribution D h Represented as a one-hot vector, where the true hop count h gold The probability of is 1, and the probability of other hops is 0: Use cross entropy loss L CE , predicted probability distribution D′ h The difference between the ground truth distribution D h Being punished for being different: in: is the probability of the actual hop count distribution of the model during training.

5. The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance according to claim 2 is characterized by: The step 22) comprises the following steps: 221) Let the vector represent q v Input the linear classification layer to calculate the probability distribution D′ of R relations in the knowledge graph r,1 : in: is a given problem representation q v The relationship c Probability: Select the top K relationships with the highest probability as the first-hop relationship path p1; R represents the total number of possible relationships in the knowledge graph; 222) In the t-th hop relationship, the t-1-th hop relationship path p t―1 The relationship path p in t―1,i It can be expressed as: p t―1,i =r i,1 |r i,2 |...|r i,t―1 ,i=1,2,...,K t―1 The question q and the relationship path p t―1,i Use "|" to connect as input sequence Q t : Q t =q|r i,1 |r i,2 |...|r i,t―1 Q t Encoded by PLM as vector table Q t,v , calculate the relationship distribution D′ in KG through the linear classification layer r,t : Q t,v =PLM(Q t ) in: Is a given input sequence representation Q t,v Time Relationship c Probability: Keep the first K relations with the highest probability as the t-1th hop relation path p t―1,i The t-th hop relationship of 223) After h-step prediction, we can get K h relationship path; relationship path p t,i The score of is the product of the probabilities of all relationships in the path: Among them: Score(p t,i ) represents the relationship path p t,i score; Different subgraphs are obtained by sorting the predicted relationship paths according to their scores.

6. The industrial knowledge generation decision-making method based on multi-granularity semantics and large model assistance according to claim 2 is characterized by: In step 7, the method steps for fine-tuning the large language model are as follows: 71) Select a pre-trained model: First, select a pre-trained model suitable for the target task; 72) Prepare fine-tuning data: Prepare fine-tuning datasets related to the target task; 73) Designing prompt information: Designing additional prompt information to guide model learning during fine-tuning; the prompt information includes labels, features or constraints related to the task to enhance the model's understanding and learning of the task; 74) Constructing a prompt learning task: combining the prompt information with the fine-tuning data to construct a prompt learning task; 75) Fine-tune the model: Fine-tune the pre-trained model using fine-tuning data and prompt learning tasks; During fine-tuning, the prompt information is input together with the model input to guide model learning; the model parameters are updated through the back-propagation algorithm and optimizer to gradually adapt the model to the target task; 76) Tuning and evaluation: During the fine-tuning process, adjust the fine-tuning strategy based on the performance feedback of the validation set; Use the test set to evaluate the performance of the fine-tuned model and further optimize and improve the model. 77) Deployment and Application: Deploy the fine-tuned model to real applications and apply it to solve specific tasks or problems. Continuously monitor the performance of the model and adjust and optimize it according to actual conditions.

Citation Information

Patent Citations

  • Primary school Chinese personalized learning system based on knowledge graph and large model

    CN116860978A

  • Fine adjustment method, system and equipment based on large language model and medium

    CN117290480A

  • Large language model enhanced question and answer generation method

    CN118013051A

  • Ancient book knowledge base intelligent question and answer method, device and equipment based on large model technology

    CN118332091A

  • Method and apparatus for automatically generating inference questions and answers

    WO2021184311A1

Cited By

  • Joint modeling method based on knowledge graph and large model retrieval enhancement

    CN121117179A

  • Large language model prompt learning method and system for engineering emergency response

    CN121365128A

  • Intelligent station emergency decision-making method based on multi-modal large model and knowledge graph

    CN122334907A

  • Intelligent station emergency decision method based on multi-modal large model and knowledge graph

    CN122334907B