Maritime business standard operation program intelligent construction method and system based on multi-source knowledge fusion

Abstract

The invention provides a maritime business standard operation program intelligent construction method and system based on multi-source knowledge fusion, and relates to the technical field of maritime business intelligent construction. Multi-source knowledge and multi-modal data of maritime business are collected firstly and then processed through a Transform technology to obtain a business knowledge graph and scene interaction data; a business process model is established based on the two, a business association network is constructed through a graph neural network to generate an operation program conforming to international and domestic dual standards, and then a standard operation program is formed after test optimization of an application test terminal integrating AR and a voice module, so that actual maritime affair requirements can be accurately met; and standardized and intelligent development of maritime affairs is facilitated.

Description

Technical Field

[0001] This application relates to the technical field of intelligent construction of maritime operations, and in particular to a method and system for intelligent construction of standard operating procedures for maritime operations based on multi-source knowledge fusion. Background Technology

[0002] In maritime management scenarios, business standardization is the core of ensuring safety and improving efficiency. It covers multiple scenarios such as ship operation and safety supervision. Related intelligent construction methods can meet the needs of industry digital transformation and have a very broad application prospect.

[0003] Currently, the construction of standard operating procedures for maritime operations largely relies on manual sorting of regulations, cases, and other materials. Some methods also use basic data integration tools to aggregate scattered business information and form fixed process texts.

[0004] However, these existing methods are difficult to fully integrate various types of business information, resulting in insufficient ability of the generated operating procedures to adapt to actual scenarios, and inflexible process adjustments when faced with complex law enforcement needs, making it impossible to respond efficiently to changes in on-site business. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for intelligent construction of maritime business standard operating procedures based on multi-source knowledge fusion, so as to solve the problems of insufficient adaptability and practical efficiency of existing maritime business standard operating procedures.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for intelligently constructing maritime business standard operating procedures based on multi-source knowledge fusion, comprising:

[0007] Collect multi-source knowledge and multi-modal data on maritime operations;

[0008] The Transformer technology is used to perform deep semantic understanding on the multi-source knowledge to obtain a business knowledge graph. The multimodal data is then semantically enhanced to obtain scene interaction data.

[0009] Based on the business knowledge graph and the scenario interaction data, a business process model is established, and based on the business process model, a business association network is constructed using graph neural network technology.

[0010] Based on the aforementioned business association network, an operating procedure that conforms to the preset dual standards for maritime business is generated.

[0011] The job program is deployed on the application test terminal for testing, and the job program is optimized based on the feedback data to obtain the standard job program.

[0012] Optionally, the step of using Transformer technology to perform deep semantic understanding on the multi-source knowledge to obtain a business knowledge graph, and performing semantic enhancement processing on the multimodal data to obtain scene interaction data, including:

[0013] The Transformer technology is used to perform multi-level semantic parsing on the multi-source knowledge to obtain the semantic parsing results.

[0014] Based on the semantic parsing results, a business knowledge graph is constructed with maritime business entities as nodes and semantic relationships as edges.

[0015] The multimodal data is converted into semantic representation data in a unified semantic space;

[0016] Based on the contextual information of the target scene, the semantic representation data is processed with context awareness to obtain scene interaction data that is dynamically adapted to the target scene.

[0017] Optionally, the step of using Transformer technology to perform multi-level semantic parsing on the multi-source knowledge to obtain semantic parsing results includes:

[0018] The feature extraction module in the Transformer architecture is used to extract features from the multi-source knowledge to obtain a semantic feature representation with contextual relevance.

[0019] Semantic association analysis is performed on the semantic feature representation using the multi-head attention mechanism in the Transformer architecture to obtain the semantic association network between maritime business entities.

[0020] A semantic parsing framework is constructed based on the semantic association network using the semantic parsing module in the Transformer architecture.

[0021] The semantic fusion module in the Transformer architecture performs semantic fusion processing on the semantic parsing framework to form a semantic parsing result.

[0022] Optionally, the step of constructing a semantic parsing framework based on the semantic association network includes:

[0023] Based on the semantic association network, the semantic relationship types between maritime business entities are extracted to obtain the entity relationship set;

[0024] Descriptive information of each maritime business entity is extracted from the multi-source knowledge to obtain a set of entity attributes;

[0025] The entity relationship set and the entity attribute set are associated and integrated to construct a business relationship graph;

[0026] The business relationship graph is structured and arranged to obtain a semantic parsing framework.

[0027] Optionally, the step of establishing a business process model based on the business knowledge graph and the scenario interaction data, and constructing a business association network based on the business process model using graph neural network technology, includes:

[0028] A set of business rules is obtained by performing multi-dimensional analysis based on the aforementioned business knowledge graph.

[0029] Based on the on-site environmental features and operation records in the scene interaction data, scene business rules that match the target scene are selected from the business rule set;

[0030] Based on the aforementioned scenario business rules, a business process model supporting augmented reality (AR) visual guidance and voice interaction control is constructed. Based on the business process model, the logical relationships between nodes are defined, and an initial logic graph is constructed based on the logical relationships.

[0031] The initial logic graph is processed by feature propagation using graph neural network technology to obtain node features, and the correlation between nodes is calculated based on the node features.

[0032] The initial logic graph is adjusted based on the correlation to obtain the target logic graph, which is a business correlation network.

[0033] Optionally, the step of generating an operating procedure that conforms to preset maritime business dual standards based on the business association network includes:

[0034] The critical business execution path is extracted from the business-related network to obtain the job framework;

[0035] The first program unit is generated based on international standards, and the second program unit is generated based on domestic standards.

[0036] The first program unit and the second program unit are combined and arranged according to the business logic sequence preset in the operation framework to obtain the program framework;

[0037] The job program is instantiated based on the program framework.

[0038] Optionally, the step of deploying the job program on an application testing terminal for testing, and optimizing the job program based on feedback data to obtain a standard job program includes:

[0039] The operating procedure is deployed to an application test terminal that integrates an AR display module and a voice interaction module.

[0040] The application testing terminal is used to conduct application testing in the target scenario to obtain feedback data including AR operation feedback and voice operation feedback.

[0041] An optimized program scheme is generated based on the feedback data;

[0042] Based on the program optimization scheme, the work program is adjusted to generate a standard work program.

[0043] Secondly, this application provides an intelligent construction system for maritime business standard operating procedures based on multi-source knowledge fusion, including:

[0044] The data acquisition module is used to collect multi-source knowledge and multi-modal data related to maritime operations.

[0045] The enhancement module is used to perform deep semantic understanding on the multi-source knowledge using Transformer technology to obtain a business knowledge graph, and to perform semantic enhancement processing on the multimodal data to obtain scene interaction data.

[0046] The construction module is used to establish a business process model based on the business knowledge graph and the scenario interaction data, and to construct a business association network based on the business process model using graph neural network technology.

[0047] The generation module is used to generate operating procedures that conform to preset maritime business dual standards based on the business association network.

[0048] The testing module is used to deploy the job program on the application testing terminal for testing, and optimize the job program based on feedback data to obtain a standard job program.

[0049] Thirdly, this application provides an electronic device, comprising:

[0050] Memory, used to store computer programs;

[0051] A processor is used to execute the computer program to implement the steps of the intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion as described in the first aspect above.

[0052] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion as described in the first aspect above.

[0053] The intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion provided in this application has the following beneficial effects: First, by collecting multi-source knowledge and multi-modal data of maritime business and integrating various related information, comprehensive and rich data support can be provided for the construction of standard operating procedures. Next, by using Transformer technology to process multi-source knowledge and obtain a business knowledge graph, the dispersed knowledge can be presented in a structured way. Simultaneously, semantic enhancement of the multi-modal data yields scenario interaction data, making the scenario-related data more aligned with actual usage needs. Then, a business process model is established based on the business knowledge graph and scenario interaction data, and a business association network is constructed using a graph neural network. This clearly outlines the maritime business logic and process relationships, thus laying a solid foundation for program generation. Subsequently, operating procedures conforming to dual standards are generated based on the business association network, ensuring that the procedures simultaneously meet relevant international and domestic requirements, thereby adapting to maritime industry standards. Finally, the operating procedures are deployed, tested, and optimized based on feedback, improving the operability and reliability of the procedures, ultimately forming standard operating procedures adapted to practical applications.

[0054] Furthermore, this application enables the formation of a systematic and clear association system of maritime business-related knowledge through multi-level semantic parsing and structured construction of multi-source knowledge. At the same time, through unified representation and context-aware processing of multi-modal data, scenario data can be dynamically adapted to the target usage scenario, thereby providing accurate and demand-oriented data support for subsequent business process modeling and association network construction, and further helping to improve the adaptability of subsequent standard operating procedures. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 A flowchart illustrating an intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion, provided for an embodiment of this application;

[0057] Figure 2 A schematic diagram illustrating a specific implementation of an intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion, provided in this application embodiment;

[0058] Figure 3 A schematic diagram of the structure of an intelligent construction system for maritime business standard operating procedures based on multi-source knowledge fusion, provided for an embodiment of this application;

[0059] Figure 4This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0060] In maritime administration, standardized operational procedures are key to ensuring navigation safety and improving law enforcement efficiency. However, existing methods rely heavily on manual sorting of regulations, cases, and other materials, and some methods only use basic data integration tools to summarize information. This makes it difficult to fully integrate various scattered business resources, resulting in operational procedures that are often not well adapted to actual law enforcement scenarios. Furthermore, these procedures are not flexible enough to adapt to complex business needs and cannot respond efficiently to changes on the ground, making it difficult to meet the digital and intelligent transformation needs of the maritime industry.

[0061] To address this, this application proposes an intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion. The core of this method lies in: firstly, comprehensively collecting various maritime-related knowledge and scenario data; secondly, using intelligent technology to organize scattered knowledge into a structured, interconnected system, while simultaneously enhancing the practicality of scenario data; thirdly, relying on this processed data to construct a clear business logic network, generating operating procedures that conform to both international and domestic standards; and finally, optimizing through actual testing to form the final solution. This method not only solves the problem of insufficient information fusion in existing technologies, making the procedures more aligned with actual scenario needs, but also improves the efficiency and standardization of procedure construction, successfully overcoming the shortcomings of existing methods in adaptability and practical efficiency.

[0062] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0063] The core of this application is to provide an intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0064] S101. Collect multi-source knowledge and multimodal data of maritime operations.

[0065] Among them, multi-source knowledge includes information such as international conventions, domestic regulations, historical law enforcement cases, and business operation specifications related to maritime affairs, which serve as the basis for building standard operating procedures; multi-modal data includes information in different forms such as images, voice, and operation records from maritime law enforcement sites, which are used to reflect the specific circumstances of actual business scenarios.

[0066] In this application example, multi-source knowledge is first obtained by sorting out the existing data and historical archives of the maritime administration department, and then multi-modal data is collected through the collection equipment at the law enforcement site to ensure that the data not only covers business rules, but also restores the actual operation scenario.

[0067] S102. Using Transformer technology, perform deep semantic understanding on the multi-source knowledge to obtain a business knowledge graph, and perform semantic enhancement processing on the multimodal data to obtain scene interaction data.

[0068] In one specific implementation, such as Figure 2 As shown, step S102 includes:

[0069] Step 1021: Use Transformer technology to perform multi-level semantic parsing on the multi-source knowledge to obtain the semantic parsing results.

[0070] The Transformer technology is implemented using the Transformer architecture.

[0071] In this embodiment of the application, step 1021 may specifically include the following steps:

[0072] Step a1: The multi-source knowledge is processed by the feature extraction module in the Transformer architecture to obtain a semantic feature representation with contextual relevance.

[0073] Semantic feature representation refers to feature data extracted from multi-source knowledge that contains contextual information, used to reflect the core meaning and relational logic of knowledge.

[0074] In step a1, the feature extraction module of the Transformer architecture performs hierarchical analysis of multi-source knowledge segment by segment and sentence by sentence, and captures the core information units within the text as well as the semantic relationships between different information units, so as to generate semantic feature representations with context-dependent characteristics.

[0075] For example, three types of multi-source knowledge were selected: international maritime conventions, domestic maritime regulations, and historical law enforcement cases. The total number of text paragraphs for these three types of knowledge was 100. Then, feature extraction was performed on these 100 text paragraphs, and one set of core semantic features was extracted from each text paragraph, resulting in a total of 100 sets of semantic feature representations.

[0076] Step a2: Perform semantic association analysis on the semantic feature representation through the multi-head attention mechanism in the Transformer architecture to obtain the semantic association network between maritime business entities.

[0077] Among them, the semantic association network refers to the network structure formed by sorting out the semantic relationships between maritime business entities with maritime business entities as the core, which can intuitively present the knowledge association pattern; it should be noted that the specific types of maritime business entities are not specifically limited in the embodiments of this application.

[0078] In step a2, the multi-head attention mechanism of the Transformer architecture is used to perform multi-dimensional correlation detection and sorting on the generated semantic feature representation, and to identify the core entities in the maritime business field and the various correlation forms between entities. Then, these correlation information are organized into a network structure according to logical relationships to form a semantic correlation network.

[0079] For example, by analyzing these features using a multi-head attention mechanism, maritime business entities such as ship crew law enforcement procedures and regulations were identified. A total of 8 entities were counted, and 32 relationships between entities were identified to form a semantic association network.

[0080] Step a3: Construct a semantic parsing framework based on the semantic association network using the semantic parsing module in the Transformer architecture.

[0081] The semantic parsing framework refers to a structured framework that integrates entity relationships and attribute information, including entity attribute features and semantic relationship types, which can standardize the logic and scope of semantic parsing. It should be noted that this application does not impose specific limitations on the specific structure of the semantic parsing framework; it can be configured according to actual circumstances.

[0082] Step a3 may specifically include the following steps:

[0083] Step b1: Based on the semantic association network, extract the semantic relationship types between maritime business entities to obtain the entity relationship set.

[0084] Among them, the entity relationship set refers to the set of data extracted from the semantic association network that reflects the various types of associations between maritime business entities.

[0085] In step b1, the semantic association network is fully traversed, and different types of association patterns between entities are extracted to form a set of entity relationships. For example, three types of entity relationships—subordinate, constraint, and execution—are extracted from the network to form a set of entity relationships, with a total of three types of relationships.

[0086] Step b2: Extract the descriptive information of each maritime business entity from the multi-source knowledge to obtain the entity attribute set.

[0087] Among them, the entity attribute set refers to the set of information extracted from multi-source knowledge that describes the characteristics of the maritime business entity itself.

[0088] In step b2, multi-source knowledge is targeted and filtered, and descriptive information corresponding to each entity is extracted to form an entity attribute set.

[0089] For example, three types of attribute information for each entity are extracted from multi-source knowledge: name, scope of application, and operational requirements. Each entity has 6 attribute records extracted, and for 8 entities, a total of 48 attribute records are obtained, forming an entity attribute set.

[0090] Step b3: Associate and integrate the entity relationship set with the entity attribute set to construct a business relationship graph.

[0091] Among them, the business relationship graph refers to a graph constructed by integrating entity relationships and attribute information. It can include entity relationships such as ship management, crew training and safety inspection. It is used to present maritime business knowledge in a structured way, and the business relationship graph is the core foundation for the construction of business knowledge graph.

[0092] In step b3, the unique identifier of the entity is used as the connecting link to systematically match and integrate the entity relationship set and the entity attribute set, and to construct a business relationship graph that includes three dimensions: entity, relationship and attribute.

[0093] Step b4: Arrange the business relationship graph in a structured manner to obtain a semantic parsing framework.

[0094] In step b4, the business relationship graph is logically hierarchically sorted and its structure optimized, and the subordinate and relational logic between different levels of information is clarified to form a well-structured semantic parsing framework. For example, the graph is structured and arranged according to the logic of entity attribute relationships to obtain the semantic parsing framework.

[0095] Step a4: Perform semantic fusion processing on the semantic parsing framework through the semantic fusion module in the Transformer architecture to form a semantic parsing result with structured representation.

[0096] In step a4, the semantic fusion module of the Transformer architecture deeply integrates and removes redundancy from the entity, relation, and attribute information in the semantic parsing framework, and extracts the core semantic logic that runs through multi-source knowledge, ultimately forming the semantic parsing result. For example, the framework information is fused to form the final semantic parsing result.

[0097] Step 1022: Based on the semantic parsing results, construct a business knowledge graph with maritime business entities as nodes and semantic relationships as edges.

[0098] Among them, business knowledge graph refers to a visual knowledge carrier that is structurally complete, logically clear, and adapted to business scenarios, formed by further filtering core entities and semantic relationships, supplementing attribute annotations, and optimizing logical hierarchy based on business relationship graph and combined with semantic parsing results.

[0099] In step 1022, the semantic parsing results are first systematically sorted out, and clear maritime business entities are extracted as the core nodes of the graph. At the same time, these nodes are classified and labeled. Then, semantic relationships that can reflect the connection logic between nodes are selected from the semantic parsing results. Subsequently, these relationships are used as edges connecting the corresponding nodes, and the relationship type of each edge is clearly defined.

[0100] Subsequently, entity nodes and semantic relationship edges are organized and connected in an orderly manner according to the basic structural pattern of nodes, edges, and nodes, thus forming a preliminary graph structure. Then, each node is labeled with core attributes such as name and type, and each edge is labeled with attribute information such as relationship type and association strength. Then, the information dimensions of the graph are improved through attribute labeling, and finally a business knowledge graph with a complete structure and clear logic is constructed.

[0101] For example, based on the semantic parsing results, two types of ship entities (bulk carriers and container ships), two types of law enforcement entities (safety inspection and qualification verification), and two types of legal entities (international convention clauses and domestic regulations clauses) were extracted from eight maritime business entities, totaling six nodes; then, from the 32 relationships between entities, semantic relationships such as safety inspection based on international convention clauses, qualification verification based on domestic regulations clauses, and bulk carriers needing to undergo safety inspections were selected, totaling eight edges.

[0102] Subsequently, the 6 nodes and 8 edges are organized according to the structure of nodes, edges, and nodes; then, the nodes are labeled with name attributes and the edges are labeled with relationship type attributes. Based on the above node, edge, and attribute information, a business knowledge graph is constructed, which will be used to build a business process model in the future.

[0103] Step 1023: Convert the multimodal data into semantic representation data in a unified semantic space.

[0104] Semantic representation data refers to data that has been converted from different types of multimodal data into a unified semantic format, in order to eliminate information gaps caused by modal differences.

[0105] In step 1023, firstly, a unified semantic description rule is formulated based on the scenario requirements and data characteristics of maritime operations, and the semantic conversion standards and format requirements for different types of multimodal data such as images, voice, and operation records are clarified; then, for image-type multimodal data, visual features are extracted and converted into text description information that conforms to the unified rule; for voice-type multimodal data, voice information is converted into text form through speech-to-text technology, and then standardized according to the unified rule; for operation record-type multimodal data, core operation information is selected and standardized according to the unified semantic rule.

[0106] Subsequently, all converted text data undergoes format validation to correct expressions that do not conform to a unified standard and to ensure that all data are within the same semantic space, thereby completely eliminating information gaps between different modal data and ultimately obtaining standardized semantic representation data.

[0107] For example, three types of multimodal data from the law enforcement scene are first selected: ship images, voice recordings, and operation logs. These include 20 ship images, 15 voice recordings, and 25 operation logs. A unified semantic rule is then established: ship images describe the appearance and equipment status; voice recordings are transcribed into question-and-answer text; and operation logs retain core operational actions. Based on this rule, visual features such as the ship's integrity and complete equipment are extracted from the 20 ship images and converted into 20 text descriptions. The 15 voice recordings are transcribed into question-and-answer text using speech-to-text technology, resulting in 15 text entries. Core information such as the initiation of inspections and recording results is selected from the 25 operation logs, resulting in 25 text entries. Finally, the 20 + 15 + 25 = 60 converted text entries are standardized to obtain 60 semantically represented data entries with a unified format.

[0108] Step 1024: Based on the context information of the target scene, perform context-aware processing on the semantic representation data to obtain scene interaction data that is dynamically adapted to the target scene.

[0109] Context-aware processing refers to the processing method of adapting and adjusting semantic representation data by combining contextual information of the target scenario. This method can make the data more suitable for specific use scenarios. Target scenarios include ship management, crew training and on-site environments for safety inspection and law enforcement. Scenario interaction data refers to data that has been dynamically adapted to the target scenario after context-aware processing, and is used to support the construction of scenario-based business processes.

[0110] In step 1024, contextual information of the target scenario is first collected. This contextual information includes key dimensions such as the scenario's business type, business objectives, and execution environment. Then, a contextual information set of the scenario is constructed, and the semantic representation data is correlated with the contextual information set to filter out core data that is highly relevant to the business needs of the target scenario and remove redundant data that is irrelevant to the scenario's needs. Next, the filtered core data is adapted and adjusted in a scenario-specific way to supplement auxiliary information such as business rules and operation priorities specific to the target scenario. This allows the data to accurately match the actual business processes and execution requirements of the target scenario, ultimately resulting in scenario interaction data that is dynamically adapted to the target scenario.

[0111] For example, the target scenario is set as port vessel safety inspection, and contextual information of this scenario is collected, including the business type as safety verification, the business objective as identifying potential equipment hazards, and the execution environment as a port area. Then, 60 semantic representation data are correlated with the contextual information to filter out core data related to vessel equipment safety. For example, the core data can be filtered using the formula F=E×R, where F is the number of core data, E is the number of semantic representation data, and R is the scenario correlation coefficient. If the coefficient is 0.5, substituting it into the formula, we get F=60×0.5=30, thus filtering out 30 core data.

[0112] In addition, two types of auxiliary information were added: special safety requirements for port areas and priority ranking of equipment inspections. Furthermore, 30 core data points were adapted and adjusted to obtain 30 sets of scenario interaction data that are dynamically adapted to port vessel safety inspection scenarios.

[0113] This application achieves multi-level, deep parsing of multi-source knowledge, transforming scattered knowledge into structured, logically clear semantic parsing results, thus providing a solid foundation for subsequent business knowledge graph construction. Next, by converting the semantic parsing results into a visualized, structured business knowledge graph, the logical connections between maritime business knowledge can be intuitively presented, facilitating subsequent business process model construction. By eliminating modal differences in multimodal data and transforming various scenario data into semantic representation data in a unified format, a unified data foundation can be provided for subsequent context-aware processing. Finally, by accurately adapting semantic representation data to target scenarios, the scenario fit and usability of the data can be improved, thus supporting the scenario-based adaptation of subsequent business process models.

[0114] S103. Based on the business knowledge graph and the scenario interaction data, establish a business process model, and based on the business process model, construct a business association network using graph neural network technology.

[0115] In one specific implementation, step S103 includes:

[0116] Step 1031: Perform multi-dimensional analysis based on the business knowledge graph to obtain a set of business rules.

[0117] The multi-dimensional aspects can include ship management, crew training, and safety inspections, as well as information from other dimensions. This application does not specifically limit these aspects. The business rule set refers to a set of rules extracted based on multi-dimensional analysis of the business knowledge graph, which includes maritime business execution requirements, operating procedures, constraints, etc., and is used to regulate the execution logic of business processes.

[0118] In step 1031, the entity types, relationship attributes, and associated logic in the business knowledge graph are first sorted out to clarify the corresponding core business dimensions. The core business dimensions include: business execution scope, operation triggering conditions, definition of responsible entities, process boundary restrictions, etc. Then, these dimensions are cross-analyzed through clustering, association mining, etc., and the core execution requirements, standardized operation steps, and rigid constraints of various maritime businesses are extracted. Subsequently, they are classified and organized according to the structural specifications of "execution conditions, operation steps, and constraint requirements" to form a logically coherent and directly implementable set of business rules.

[0119] Step 1032: Based on the on-site environmental features and operation records in the scene interaction data, select scene business rules that match the target scene from the business rule set.

[0120] Among them, scenario business rules refer to the rules selected from the business rule set that match the on-site environmental characteristics and operation records of the target scenario, and are used to adapt the business process construction of specific scenarios.

[0121] In step 1032, on-site environmental features and operation records are first extracted from the scene interaction data. The on-site environmental features include key information such as execution location attributes, equipment operating status, and external environmental conditions. The operation records include historical execution step sequences, high-frequency operation behaviors, and common abnormal situations. Then, a mapping relationship between features and rules is established, and each rule in the business rule set is quantitatively evaluated according to the scene adaptability score to filter out rules with scores higher than the adaptability threshold. Finally, conflicting rules are eliminated through scene rationality verification to form scene business rules that fit the actual needs of the target scene.

[0122] Step 1033: Based on the scenario business rules, construct a business process model that supports AR visualization guidance and voice interaction control. Based on the business process model, define the logical relationships between nodes and construct an initial logic diagram based on the logical relationships.

[0123] Among them, the business process model refers to a process framework built based on scenario business rules that supports AR visualization guidance and voice interaction control, used to clarify the step sequence and interaction method of business operations; the initial logic diagram refers to a graphical structure built based on the nodes and logical relationships between nodes in the business process model, used to intuitively present the logical connections of the business process.

[0124] In step 1033, the core business operation is first broken down into its key steps and sequence, with the scenario business rules as the core, to clarify the execution conditions, input and output information and exception handling mechanisms of each step. Then, the AR visualization guidance is designed according to the logic of "step decomposition, hierarchical information display and real-time status feedback" to ensure that the operation steps are presented intuitively. Then, the voice interaction control presets a library of commonly used commands and question-and-answer mapping relationships, and supports voice-triggered step switching, information query and other operations. Finally, these elements are integrated to build a business process model.

[0125] Next, key nodes in the model are extracted according to the classification criteria of "core operations, auxiliary operations, and decision nodes". Logical relationships such as the execution order, dependencies, and branch jump conditions between nodes are defined. Then, the nodes are connected in an orderly manner according to the logical relationships to supplement the node attribute descriptions and relationship descriptions. Finally, an initial logic diagram that can fully reflect the core logic of the business process is constructed.

[0126] Step 1034: Use graph neural network technology to perform feature propagation processing on the initial logic graph to obtain node features, and calculate the correlation degree between nodes based on the node features.

[0127] Among them, node features refer to the feature data that reflects the core attributes and related information of nodes after the initial logical graph is processed by feature propagation using graph neural network technology; correlation degree refers to the index calculated based on node features, which is used to measure the degree of correlation between different nodes in the initial logical graph.

[0128] In step 1034, training data is first prepared, and logical graph samples of historical maritime business processes, along with corresponding node feature annotations and labels of real relationships between nodes, are collected. Then, based on the training data, a gradient descent optimizer and cross-entropy loss function are used to train a graph neural network, and the model parameters are iteratively adjusted until the loss value converges stably.

[0129] Subsequently, the adjacency matrix of the initial logical graph and the initial attributes of the nodes are input into the trained graph neural network. Through the inter-layer propagation mechanism of the model, the adjacency matrix is ​​used to transmit the association information between nodes, and the node's own attributes and the features of neighboring nodes are fused to generate node features containing multi-dimensional information. Then, based on the node features, the cosine similarity algorithm is used to calculate the similarity between any two node feature vectors to obtain the correlation degree that quantitatively reflects the closeness of the association between nodes.

[0130] Step 1035: Adjust the initial logic graph based on the correlation degree to obtain the target logic graph, which is a business correlation network.

[0131] Among them, the target logic diagram refers to the graphical network that accurately reflects the inherent logical relationships of each link in maritime operations after adjusting the initial logic diagram based on the degree of correlation, and is used to support the generation of operating procedures; the business relationship network can reflect the inherent logical relationships of each link in maritime operations.

[0132] In step 1035, the correlation threshold is first dynamically determined based on the importance weight of the business scenario, and then the correlation between nodes is compared with the threshold to retain valid connections with correlation higher than the threshold and eliminate redundant connections with correlation lower than the threshold to simplify the network structure.

[0133] Next, the weight representation of the connection between nodes is adjusted according to the magnitude of the correlation. Nodes with extremely high correlation are merged with similar functions to simplify the process. Then, transition logic is added to nodes with insufficient correlation but necessary for business to ensure the continuity of the process. Subsequently, the structural integrity of the adjusted logic graph is verified, and the business logic closed-loop detection is used to ensure that there are no process breakpoints. Finally, the target logic graph with optimized structure and rigorous logic is obtained as the business correlation network.

[0134] This application provides a unified rule basis for subsequent scenario-based rule selection and business process modeling by forming a standardized rule set covering core maritime business. Then, by achieving precise adaptation between business rules and target scenarios, it lays the foundation for building a business process model that fits actual needs. Next, by constructing a business process model and initial logic diagram that combines standardization and scenario-based approaches, it provides a basic framework for building subsequent business association networks. Subsequently, by extracting precise node features and quantifying the correlation between nodes, it provides data support for optimizing and adjusting the initial logic diagram. Finally, by constructing a logically rigorous and precisely correlated business association network, it provides core support for generating operational procedures that conform to actual business logic.

[0135] S104. Based on the business association network, generate an operating procedure that conforms to the preset dual standards of maritime business.

[0136] In one specific implementation, step S104 includes:

[0137] Step 1041: Extract the key business execution path from the business-related network to obtain the job framework.

[0138] The operational framework refers to the key business execution paths extracted from the business-related network. The operational framework can be a framework structure that includes core business links, step sequence, and node associations, and is used to standardize the overall logical layout of the operational procedure.

[0139] In step 1041, a path search algorithm is first used to fully traverse all node connection paths in the business-related network. Then, the business weight of each path is calculated by considering three dimensions: node importance, correlation strength, and execution frequency. After that, a weight threshold is set and critical paths above the threshold are selected, while redundant branch paths that have no impact on the core process are simultaneously removed.

[0140] Subsequently, the selected critical paths are systematically sorted according to the natural execution sequence of maritime operations to clarify the connection logic of each core link and the functional positioning of each node. Then, they are organized and sorted according to the inherent structure of "starting node, core link, and ending node" to finally form an operational framework with clear logical hierarchy and clear link connection.

[0141] Step 1042: Generate the first program unit based on international standards and the second program unit based on domestic standards.

[0142] The first procedure unit refers to a procedure module that is based on international maritime standards and includes the operational steps, compliance conditions, and implementation specifications required by international standards; the second procedure unit refers to a procedure module that is based on domestic maritime standards and includes the operational steps, compliance conditions, and implementation specifications required by domestic standards.

[0143] In step 1042, the core requirements directly related to maritime business in international and domestic standards are first systematically reviewed, and key information such as operating procedures, compliance judgment conditions, liability definition, and record specifications are comprehensively extracted. Then, this scattered information is transformed into structured, executable operating instructions and rigid constraint rules.

[0144] Next, the operational instructions and constraint rules corresponding to the international standards are classified and integrated according to the specific types of maritime business to ensure that the content logic under each category is consistent, forming the first procedure unit that conforms to the international standards. At the same time, the relevant content of the domestic standards is processed with the same classification and integration logic to form the second procedure unit that conforms to the domestic standards. Both procedure units reserve adaptation interfaces that precisely correspond to each node of the operation framework to ensure smooth subsequent combination.

[0145] Step 1043: Combine and arrange the first program unit and the second program unit according to the business logic sequence preset in the job framework to obtain the program framework.

[0146] The program framework refers to the overall program structure that meets both international and domestic standards, formed by combining and arranging the first and second program units according to the business logic sequence of the work framework.

[0147] In step 1043, the business logic sequence of the job framework and the adaptation requirements of each node are first analyzed in depth to clarify the functional positioning and operation requirements of each core node. Then, the operation steps in the first program unit and the second program unit are matched precisely with the core nodes of the job framework one by one, and the matched steps are ordered in an orderly manner according to the principle of "basic operations first, then special operations, and general requirements first, then specific requirements".

[0148] Next, the repetitive operation steps in the two units are merged, and the conflicting operation steps are integrated according to the rule of "prioritizing dual compliance and adapting to special scenarios" to supplement the connection logic and transition explanation between each step, eliminate process breakpoints, and finally form a program framework that is structurally complete, logically coherent and meets the dual standard requirements at the same time.

[0149] Step 1044: Instantiate and generate the job program based on the program framework.

[0150] The operating procedures may include ship safety inspection procedures, dangerous goods supervision procedures, crew qualification verification procedures, etc. This application embodiment does not specifically limit the content of this level, and can be set according to the actual situation.

[0151] In step 1044, specific and actionable operational details are first filled in for each core node of the program framework, including key practical information such as the executing entity, operation duration, required tools, judgment criteria, and recording methods, to ensure that each step has clear execution guidelines. Then, an exception handling process is added to clarify the corresponding corrective measures and alternative solutions for possible deviations in each operation step, so as to ensure the fault tolerance of the process.

[0152] Next, the procedures are standardized and organized according to a unified format of "step number, operation content, compliance basis, and precautions" to make the presentation of the procedures clear and easy to quickly refer to and follow during actual execution, ultimately forming an operating procedure that meets both dual standards and has strong practical feasibility.

[0153] This application extracts the core execution path of maritime operations, providing a structured framework for subsequent combination of procedural units and generation of operational procedures. Then, by forming procedural modules adapted to both international and domestic standards, it provides standardized unit materials for subsequent combination and arrangement. Next, through the organic integration of procedural units conforming to both standards, it constructs an overall procedural structure that is both compliant and logical. Finally, it generates operational procedures that are both compliant with both standards and practical, providing direct guidance for the standardized execution of maritime operations.

[0154] S105. Deploy the job program on the application test terminal for testing, and optimize the job program based on the feedback data to obtain a standard job program.

[0155] In one specific implementation, step S105 includes:

[0156] Step 1051: Deploy the operating program to an application test terminal that integrates an AR display module and a voice interaction module.

[0157] The application testing terminal refers to a hardware device that integrates an AR display module and a voice interaction module. The application testing terminal includes core components such as a processor, sensors, and data transmission interfaces. It is used to carry out work programs and simulate actual business scenarios for testing. It can realize visual guidance of work steps and interactive response to voice commands.

[0158] In step 1051, the application test terminal is first configured with a systematic environment and the runtime dependency library and system patch adapted to the operating program are installed to complete the resolution calibration, display delay debugging and viewing angle adaptation of the AR display module. This ensures that the visual content is presented clearly and without screen stuttering. At the same time, the voice interaction module’s sound reception sensitivity, noise filtering threshold and command recognition trigger conditions are adjusted to improve the basic interaction accuracy.

[0159] Next, the structured code and resource files of the job program were completely imported through the terminal's high-speed data transmission interface, and a standardized interface adaptation script was written to achieve precise linkage between program instructions and the AR display and voice interaction modules. Then, the rendering effect of the AR display module on the job steps, the smoothness of page switching, and the stability of information overlay were verified, and the recognition response rate and execution accuracy of the voice interaction module for the preset instructions in the program were tested to confirm that the program can run continuously and stably on the terminal and that all core functions are fully activated, laying a solid foundation for subsequent scenario-based testing.

[0160] Step 1052: Conduct application testing in the target scenario using the application testing terminal to obtain feedback data including AR operation feedback and voice operation feedback.

[0161] Feedback data refers to the comprehensive data collected during application testing. This feedback data covers information such as operational effectiveness, functional defects, and user experience issues, and is used to comprehensively evaluate the practical adaptability and functional completeness of the operating procedures.

[0162] Furthermore, AR operation feedback refers to the data related to the AR display module recorded during the test. This feedback includes step visualization clarity score, information overlay position deviation value, operation guidance completion rate, display delay time, etc., which are used to reflect the adaptation effect of AR function; voice operation feedback refers to the data related to the voice interaction module recorded during the test. For example, voice operation feedback includes command recognition accuracy, response time, question-answer matching rationality score, recognition stability in noisy environments, etc., which are used to reflect the interaction effect of voice function.

[0163] In step 1052, a test environment consistent with actual maritime operations is first built in the target scenario, and typical test scenarios such as normal operation, abnormal handling, multi-task switching, and complex environmental interference are designed to cover different operation modes such as single-person operation and multi-person collaboration. Then, test personnel with rich experience in maritime operations are organized to follow the complete steps of the operation procedure and conduct full-process, multi-round practical operation through the application test terminal, while simultaneously starting the terminal's built-in data acquisition system.

[0164] Next, AR operation feedback is collected in real time through the terminal image sensor, display log module, and operation behavior recorder to accurately record data such as visual scores, superimposed deviations, and guidance completion status for each step. Voice operation feedback is collected through the voice module's recognition log, timing component, and environmental noise monitor to record detailed data such as command recognition results, response time, and question-and-answer matching effect. The collected raw data is deduplicated and noise-reduced, and invalid data caused by operational errors or temporary equipment failures is removed through data validity verification. The data is then organized and classified according to a standardized structure of "operation step, feedback type, core indicator, and problem description" to form complete, accurate, and analytically valuable feedback data, providing a reliable basis for subsequent optimization analysis.

[0165] Step 1053: Generate a program optimization scheme based on the feedback data.

[0166] Among them, the program optimization plan refers to a systematic plan document formed based on feedback data analysis. The program optimization plan may include a problem list, priority ranking, optimization direction, specific measures, implementation steps, etc., and is used to accurately guide the adjustment and improvement of the work procedure.

[0167] In step 1053, the feedback data is first automatically classified using the K-means clustering algorithm, initially categorized into four main types: AR operation issues, voice interaction issues, process logic issues, and compliance issues. Then, manual review corrects any classification biases to ensure accurate categorization. For each type of issue, core data indicators are extracted to establish a quantitative evaluation system, and a problem priority scoring model is constructed to calculate the optimization priority of each issue. For example, the scoring formula corresponding to the problem priority scoring model is: Where P is the problem priority score. As the influence range weighting coefficient, This is a severity weighting coefficient. The frequency weighting coefficients are set according to the degree of business impact and their sum is 1. a is the impact range coefficient, I is the severity coefficient, and b is the frequency coefficient. Their coefficient values ​​are divided into 1 to 5 points according to business standards.

[0168] Next, the core issues were sorted out according to their priority scores from high to low. For each issue, the root cause was located through causal analysis. Based on the needs of maritime business, the optimization direction was clarified and specific and implementable adjustment measures were formulated. For example, AR display issues were addressed by adjusting rendering parameters or optimizing content layout, voice issues by expanding the command library or optimizing the recognition model, and process issues by adjusting the order of steps or supplementing the connection logic. Finally, a complete process optimization plan was formed, which included issue details, priority, root cause analysis, optimization measures, responsible parties, and completion deadlines. The plan was then finalized after being reviewed and confirmed by business experts.

[0169] Step 1054: Adjust the job program based on the program optimization scheme to generate a standard job program.

[0170] Standard operating procedures (SOPs) refer to the final operating procedures that have been tested and optimized to meet the actual enforcement needs of ship management, crew training, and safety inspections. They can combine international and domestic standards for compliance, practical feasibility, and functional stability, and can directly guide the standardized execution of maritime operations.

[0171] In step 1054, the work process adjustment is initiated according to the priority order of the program optimization scheme. High-priority issues are addressed first. The AR display module's step visualization content resolution, overlay coordinate parameters, and rendering logic are optimized by modifying the program code. Then, voice interaction issues are resolved by expanding the voice command library, optimizing the recognition algorithm model, and adjusting noise filtering parameters. Subsequently, the process node logic is reconstructed to adjust the step order and supplement the connection explanation. Finally, compliance issues are resolved by comparing the compliance basis details with the original standard text. A version control mechanism is used throughout the process to record every modification.

[0172] Next, after the adjustments were completed, the optimized program was redeployed to the application testing terminal, and multiple rounds of verification testing were conducted according to the original testing scenario. New feedback data was collected and quantitatively compared with the data before optimization to confirm that all high-priority issues were resolved, medium and low-priority issues were effectively improved, and no new issues were added. Subsequently, a review group composed of maritime business experts, compliance specialists, and front-line operators was organized to conduct a compliance review of the program against the original texts of both international and domestic standards. The feasibility and convenience of the program were then evaluated based on practical feedback, and final adjustments were made to the minor issues found during the review. Finally, a standard operating procedure that combines compliance, operability, and stability was formed and archived.

[0173] This application achieves deep adaptation between the operating program and the testing terminal through the above steps, and establishes the hardware and software foundation for scenario-based testing, which can ensure the smooth progress of subsequent testing work. Then, by comprehensively capturing the functional shortcomings and experience problems of the operating program in the actual scenario, it generates high-quality feedback data that can support optimization decisions. Finally, by accurately locating the core problems of the operating program and clarifying the optimization path, it can provide a scientific and operational guidance plan for program adjustment.

[0174] Figure 3 This application provides a schematic diagram of a specific implementation of an intelligent construction system for maritime business standard operating procedures based on multi-source knowledge fusion, as shown in the following embodiment. Figure 3 The system may include:

[0175] The acquisition module 31 is used to acquire multi-source knowledge and multimodal data of maritime business.

[0176] The enhancement module 32 is used to perform deep semantic understanding on the multi-source knowledge using Transformer technology to obtain a business knowledge graph, and to perform semantic enhancement processing on the multimodal data to obtain scene interaction data.

[0177] The construction module 33 is used to establish a business process model based on the business knowledge graph and the scenario interaction data, and to construct a business association network based on the business process model using graph neural network technology.

[0178] The generation module 34 is used to generate an operating procedure that conforms to the preset dual standards of maritime business based on the business association network.

[0179] The testing module 35 is used to deploy the job program on the application testing terminal for testing, and optimize the job program based on feedback data to obtain a standard job program.

[0180] The intelligent construction system for maritime business standard operating procedures based on multi-source knowledge fusion in this application embodiment is used to implement the aforementioned intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion. Therefore, the specific implementation of the intelligent construction system for maritime business standard operating procedures based on multi-source knowledge fusion can be found in the embodiment section of the intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0181] like Figure 4 As shown, this application also provides an electronic device, including: a memory 41 for storing a computer program; and a processor 42 for executing the computer program to implement the steps of the above-described intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion.

[0182] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion.

[0183] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0184] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the intelligent construction method for maritime business standard operating procedures based on multi-source knowledge fusion.

[0185] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0186] The above provides a detailed description of the intelligent construction method and system for maritime business standard operating procedures based on multi-source knowledge fusion provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for intelligent construction of maritime business standard operation procedure based on multi-source knowledge fusion, characterized in that, The method comprises the following steps: Collecting multi-source knowledge and multi-modal data of maritime business; Using the Transformer technology to perform deep semantic understanding on the multi-source knowledge to obtain a business knowledge graph, and performing semantic enhancement processing on the multi-modal data to obtain scene interaction data; Based on the business knowledge graph and the scene interaction data, a business process model is established, and based on the business process model, a business association network is constructed using graph neural network technology; Based on the business association network, an operation procedure conforming to the preset maritime business double standard is generated; The operation procedure is deployed on an application test terminal for testing, and the operation procedure is optimized according to the feedback data to obtain a standard operation procedure.

2. The method of claim 1, wherein, The method comprises the following steps: Using the Transformer technology to perform multi-level semantic analysis on the multi-source knowledge to obtain a semantic analysis result; Based on the semantic analysis result, a business knowledge graph is constructed with maritime business entities as nodes and semantic relationships as edges; The multi-modal data is converted into semantic representation data in a unified semantic space; Based on the context information of the target scene, the semantic representation data is subjected to context-aware processing to obtain scene interaction data dynamically adapted to the target scene.

3. The method of claim 2, wherein, The method comprises the following steps: Through a feature extraction module in the Transformer architecture, the multi-source knowledge is subjected to feature extraction processing to obtain semantic feature representations with context association; Through a multi-head attention mechanism in the Transformer architecture, semantic association analysis is performed on the semantic feature representations to obtain a semantic association network between maritime business entities; Through a semantic analysis module in the Transformer architecture, a semantic analysis framework is constructed based on the semantic association network; Through a semantic fusion module in the Transformer architecture, semantic fusion processing is performed on the semantic analysis framework to form a semantic analysis result.

4. The method of claim 3, wherein, The method comprises the following steps: Based on the semantic association network, the semantic relationship types between maritime business entities are extracted to obtain an entity relationship set; From the multi-source knowledge, the description information of each maritime business entity is extracted to obtain an entity attribute set; The entity relationship set and the entity attribute set are associated and integrated to construct a business relationship graph; The business relationship graph is structured and arranged to obtain a semantic analysis framework.

5. The method of claim 1, wherein, The method comprises the following steps: Based on the business knowledge graph, multi-dimensional analysis is performed to obtain a business rule set; Screening a scene business rule matched with the target scene from the business rule set based on the live environment features and operation records in the scene interaction data; Based on the scene business rule, a business process model supporting AR visual guidance and voice interaction control is constructed, based on the business process model, a logical relationship between nodes is defined, and based on the logical relationship, an initial logical graph is constructed; Feature propagation processing is performed on the initial logical graph using graph neural network technology to obtain node features, and based on the node features, the correlation degree between nodes is calculated; Based on the correlation degree, the initial logical graph is adjusted to obtain a target logical graph, which is a business association network.

6. The method of claim 1, wherein, Based on the business association network, an operation procedure conforming to the preset maritime business double standard is generated, including: Performing key business execution path extraction processing on the business association network to obtain an operation framework; Generating a first program unit based on international standards and a second program unit based on domestic standards; Combining and arranging the first program unit and the second program unit according to the business logic sequence preset by the operation framework to obtain a program framework; Based on the program framework, an operation procedure is instantiated.

7. The method of claim 1, wherein, The operation procedure is deployed on an application test terminal for testing, and the operation procedure is optimized based on feedback data to obtain a standard operation procedure, including: Deploying the operation procedure to an application test terminal integrated with an AR display module and a voice interaction module; Performing application testing in a target scene through the application test terminal to obtain feedback data containing AR operation feedback and voice operation feedback; Generating a program optimization scheme based on the feedback data; Based on the program optimization scheme, the operation procedure is adjusted to generate a standard operation procedure.

8. A maritime business standard operating procedure intelligent construction system based on multi-source knowledge fusion, characterized in that, It includes: The acquisition module is used for collecting multi-source knowledge and multi-modal data of maritime business; The enhancement module is used for deep semantic understanding of the multi-source knowledge by using the Transformer technology to obtain a business knowledge graph, and semantic enhancement processing of the multi-modal data to obtain scene interaction data; The construction module is used for establishing a business process model based on the business knowledge graph and the scene interaction data, and constructing a business association network based on the business process model using graph neural network technology; The generation module is used for generating an operation procedure conforming to the preset maritime business double standard based on the business association network; The test module is used for deploying the operation procedure on an application test terminal for testing, and optimizing the operation procedure based on feedback data to obtain a standard operation procedure.

9. An electronic device, comprising: It includes: The memory is used for storing a computer program; The processor is used for executing the computer program to implement the steps of the maritime business standard operation procedure intelligent construction method based on multi-source knowledge fusion according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the maritime business standard operation procedure intelligent construction method based on multi-source knowledge fusion according to any one of claims 1 to 7.

Images