Integrated kitchen auxiliary design optimization method based on home knowledge graph

Abstract

The application discloses an integrated kitchen auxiliary design optimization method based on a household knowledge graph. The steps include: constructing an integrated kitchen auxiliary design model based on a household knowledge graph; inputting a preset interactive command into a kitchen design knowledge retrieval system, and the kitchen design knowledge retrieval system extracts a plurality of integrated kitchen knowledge points from the household knowledge graph in the integrated kitchen professional field according to the preset interactive command to perform auxiliary design and generate an integrated kitchen design scheme diagram, and then stores the integrated kitchen design scheme diagram in a data analysis and management system, so that the integrated kitchen auxiliary design optimization is realized. The application effectively solves the problems of poor kitchen household design scheme quality, serious homogeneity and the like, solves the problems of lack of shareable resources and low knowledge reuse efficiency in the kitchen professional field, supports deep knowledge association and mining of the integrated kitchen, generates an intelligent auxiliary design scheme, reduces the number of repeated design iteration modifications, and solves the problem of low household knowledge retrieval efficiency in the current integrated kitchen professional field.

Description

Technical Field

[0001] This invention relates to an integrated kitchen auxiliary design optimization method, specifically an integrated kitchen auxiliary design optimization method based on a home knowledge graph. Background Technology

[0002] Currently, integrated kitchen design capabilities have established significant advantages in design standards, service processes, professional software development and application, personnel training, and system construction. However, several pain points remain, which are also common problems in the industry. For example, due to varying levels of designer skill and a lack of design support tools, kitchen cabinet layouts are often unreasonable, poorly matched, and the design style does not match the environment; excellent design experience cannot be solidified. Clients lack sufficient understanding of professional design knowledge, resulting in actual solutions that fail to meet their needs and are highly homogenized. Furthermore, companies lack data management and analysis platforms, resulting in low levels of information management and often ad-hoc decisions regarding new product development. Consequently, the design and development of kitchen cabinets / appliances suffers from scattered knowledge distribution, inconsistent content quality, and an inability to meet the complex, diverse, and personalized needs of clients. Moreover, the entire design process requires iterative iterations due to the lack of intelligent design support tools, consuming substantial resources and time. To improve designers' design skills and enterprise information management, and to achieve more efficient knowledge management and services, a big data knowledge service framework based on knowledge fusion has been proposed, and a knowledge fusion model has been designed to improve the level of knowledge service management. A knowledge service model and user system have also been established, and a 3D printing knowledge ontology has been constructed to enable personalized knowledge services for customers, which have been applied to the appearance development of toy cars. Addressing the difficulty of providing personalized retrieval in the express delivery and logistics sector, a standardized description of the logistics service process has been developed, a relevant knowledge graph has been constructed, and semantic analysis has been used to obtain complex user needs, enabling semantic retrieval of express delivery services. However, it remains difficult to solve the problems faced in the design phase of the integrated kitchen professional field, such as scattered and redundant knowledge distribution, lack of auxiliary design tools for designers, and lack of data management and analysis support platforms for enterprises. Summary of the Invention

[0003] To address the problems existing in the background technology, the present invention provides an integrated kitchen auxiliary design optimization method based on home knowledge graph.

[0004] The technical solution adopted in this invention is:

[0005] The integrated kitchen auxiliary design optimization method of the present invention includes the following steps:

[0006] Step 1) Construct an integrated kitchen auxiliary design model based on a home furnishing knowledge graph. The integrated kitchen auxiliary design model includes a home furnishing knowledge graph in the field of integrated kitchens, a kitchen design knowledge retrieval system, and an enterprise data analysis and management system.

[0007] Step 2) Input the preset interactive command into the kitchen design knowledge retrieval system. Based on the preset interactive command, the system extracts several integrated kitchen knowledge points from the home furnishing knowledge graph in the integrated kitchen professional field to assist in the design and generate an integrated kitchen design scheme diagram, which is then stored in the data analysis and management system, thus optimizing the integrated kitchen design. In actual operation, when the user confirms and submits the integrated kitchen design scheme diagram generated in this retrieval, it is considered that the user needs to confirm the feasibility of the integrated kitchen design scheme diagram and order the various kitchen cabinets / appliances products in the integrated kitchen design scheme diagram.

[0008] In step 1), the construction of the home furnishing knowledge graph in the field of integrated kitchens is specifically as follows:

[0009] This paper uses an abstract inductive method to abstract and summarize kitchen cabinet / appliance related content and its related carriers into several entity types, thereby constructing a home furnishing knowledge graph ontology for the integrated kitchen professional field. A top-down knowledge construction method is used to map the concepts, instances, features, and systems in incremental kitchen cabinet / appliance knowledge to the home furnishing knowledge graph ontology for the integrated kitchen professional field. An R2RML-based knowledge mapping method is used to map the tables, records, columns, and foreign keys in existing kitchen cabinet / appliance knowledge to the home furnishing knowledge graph ontology for the integrated kitchen professional field, thus constructing the home furnishing knowledge graph for the integrated kitchen professional field. The home furnishing knowledge graph for the integrated kitchen professional field includes several integrated kitchen knowledge points. Each integrated kitchen knowledge point serves as an entity node in the integrated kitchen professional field home furnishing knowledge graph according to its abstracted and summarized type. When there is a relationship between two entity nodes, a connection exists between the two entity nodes, and each connection serves as an edge of the integrated kitchen professional field home furnishing knowledge graph.

[0010] The home furnishing knowledge graph ontology centers on the integrated kitchen design process, with content related to kitchen cabinets and appliances as its main components. The top-down knowledge construction method starts with the top-level design of the integrated kitchen, defining the classes, entities, attributes, relationships, and hierarchical structure of the home furnishing knowledge graph within the integrated kitchen professional domain based on the home furnishing knowledge graph ontology. A knowledge mapping method based on R2RML (RDB to RDF Mapping Language) is used to extract knowledge stored in the enterprise's relational database and convert it into triplet data, thereby constructing the home furnishing knowledge graph for the integrated kitchen professional domain.

[0011] The kitchen cabinet / appliance related content specifically includes the integrated kitchen design process steps, types and information of kitchen cabinet / appliance products, integrated kitchen design scheme diagrams generated before the current moment, product information of each kitchen cabinet / appliance product, user information, and integrated kitchen design experience texts. The product information of the kitchen cabinet / appliance products includes the style and size of the kitchen cabinet / appliance. The relevant carriers of the kitchen cabinet / appliance related content are specifically videos and documents related to kitchen cabinet / appliance.

[0012] The incremental kitchen cabinet / appliance knowledge specifically includes unstructured data such as style and color matching of kitchen cabinet / appliance products not included in the kitchen cabinet / appliance knowledge system, specifically design experience and product orders; the existing kitchen cabinet / appliance knowledge specifically includes structured data in the relational database of the kitchen cabinet / appliance knowledge system, as well as several pre-set product information and category information, that is, detailed information and category information of various products currently existing in the enterprise.

[0013] In step 1), the data analysis and management system includes a data analysis module and a data management module. The data analysis and management system manages the home furnishing knowledge graph in the integrated kitchen professional field through the data management module. Specifically, the data management module edits, downloads, or deletes several integrated kitchen knowledge points in the home furnishing knowledge graph in the integrated kitchen professional field, or uploads several integrated kitchen knowledge points that are not in the home furnishing knowledge graph in the integrated kitchen professional field.

[0014] The development of the data management module is divided into a user management module and an integrated kitchen auxiliary design knowledge base management and retrieval module. The user management module is further divided into super administrators (who can perform all operations), administrators (who manage user permissions and the knowledge base), and ordinary users (who have retrieval permissions). The knowledge base management function includes knowledge base upload, editing, download, and deletion operations. Metadata related to integrated kitchen design is stored in a MySQL database. Information is preprocessed according to a pre-configured model and stored in the graph database Neo4j in graph form. The number of relevant nodes and triples is calculated and stored in MySQL. Retrieval involves semantically encoding relevant entities in the integrated kitchen metadata and storing them in the search and data analysis engine ElasticSearch. Step 3) filters out the 50 entities with the highest similarity to the input string, returns the corresponding encoded vectors, and then sorts the information of the top 20 entities with the highest similarity values ​​before pushing it to the user.

[0015] The data analysis and management system is based on a B / S architecture and is provided to enterprise decision-makers in the form of a web browser. The data management module is based on the Spring Boot framework, the relational database MySQL, and the graph database Neo4j, and develops user management, knowledge base management, and retrieval functions.

[0016] In step 1), the data analysis and management system includes a data analysis module and a data management module. It uses a graph traversal algorithm to analyze and generate a hotspot home furnishing graph within the integrated kitchen professional field, and then visualizes it. Specifically, the data analysis module statistically analyzes the integrated kitchen professional field's home furnishing knowledge graph based on previously generated integrated kitchen design schemes and product information for various kitchen cabinets / appliances, according to the category, region, and style of the kitchen cabinets / appliances. Finally, it generates a hotspot home furnishing graph in a visual form for users to browse and analyze. The data analysis and management system performs statistical analysis and then visualizes the hotspot home furnishing graph.

[0017] The data analysis module uses integrated kitchen knowledge points, specifically historical enterprise data. After statistical analysis, it generates data such as sold furniture, type, style, sales area, and matching product information. This data is then filtered, analyzed, optimized, and presented as an improved kitchen cabinet / appliance design solution. The matching product information specifically refers to the products that are used in conjunction with sold furniture, such as screws and bolts used with sold kitchen cabinets.

[0018] The data analysis module utilizes historical enterprise data stored in the home furnishing knowledge graph, primarily including sales regions, product styles, complementary product information, and customer information from order product information. It employs a graph traversal algorithm to perform integrated kitchen product spectrum analysis, identifying regional and style hotspots for sold products, and then presents the data through graph visualization. Embedded within the retrieval system, data analysis is mainly located in the sidebar of the main interface, including regional statistics, style statistics, and customer profiles.

[0019] In step 2), the kitchen design knowledge retrieval system is equipped with a pre-trained model and a visualization module. The visualization module of the kitchen design knowledge retrieval system visualizes the home knowledge graph of the integrated kitchen professional field, that is, it displays the categories of kitchen cabinets / kitchen appliances.

[0020] In step 2), the kitchen design knowledge retrieval system includes a pre-trained model and a visualization module. Preset interactive commands are input into the pre-trained model for processing. The pre-trained model outputs an integrated kitchen design scheme diagram, which is then input into the visualization module for visualization, generating an excellent design scheme. This excellent design scheme includes integrated kitchen design knowledge, such as how to waterproof kitchen cabinets and the power supply models required for certain types of kitchen appliances, all visualized using a graph retrieval system. The kitchen design knowledge retrieval system allows users to interact with it through keywords or natural language queries within the integrated kitchen domain. It retrieves knowledge points related to the user's information needs from the home furnishing knowledge graph to assist in design. The preset interactive commands specifically refer to keywords or natural language within the integrated kitchen domain.

[0021] The pre-trained model is an unsupervised contrastive learning model built upon the fusion of knowledge from the integrated kitchen domain. This model encodes semantically similar sentences related to integrated kitchens, resulting in higher similarity in the vector space, while sentences with different semantics show lower similarity, thus improving the retrieval accuracy of integrated kitchen-related knowledge. The visualization interface of the graph is developed using the D3.js and Vue frameworks, displaying the search results as a graph in the user interface. The user interface is developed using JavaScript and employs a force-directed graph as its main structure. This feature not only supports the display of search results but also allows for dragging, locking, and unlocking of nodes, as well as customization of node colors, sizes, and text, and customization of line styles, widths, and text. In addition to demonstrating search capabilities, the visualization module also supports interactive operations, such as clicking on nodes or lines to view entity and relationship information.

[0022] The pre-trained model mentioned above is specifically the pre-trained BERT model, and the training process is as follows:

[0023] We constructed a corpus dataset for the integrated kitchen professional field, using literature, books, and web pages, and divided it into training, validation, and test sets. The training set was unlabeled because the method uses an unsupervised learning mechanism, so there was no need to label the training samples. The validation and test sets were labeled by forming sentence pairs from every two sentences in the validation and test sets. Each sentence pair was labeled with N+1 integers from 0 to N. The larger the label number, the more related the two sentences in the sentence pair are. The training, validation, and test sets were then sequentially input into the BERT pre-trained model to obtain the trained BERT pre-trained model.

[0024] A BERT (Bidirectional Encoder Representation from Transformers) pre-trained model is employed: the sampling layer samples from the integrated kitchen professional domain sample dataset and the general domain sample dataset at a specific ratio to form training samples for each batch. Each training sample passes through the encoding layer twice independently, generating an encoding vector corresponding to the sample and an encoding vector corresponding to the positive sample. The loss calculation layer calculates the loss for each sample, and the sum of the losses of all samples in the batch is the total loss for that round of training. After training, the model exhibits strong encoding capabilities on the integrated kitchen professional domain corpus; that is, sentences with similar semantics in the integrated kitchen professional domain show higher similarity in the vector space after model encoding, while sentences with different semantics show lower similarity in the vector space. The model can complete the task of integrated kitchen design knowledge retrieval with higher accuracy.

[0025] The beneficial effects of this invention are:

[0026] 1) This invention effectively solves the problems of poor quality and serious homogenization of kitchen home design schemes, and also solves the problems of scarce shareable resources and low efficiency of knowledge reuse in the professional field of kitchens.

[0027] 2) An integrated kitchen design knowledge retrieval system was built based on the constructed home furnishing knowledge graph, which effectively solved the problem of low efficiency in home furnishing knowledge retrieval in the current integrated kitchen professional field.

[0028] 3) Based on the integrated kitchen professional field home furnishing knowledge graph, the enterprise data analysis and management system was built, which improved the enterprise's information management level and provided decision support for guiding future new product development. Attached Figure Description

[0029] Figure 1 This is the auxiliary design model and overall process diagram of the present invention.

[0030] Figure 2 This is a diagram of classes and their hierarchical structure within the home furnishing knowledge ontology of the integrated kitchen professional field.

[0031] Figure 3 It is a relational, domain, and range graph defined in the home knowledge ontology of the integrated kitchen professional field.

[0032] Figure 4 This is a general framework diagram of home furnishing knowledge extraction methods in the field of integrated kitchens.

[0033] Figure 5 This is a schematic diagram of home furnishing knowledge in the field of integrated kitchens. Detailed Implementation

[0034] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] The integrated kitchen auxiliary design optimization method of the present invention includes the following steps:

[0036] Step 1) as Figure 1 As shown, an integrated kitchen auxiliary design model based on a home furnishing knowledge graph is constructed. The integrated kitchen auxiliary design model includes a home furnishing knowledge graph in the field of integrated kitchens, a kitchen design knowledge retrieval system, and an enterprise data analysis and management system.

[0037] In step 1), the construction of the home furnishing knowledge graph in the field of integrated kitchens is as follows:

[0038] Using abstract induction methods, kitchen cabinet / appliance-related content and its related carriers are abstracted and summarized into several entity types, thereby constructing an integrated home furnishing knowledge graph ontology in the professional field of kitchens; such as... Figure 4 As shown, a top-down knowledge construction method is used to map the concepts, instances, features, and systems in incremental kitchen cabinet / appliance knowledge to the home furnishing knowledge graph ontology of the integrated kitchen professional field. An R2RML-based knowledge mapping method is used to map the tables, records, columns, and foreign keys in existing kitchen cabinet / appliance knowledge to the home furnishing knowledge graph ontology of the integrated kitchen professional field, thus constructing the home furnishing knowledge graph of the integrated kitchen professional field. The home furnishing knowledge graph of the integrated kitchen professional field includes several integrated kitchen knowledge points. Each integrated kitchen knowledge point serves as an entity node of the home furnishing knowledge graph of the integrated kitchen professional field according to its abstracted and summarized type. When there is a relationship between two entity nodes, a connection exists between the two entity nodes, and each connection serves as an edge of the home furnishing knowledge graph of the integrated kitchen professional field.

[0039] The home furnishing knowledge graph ontology centers on the integrated kitchen design process, with content related to kitchen cabinets and appliances as its main components. The top-down knowledge construction method starts with the top-level design of the integrated kitchen, defining the classes, entities, attributes, relationships, and hierarchical structure of the home furnishing knowledge graph within the integrated kitchen professional domain based on the home furnishing knowledge graph ontology. A knowledge mapping method based on R2RML (RDB to RDF Mapping Language) is used to extract knowledge stored in the enterprise's relational database and convert it into triplet data, thereby constructing the home furnishing knowledge graph for the integrated kitchen professional domain.

[0040] The content related to kitchen cabinets / appliances specifically includes the integrated kitchen design process steps, types and information of kitchen cabinets / appliances, integrated kitchen design scheme diagrams generated up to the current moment, product information of each kitchen cabinet / appliance product, as well as user information and integrated kitchen design experience texts, etc. The product information of kitchen cabinets / appliances includes the style and size of the kitchen cabinets / appliances, etc.

[0041] The specific carriers of the kitchen cabinet / kitchen appliance related content are videos and documents related to kitchen cabinets / kitchen appliances.

[0042] Incremental kitchen cabinet / appliance knowledge specifically includes unstructured data such as style and color matching of kitchen cabinet / appliance products not included in the kitchen cabinet / appliance knowledge system, specifically design experience and product orders; existing kitchen cabinet / appliance knowledge specifically includes structured data in the relational database of the kitchen cabinet / appliance knowledge system, as well as several pre-set product and category information, that is, detailed information and category information of various products currently existing in the enterprise.

[0043] In step 1), the data analysis and management system includes a data analysis module and a data management module. The data analysis and management system manages the home furnishing knowledge graph in the integrated kitchen professional field through the data management module. Specifically, the data management module edits, downloads, or deletes several integrated kitchen knowledge points in the home furnishing knowledge graph in the integrated kitchen professional field, or uploads several integrated kitchen knowledge points that are not in the home furnishing knowledge graph in the integrated kitchen professional field.

[0044] The development of the data management module is divided into a user management module and an integrated kitchen auxiliary design knowledge base management and retrieval module. The user management module is further divided into super administrators (who can perform all operations), administrators (who manage user permissions and the knowledge base), and ordinary users (who have retrieval permissions). The knowledge base management function includes knowledge base upload, editing, download, and deletion operations. Metadata related to integrated kitchen design is stored in a MySQL database. Information is preprocessed according to a pre-configured model and stored in the graph database Neo4j in graph form. The number of relevant nodes and triples is calculated and stored in MySQL. Retrieval involves semantically encoding relevant entities in the integrated kitchen metadata and storing them in the search and data analysis engine ElasticSearch. Step 3) filters out the 50 entities with the highest similarity to the input string, returns the corresponding encoded vectors, and then sorts the information of the top 20 entities with the highest similarity values ​​before pushing it to the user.

[0045] The data analysis and management system is based on a B / S architecture and is provided to enterprise decision-makers in the form of a web browser. The data management module is based on the Spring Boot framework, the relational database MySQL, and the graph database Neo4j, and develops user management, knowledge base management, and retrieval functions.

[0046] In step 1), the data analysis and management system includes a data analysis module and a data management module. It uses a graph traversal algorithm to analyze and generate a hotspot home furnishing graph within the integrated kitchen professional field, and then visualizes it. Specifically, the data analysis module statistically analyzes the integrated kitchen professional field's home furnishing knowledge graph based on previously generated integrated kitchen design schemes and product information for various kitchen cabinets / appliances, according to the category, region, and style of the kitchen cabinets / appliances. Finally, it generates a hotspot home furnishing graph in a visual form for users to browse and analyze. The data analysis and management system performs statistical analysis and then visualizes the hotspot home furnishing graph.

[0047] The data analysis module uses integrated kitchen knowledge points, specifically historical enterprise data. After statistical analysis, it generates data such as sold furniture, type, style, sales area, and matching product information. This data is then filtered, analyzed, optimized, and presented as an improved kitchen cabinet / appliance design solution. The matching product information specifically refers to the products that are used in conjunction with sold furniture, such as screws and bolts used with sold kitchen cabinets.

[0048] The data analysis module utilizes historical enterprise data stored in the home furnishing knowledge graph, primarily including sales regions, product styles, complementary product information, and customer information from order product information. It employs a graph traversal algorithm to perform integrated kitchen product spectrum analysis, identifying regional and style hotspots for sold products, and then presents the data through graph visualization. Embedded within the retrieval system, data analysis is mainly located in the sidebar of the main interface, including regional statistics, style statistics, and customer profiles.

[0049] Step 2) Input the preset interactive command into the kitchen design knowledge retrieval system. Based on the preset interactive command, the system extracts several integrated kitchen knowledge points from the home furnishing knowledge graph in the integrated kitchen professional field to assist in the design and generate an integrated kitchen design scheme diagram, which is then stored in the data analysis and management system, thus optimizing the integrated kitchen design. In actual operation, when the user confirms and submits the integrated kitchen design scheme diagram generated in this retrieval, it is considered that the user needs to confirm the feasibility of the integrated kitchen design scheme diagram and order the various kitchen cabinets / appliances products in the integrated kitchen design scheme diagram.

[0050] In step 2), the kitchen design knowledge retrieval system is equipped with a pre-trained model and a visualization module. The visualization module of the kitchen design knowledge retrieval system visualizes the home knowledge graph of the integrated kitchen professional field, that is, it displays the categories of kitchen cabinets / kitchen appliances.

[0051] In step 2), the kitchen design knowledge retrieval system includes a pre-trained model and a visualization module. Preset interactive commands are input into the pre-trained model for processing. The pre-trained model outputs an integrated kitchen design scheme diagram, which is then input into the visualization module for visualization, generating an excellent design scheme. This excellent design scheme includes integrated kitchen design knowledge, such as how to waterproof kitchen cabinets and the power supply models required for certain types of kitchen appliances, all visualized using a knowledge graph. The kitchen design knowledge retrieval system allows users to interact with it through keywords or natural language queries within the integrated kitchen domain. It retrieves knowledge points related to the user's information needs from the home furnishing knowledge graph to assist in design. The preset interactive commands are specifically keywords or natural language queries within the integrated kitchen domain.

[0052] The pre-trained model is an unsupervised contrastive learning model built upon the fusion of knowledge from the integrated kitchen domain. This model encodes semantically similar sentences related to integrated kitchens, resulting in higher similarity in the vector space, while sentences with different semantics show lower similarity, thus improving the retrieval accuracy of integrated kitchen-related knowledge. The visualization interface of the graph is developed using the D3.js and Vue frameworks, displaying the search results as a graph in the user interface. The user interface is developed using JavaScript and employs a force-directed graph as its main structure. This feature not only supports the display of search results but also allows for dragging, locking, and unlocking of nodes, as well as customization of node colors, sizes, and text, and customization of line styles, widths, and text. In addition to demonstrating search capabilities, the visualization module also supports interactive operations, such as clicking on nodes or lines to view entity and relationship information.

[0053] The pre-trained model specifically refers to the pre-trained BERT model, and the training process is as follows:

[0054] We constructed a corpus dataset for the integrated kitchen professional field, using literature, books, and web pages, and divided it into training, validation, and test sets. The training set was unlabeled because the method uses an unsupervised learning mechanism, so there was no need to label the training samples. The validation and test sets were labeled by forming sentence pairs from every two sentences in the validation and test sets. Each sentence pair was labeled with N+1 integers from 0 to N. The larger the label number, the more related the two sentences in the sentence pair are. The training, validation, and test sets were then sequentially input into the BERT pre-trained model to obtain the trained BERT pre-trained model.

[0055] In practice, both the validation and test sets each contain 120 samples, labeled with six integers from 0 to 5. A BERT (Bidirectional Encoder Representation from Transformers) pre-trained model is used: the sampling layer samples from the integrated kitchen professional domain dataset and the general domain dataset at a specific ratio to form batches of training samples. Each training sample passes through the encoding layer twice independently, generating an encoding vector for the sample and an encoding vector for the positive example. The loss calculation layer calculates the loss for that sample, and the sum of the losses for all samples in that batch is the total loss for that round of training. After training, the model exhibits strong encoding capabilities on the integrated kitchen professional domain corpus; that is, sentences with similar semantics in the integrated kitchen professional domain show higher similarity in the vector space after model encoding, while sentences with different semantics show lower similarity in the vector space. The model can complete the task of integrated kitchen design knowledge retrieval with higher accuracy.

[0056] The classes and their hierarchical structure in the constructed integrated kitchen professional field home knowledge ontology are as follows: Figure 2 As shown, the class "Home Furnishing Knowledge in the Integrated Kitchen Professional Field" is the common ancestor of all classes, with four subclasses: "Kitchen Products," "Customer Orders," "Showroom and Exhibition Design," and "Design Experience and Engineering." The "Kitchen Products" class represents kitchen cabinets / appliances within Fotile's integrated kitchen sector and forms the core of the home furnishing knowledge graph ontology for the integrated kitchen professional field. This class has three subclasses: "Kitchen Cabinet Product Information," "Supporting Functional Component Information," and "Supporting Kitchen Appliance Product Information." The "Customer Orders" class represents order information generated from Fotile's historical sales data. This class has three subclasses: "Customer," "Main Product of the Order," and "Supporting Products of the Order." The "Showroom and Exhibition Design" class represents the carrier of Fotile's home furnishing knowledge in the integrated kitchen professional field and supports the ontology construction. This class has four subclasses: "Documents," "Videos," "Slides," and "Web Pages." The "Design Experience and Engineering Knowledge" class represents the excellent design and engineering experience of Fotile's designers and engineers. This class has two subclasses: "Documents" and "Images."

[0057] In the home furnishing knowledge graph constructed by this invention within the integrated kitchen professional field, specific knowledge primarily exists in the form of attribute values. The definition, category, examples, functions, and characteristics of a particular product are almost entirely stored as attribute values ​​within the knowledge graph. Therefore, the definition of attributes is crucial for the quality and completeness of home furnishing knowledge in the integrated kitchen professional field.

[0058] In the constructed integrated kitchen-specific home furnishing knowledge graph, relationships are primarily used to describe the hierarchical and object relationships between "kitchen products," the relevance between knowledge items, and the carriers of showrooms and exhibitions. The relationships defined and their descriptions in the constructed kitchen-specific home furnishing knowledge ontology are as follows: Figure 3 As shown, the domain-domain of the relationship "belonging to" is "kitchen product—kitchen product," and the relationship description is that a lower-level kitchen product belongs to a higher-level kitchen product. For example, the entity "New Chinese Style Kitchen Cabinet" includes the entity "Integrated Panel Kitchen Cabinet." The domain-domain of the relationship "matching (reversible)" is "kitchen product—kitchen product," and the relationship description is that different kitchen products have a certain correlation. For example, the entities "Chinese Style Kitchen Cabinet" and "Solid Wood Panels" are related entities. The domain-domain of the relationship "same type (reversible)" is "kitchen product—kitchen product," and the relationship description is that kitchen products of the same type, such as the entities "Integrated Panel Series" and "Integrated Panel Series." The "Solid Wood Series" is a related knowledge point; the related domain-value domain is customer orders-kitchen products, and the relationship description is that there is a certain correlation between the class "customer orders" and the class "kitchen products". For example, the entity "customer A" ordered the kitchen cabinet product entity "screen appliance cabinet"; the included domain-value domain is kitchen products-design experience and engineering, and the relationship description is that the design and construction process of kitchen products includes design experience and engineering; the carrier domain-value domain is kitchen products-showroom and exhibition design, and the relationship description is that kitchen products rely on showroom and exhibition design. For example, the carrier of "New Chinese Style Kitchen Cabinets" is "webpage".

[0059] For incremental knowledge, a top-down knowledge graph construction technique is used to generate it. Starting from the top-level design of the integrated kitchen, classes, attributes, relationships, and hierarchical structures are defined, and entities are further instantiated to generate them. For example, "Chinese Zen" is instantiated from the class "New Chinese Style". The ontology construction tool Protégé is used to model the incremental knowledge. The ontology layer is constructed with concrete things as the root node. The product category (composed of the core category tree of the product) and intangible things (business scenarios, product functions, product styles, product brands, target audiences, and styles) are the subclasses of things. The target audience of the category, the domain containing the material composition, and the domain containing the style are the categories. The range is the intangible thing. The name and alias are the data attributes of the entity node. The domain is the entity node, and the relationship is the type. The instance layer is the concretization of the ontology layer.

[0060] The integrated kitchen knowledge graph constructed in this invention is uploaded to the integrated kitchen design knowledge retrieval system, allowing for the visualization of integrated kitchen knowledge. The visualization clearly shows the relationships between integrated kitchen knowledge entities. Common errors and design problems are categorized as follows: kitchen appliances, hardware / cabinets, water / electricity / gas levels, size-related issues, carelessness-related issues, and other problems. Specifically, the error relationship between kitchen appliances and tower-type (European-style) range hoods, range hoods, and water heaters is related to installation location. Figure 5 As shown.

[0061] This invention constructs a home furnishing knowledge graph integrating the professional field of kitchen design; based on the constructed home furnishing knowledge graph, a kitchen design knowledge retrieval system is built to achieve rapid and in-depth association and retrieval of design knowledge related to kitchen cabinets and appliances within the professional field of kitchen design; based on the constructed home furnishing knowledge graph, an enterprise data analysis and management system is built, which processes and interprets the enterprise's historical order information through graph visualization technology to guide new product development and enterprise decision-making, and realize the digital management of enterprise data. This invention not only increases the knowledge base of home furnishing design experience, supports in-depth association and mining of integrated kitchen knowledge, and intelligently assists in the generation of design schemes, reducing the number of repeated design iterations and modifications, but also supports enterprises in managing, analyzing, and making decisions based on data.

Claims

1. An integrated kitchen-aided design optimization method based on home knowledge graph, characterized in that: The method includes the following steps: Step 1) Construct an integrated kitchen auxiliary design model based on a home furnishing knowledge graph. The integrated kitchen auxiliary design model includes a home furnishing knowledge graph in the field of integrated kitchens, a kitchen design knowledge retrieval system, and a data analysis and management system. Step 2) Input the preset interactive command into the kitchen design knowledge retrieval system. The kitchen design knowledge retrieval system extracts several integrated kitchen knowledge points from the home furnishing knowledge graph in the field of integrated kitchens according to the preset interactive command to assist in the design, generate integrated kitchen design scheme diagrams, and then store them in the data analysis and management system to achieve optimization of integrated kitchen auxiliary design. In step 1), the construction of the home furnishing knowledge graph in the field of integrated kitchens is specifically as follows: This paper uses an abstract inductive method to abstract and summarize kitchen cabinet / appliance related content and its related carriers into several entity types, thereby constructing a home furnishing knowledge graph ontology for the integrated kitchen professional field. A top-down knowledge construction method is used to map incremental kitchen cabinet / appliance knowledge onto the home furnishing knowledge graph ontology for the integrated kitchen professional field, and an R2RML-based knowledge mapping method is used to map existing kitchen cabinet / appliance knowledge onto the home furnishing knowledge graph ontology for the integrated kitchen professional field, thus constructing a home furnishing knowledge graph for the integrated kitchen professional field. The home furnishing knowledge graph for the integrated kitchen professional field includes several integrated kitchen knowledge points, each of which serves as an entity node. When there is a connection between two entity nodes, a connection line exists between them, and each connection line serves as an edge of the home furnishing knowledge graph for the integrated kitchen professional field. In step 1), the data analysis and management system includes a data analysis module and a data management module. The data analysis and management system manages the home furnishing knowledge graph in the integrated kitchen professional field through the data management module. Specifically, the data management module edits, downloads, or deletes several integrated kitchen knowledge points in the home furnishing knowledge graph in the integrated kitchen professional field, or uploads several integrated kitchen knowledge points that are not in the home furnishing knowledge graph in the integrated kitchen professional field. In step 2), the kitchen design knowledge retrieval system is equipped with a pre-trained model and a visualization module. The visualization module of the kitchen design knowledge retrieval system presents a visual representation of the home furnishing knowledge graph in the integrated kitchen professional field.

2. The integrated kitchen-aided design optimization method based on home knowledge graph as described in claim 1, characterized in that: The kitchen cabinet / appliance related content specifically includes the integrated kitchen design process steps, kitchen cabinet / appliance product types and product information, integrated kitchen design scheme diagrams generated before the current moment, and product information for each kitchen cabinet / appliance product, including the style and size of the kitchen cabinet / appliance; the related carriers of the kitchen cabinet / appliance related content are specifically videos and documents related to kitchen cabinet / appliance.

3. The integrated kitchen-aided design optimization method based on home knowledge graph as described in claim 1, characterized in that: The incremental kitchen cabinet / appliance knowledge specifically includes unstructured data on style and color matching of kitchen cabinet / appliance products that are not included in the kitchen cabinet / appliance knowledge system; the existing kitchen cabinet / appliance knowledge specifically includes structured data in the relational database of the kitchen cabinet / appliance knowledge system, as well as several preset product and category information.

4. The integrated kitchen auxiliary design optimization method based on home knowledge graph according to claim 1, characterized in that: In step 1), the data analysis and management system includes a data analysis module and a data management module. It uses a graph traversal algorithm to analyze and generate a hotspot home furnishing graph in the field of integrated kitchens and presents it in a visual form. Specifically, the data analysis module performs statistical analysis on the home furnishing knowledge graph in the field of integrated kitchens according to the category, region and style of the kitchen cabinets / appliances, based on the integrated kitchen design scheme diagrams generated before the current moment and the product information of each kitchen cabinet / appliance product. Finally, it generates a hotspot home furnishing graph in a visual form.

5. The integrated kitchen-aided design optimization method based on home knowledge graph according to claim 1, characterized in that: In step 2), the kitchen design knowledge retrieval system is equipped with a pre-trained model and a visualization module. The preset interactive commands are input into the pre-trained model of the kitchen design knowledge retrieval system for processing. The pre-trained model outputs an integrated kitchen design scheme diagram, which is then input into the visualization module for visualization presentation. The preset interactive commands are specifically keywords or natural language in the integrated kitchen field.

6. The integrated kitchen-aided design optimization method based on home knowledge graph according to claim 1, characterized in that: The pre-trained model mentioned above is specifically the pre-trained BERT model, and the training process is as follows: Literature, books, and web resources in the field of integrated kitchens were used to construct a corpus dataset for the integrated kitchen field, which was then divided into a training set and a validation set. The training set was unlabeled, while the validation set was labeled. Every two sentences in the validation set were combined into sentence pairs, and each sentence pair was labeled with N+1 integers from 0 to N. The larger the label number, the more related the two sentences in the sentence pair were. The training set and validation set were then sequentially input into the BERT pre-trained model to obtain the trained BERT pre-trained model.

Images