Study On Knowledge Graph For Fashion Design Resources

As an indispensable part of traditional industries in China,the clothing industry plays an essential supporting role in the growth of the national economy and social development.Although fashion design(FD)is a local part of the whole chain of the clothing industry,it impacts the entire process.Generally,FD is considered the core stage of highlighting the style of clothing products,improving the ability of design innovation,upgrading the enterprise brand,enhancing the market competitiveness,and meeting the diversified needs of consumers.The FD level,clothing aesthetic level,and clothing sales level are closely related to FD’s cognition level and knowledge reserve.With the development of the Internet,especially after WEB2.0,the active participation of users has dramatically enriched FD resources,which can provide fashion designers and related people with the most forwardlooking fashion information and design materials at the fastest speed.However,at present,the design resources on the Internet are relatively scattered,the FD knowledge is not systematic enough,and the FD knowledge from different sources is mixed,which requires a lot of time and energy to identify and select before they can be applied.Therefore,knowledge graph(KG)construction for FD resources is studied in this study,which aims to reorganize and utilize the FD resources scattered on the Internet through natural language processing technology and artificial intelligence.It is expected that KG can provide support for semantic retrieval,big data analysis,decision making,and visualization of FD.In this study,the current research status of KG in the field of FD is sorted out.Based on current studies,the definition,classification,connotation,and characteristics of FD resources are reconstructed,and the application value of KG of FD resources from users,enterprises,service providers,and society are analyzed.Additionally,the feasibility of KG construction from the perspectives of the policy environment,information resources,industry knowledge management,and artificial intelligence methods is analyzed to understand the main functions of KG of FD resources.Based on the analysis of characteristics,content,procedures and knowledge requirements of FD,the purpose and principles of why building the knowledge architecture are analyzed.By combining with the relevant principles of knowledge granularity,the framework of knowledge architecture of FD is studied.At the same time,the knowledge architecture is studied from five aspects,i.e.,the FD information resources,the FD theory resources,the fashion technology resources,the fashion material resources,and the fashion education resources,and the FD resources are comprehensively sorted.Based on traditional design theories,FD education resources are incorporated into the FD resource architecture to meet the needs of the FD industry for the knowledge and skills of designers.Under the architecture of FD resource knowledge,a domain hierarchy model is proposed to recognize and organize search interfaces from the deep web and extract web pages from open FD resources on the Internet.Additionally,a block importance model is proposed to partition page blocks and recognize important blocks to extract information highly related to the topic.Also,the conceptual structure and physical structure of the FD resource knowledge base are designed according to the architecture of FD resource knowledge and the data types of important blocks.A bottom-up method is used to extract high confidence pattern information from public data to create KG.The schema layer of KG is managed by the ontology library,and Neo4j is used to store triples in the data layer.In KG construction,knowledge transfer is carried out from the general corpus in combination with transfer learning to assist entity and relation extraction in the face of the insufficient training corpus.A clustering algorithm with automatic feature reduction is designed for entity semantic disambiguation in the knowledge fusion stage.In this study,a systematic study is carried out on the theory of FD resource knowledge,the architecture of FD resource knowledge,the intelligent construction of FD resource knowledge,and the construction of KG,including standardizing the composition of the FD resource knowledge architecture,establishing the conceptual structure and physical structure of FD resource knowledge base,and hence constructing KG of FD resource knowledge.KG of FD resource knowledge can finally support FD practitioners to efficiently and conveniently conduct semantic retrieval of design knowledge,design analysis and decision,design knowledge management,and service.