The Construction And Application Of Commodity Knowledge Graph In The Field Of Catering

In the past few years,the rapid development of China's economy and the rapid expansion of the Internet have made the Internet + economy slowly take root in the country,which has led to significant changes in the people's living,living,and living structures.In the retail sector,more and more people choose online shopping to improve efficiency,and the rise of the takeaway industry in the catering sector has also increased the demand for high standards and fast-paced diet for both students and white-collar workers.Along with this,the Internet has created an integrated platform model for online and offline life and entertainment.On the platform,it also accumulates a wealth of information about businesses and consumers,involving products,product portraits,user portraits and other aspects.How to dig out the concepts and semantic relationships behind this information and integrate them into a powerful knowledge network to provide potential consumers with more intelligent services such as accurate search,personalized recommendations,and automatic question answering,thereby empowering knowledge and becoming a This is an important research question,and the construction and application of knowledge maps in the professional field of catering products is the positioning of a solution to this problem.At present,there is a lack of open source domestic product knowledge maps in the participating fields that are close to life.There is difficulty in constructing high-quality product knowledge maps in this field.This article chooses to build product knowledge maps in the emerging food and beverage industry.This article uses the top-down approach to build a standard product knowledge map in the food knowledge map construction method in the catering industry.First of all,a schema of the product knowledge map is established based on the characteristics of the product.Seven main entities are defined,which are standard dishes and dishes.Category,ingredients,cooking method,taste,region,efficacy,named entity recognition through the use of the NLP cutting-edge model BERT-CRF,construction of a domain attribute dictionary and expansion of the dictionary,and use of a supervised BERT classification model to map entities and Relationship recognition,taking into account the particularity of the dish product name and the attribute relationship between different products,establishes the upper and lower and parity associations of the dish.The standard library is constructed from two different perspectives of characters and attributes.The proposed model and The method performs well on both the training and validation sets.The main research contents and contributions of this article are as follows:(1)Study how to build a knowledge map in the catering industry,including knowledge extraction,knowledge fusion and knowledge expansion.(2)Design the architecture of the product knowledge graph from top to bottom,that is,the entity layer,attributes and relationships between entities contained in the pattern layer of the graph.(3)Construct a domain dictionary,define multiple domain features,and implement a method for aligning dish entities and attribute entities through the domain dictionary.(4)Training the dish classification model through the BERT model.The BERTCRF model is used to realize the entity recognition task of unstructured data.A model method for domain dictionary expansion is provided.(5)Investigate the design of the upper and lower structure of the standard dish library,design different inheritance methods for different attributes,and solve the problems of the diversity of similar dish names and the lack of attributes.The experimental results show that the entity recognition and relationship classification implemented in this paper have high accuracy and recall,which can ensure the construction of a high-quality knowledge map.In order to improve the application value of the knowledge map,this article provides a variety of application schemes based on the knowledge map.At the same time,it can effectively manage and iterate the entity and semantic relationship in the knowledge map through the management of the dictionary,which can help users to fully use product knowledge.