Construction And Query Of Shopping Mall Indoor Space Knowledge Graph

With the development of new positioning technology,Indoor location services have also become a research hotspot in recent years.People's daily life is located in the indoor and outdoor space,especially,most of the time spent in the indoor environment.Different people have different cognitions and needs for indoor space services,and have different semantic restrictions.Therefore,the existing outdoor space related technologies cannot be directly applied to indoor space.As a widely used indoor location service,indoor space query is still searched by keyword matching and returns the result to the user,which is impossible to provide users with a semantically relevant,clear,and more accurate retrieval structure.The emergence of knowledge graph provides an effective solution for intelligent retrieval.This paper takes care of the semantic query of shopping mall space.In view of the existing work,the lack of knowledge map in the shopping mall space can not support the efficient indoor space query problem.We have studied the construction method of the space knowledge graph of shopping malls.The specific research results are as follows:(1)This thesis proposes a framework for building indoor space knowledge maps.The the relationship between entities and entities is more concerned.(2)This thesis constructs a shopping mall indoor space ontology,which describes the structural pattern of the indoor s pace knowledge graph of the mall.Through the analysis of indoor space knowledge,defined the concepts and attributes involved in the interior space,and completed the construction of the shopping mall ontology.(3)This thesis extracts information from online resource data,extracts the attributes of entities,entities and relationships between entities,and improves the knowledge graph of indoor space in shopping malls.Firstly,the statistics-based and rule-based methods are used to extract the entities in the data source for entity identification.Secondly,the relationship extraction is transformed into a classification problem,and the relationship is expanded.Finally,the mall attribute is expanded by constructing a wrapper.(4)This thesis implements a prototype system that supports indoor space query in shopping mall,which is divided into:corpus preprocessing function module,similarity matching module,and personalized preference query module.The system uses the knowledge graph to answer the user's indoor space query,and can provide query results that more satisfy the user's needs and personalized preferences.