Research And Implementation Of Semantic Association Spatio-Temporal Mining

In linked data,semantic association represents a complex graph-structured relationship formed by links between two or more entities.The current semantic association mining does not fully consider the rich spatio-temporal information in the linked data.The existing link data does not follow a unified spatio-temporal model,and the rich spatio-temporal information is often implied in the unstructured literals,which poses a challenge to the spatio-temporal mining of semantic association.This thesis proposes a semantic association spatio-temporal mining method based on spatio-temporal slice,which can effectively help users find semantic association to meet the constraints of time and space in the linked data.In this thesis,the text extraction method identifies the spatio-temporal information in the linked data,and uses GeoNames as a geographic knowledge base to eliminate the ambiguity of geo nouns using a geo noun disambiguation method based on bucket model,and then uses rules to identify spatio-temporal attributes and spatio-temporal entities,and model the linked data into temporal and spatial entity graph.In this thesis,a semantic association spatio-temporal mining method based on spatio-temporal slice is proposed,and a spatio-temporal interval table is constructed to describe the value distribution of various types of entities in the spatiotemporal attributes in the linked data.The user determines the granularity of the spatio-temporal slice by interacting with the spatio-temporal interval table,and slices the temporal and spatial entity graph,so that the temporal and spatial entity graph is divided into multi-dimensional subgraph sets,and the frequent sub-graph mining method is used to discover the link patterns and the corresponding semantic associations in the spatio-temporal slice.In this thesis,a multitarget semantic association retrieval method based on spatio-temporal constraints is proposed.It describes the user's query keywords as multiple query targets,and finally help users find semantic association to meet the needs through 1-Target retrieval,retrieval space management,and retrieval results fusion.The temporal and spatial entity graph model extends the expressive power of existing spatio-temporal models of linked data and forms a structured representation of spatio-temporal information.The spatio-temporal mining method of semantic association satisfies the temporal and spatial needs of users in mining process.The multi-target semantic association retrieval method based on spatio-temporal constraints not only extends the spatio-temporal expressive ability of traditional semantic association retrieval methods,but also improves the accuracy of retrieval.