Research On Regular Path Query Method On Uncertain RDF Graph

Knowledge graph is the core infrastructure of AI,and has taken RDF data model as the de facto standard of its data layer.At the same time,due to the existence of uncertainties such as noise,it is difficult to complete the knowledge representation,and uncertain information appears in the RDF Knowledge graph.The traditional RDF query method cannot meet the application requirements of the uncertain RDF Knowledge graph.Querying the uncertain information in the Knowledge graph data has become an urgent and challenging task.Navigation based path query based on regular expressions is a common query method in RDF data queries.Based on existing research,thesis designs and implements two regular path query methods on uncertain RDF graphs.The main work is as follows:(1)Aiming at the problem of regular path query on uncertain RDF Knowledge graph,a probabilistic automata based regular path query method for uncertain RDF map is proposed.This method converts the uncertain RDF graph data into a probability matrix,uses the probability equation of the state machine to describe the behavior of the state machine,and calculates the probability to determine the next state of the Finite-state machine.In space,this method is more effective in representing it as a transpose of row vectors.The experimental results show that this method outperforms the benchmark algorithm in terms of query response time and precision.(2)A finite automata based method for uncertain RDF graph regular path query is proposed to address the problem of large-scale regular path queries.This method transforms the path into a string query problem by constructing an index parse tree,and proves the correctness and Completeness of the finite Automata theory theory.The process of converting NFA into equivalent DFA in the process of converting regular expressions into NFA is called deterministic automata.The difficulty of deterministic algorithms also directly affects the matching efficiency of NFA/DFA hybrid methods.Finally,this method performs matching verification on candidate strings and sets probability thresholds to filter the final result set.Through multiple comparative experiments on different data scales,it is shown that this method has good results in index construction and regular path queries.(3)Based on the above two query methods,a regular path query system for uncertain RDF graphs has been developed.The system extends SPARQL language,and realizes regular path query on the uncertain Knowledge graph.By selecting different query thresholds,it realizes the visual display of query results,which lays a foundation for the application research of Knowledge graph under intelligent environment.