Streaming reasoning method combining a variable memory and a multivariable connector

Abstract

The invention provides a streaming reasoning method combining a variable memory and a multivariable connector. The streaming reasoning method comprises the following steps: processing RDF streaming data based on an OWL Horst rule set; According to the streaming reasoning method, under the background of a knowledge graph, firstly putting forward a variable memory, and avoiding redundant storage anddata transmission expenditure in the form of taking rules as units while the query speed in specific RDF streaming data is increased and looked up. Providing a storage method of the multivariable connector for pre-judging whether the rule is activated or not; And finally, in combination with the storage design scheme, designing a streaming reasoning algorithm EMSR, and effectively realizing efficient reasoning for the streaming RDF data. By additionally arranging the variable memory and the multivariable connector, the triad of the same type can be used for a plurality of rules, redundant storage in the form of taking the rules as units is avoided, and data network transmission is reduced.

Description

technical field [0001] The invention relates to the technical field of streaming RDF data reasoning under knowledge graphs, in particular to a streaming reasoning method combined with variable storage and multivariable connectors. Background technique [0002] In the early days, many scholars used parallel reasoning algorithms based on RDFS rule sets. Gu Rong et al. proposed an efficient and scalable semantic reasoning engine YARM based on MapReduce, which processed pattern triples and instance triples data separately, and optimized the order of rule execution, so that reasoning could be completed in one MapReduce; Chevalier et al. An effective incremental reasoner Slider is proposed, which maps rules to independent modules and performs distributed reasoning. At the same time, vertical partitioning and multiple indexes are used to solve repeated calculations to improve reasoning efficiency; Bo et al. proposed a novel method in , use Transitive Reasoning Tree (TRT) and Under...

AI Technical Summary

Problems solved by technology

However, an instance triple may satisfy the antecedents of multiple inference rules. According to the existing practice, it will be stored multiple times

In the case of massive knowledge graph data, the redundant storage of data marked by rules not only wastes storage space, but also greatly increases the overhead of network storage and transmission

[0007] (3) There are multiple connection variables in the precondition of the rule, which need to be connected multiple times

Overall inefficiency after many iterations

[0008] (4) The triples of streaming reasoning are not loaded once, and need to be loaded multiple times

If the rule cannot be activated after iteration, a series of iteration time is wasted

Images