Distributed real-time rule reasoning and scheduling method

Abstract

The invention discloses a distributed real-time rule reasoning and scheduling method, comprising the following steps: 1) establishing a rule graph corresponding to a productive rule set; 2) convertingthe rule graph into a data operation graph; 3) dividing the data operation graph into a plurality of sub-graphs according to the received fact object by the rule inference system, and generating an inference job for each sub-graph; 4) setting priorities of the inference jobs according to the processing time of the inference jobs, creating n scheduling queues in the rule inference system, settingdifferent weight attributes for each scheduling queue, and calling the inference jobs into the corresponding scheduling queues according to a corresponding relationship between the priorities and theweights; 5) when ready tasks are submitted in different scheduling queues at the same time, selecting the inference task with the maximum weight in the scheduling queue, and creating and executing theinference task, and 6) deleting the inference task corresponding to the inference task from the scheduling queue to which the inference task belongs until all the scheduling queues are empty after the inference task is executed.

Description

technical field [0001] The invention belongs to the field of rule system and real-time technology, and in particular relates to a distributed real-time rule reasoning scheduling method. Background technique [0002] Rule reasoning expresses domain knowledge as rules, triggers and executes rules by matching rules with facts, so as to make automatic decisions. It has been widely used in various expert systems such as medical diagnosis, geological exploration, and intelligent decision-making. [0003] In recent years, with the popularization of Internet of Things technology, Internet of Things has been widely used in various industrial fields. At this time, the IoT application collects external environmental data through a series of sensors, monitors the continuous event flow, identifies scenes that need attention and makes automatic decisions, so as to execute actions in real time to respond to the identified scenes. Therefore, rule reasoning has begun to be applied to typic...

AI Technical Summary

Problems solved by technology

[0005] Traditional rule reasoning systems, including CLIPS, Drools, and JESS, etc., have low rule matching efficiency due to the limitations of their stand-alone machines. The patent application with publication number CN107145949A has improved this, and designed a The real-time rule reasoning method in a multi-core environment has improved the matching efficiency of rules to a certain extent, but because its rule reasoning is still based on a local stand-alone system, its performance is still difficult to meet the system resource requirements brought by massive real-time data processing

[0006] At present, scholars have proposed some distributed real-time reasoning methods for streaming data (Zhang Qi. Research and implementation of distributed rule matching system based on MapReduce [D]. Zhejiang University, 2011; Yi Chen, BehzadBordbar. DRESS: A Rule Engine on Spark for Event Stream Processing. BDCAT 2016; Zhang J, Yang J, Li J. When Rule Engine Meets Big Data: Design and Implementation of a Distributed Rule Engine Using Spark, 2017IEEE Third International Conference on Big Data Computing Service and Applications( BigDataService).IEEE, 2017.), these methods realize the parallel reasoning of the rules by using the distributed computing framework, and the patent application with the publication number CN108549939A proposes a load-balancing task distribution algorithm on this basis, which further improves the The efficiency of rule reasoning, but these methods do not consider the deadline of reasoning, and cannot process as much data as possible before the deadline, so that they cannot guarantee a timely response to the scene of interest.

Images