Research And Design Of Intelligent Correcting System

Automatic correcting of subjective questions is a complex problem, which goes beyond any obvious algorithms ,and there is no mature solutions. In the area of automatic correcting, correcting of objective questions, such as choice, is already quite mature, while it is still lacking in the area of subjective automatic correcting.This paper researchs about automatic correcting of subjective questions, and designs an intelligent correcting system used for subjective and objective questions correcting.The intelligent of this system is due to introducing expert system based on rule engine.The system can separate data and knowledge from process control,so that the knowledge-based rules can be independent of the computer program,which can reduce the complexity of the automatic correcting of subjective questions obviously.The system is based on J2EE open-source architecture, using Jboss Drools rules engines and the HowNet semantic network for similarity calculating, and using Lucene search engine to provide the segmentation framework for the automatic correcting. This paper designed a regular archiving module to solve the potential problem of massive data, using spring AOP managed cache to cache data, in order to improve system's performance. The system uses UML modeling system for the design of use case and class. This paper also analyzes the information extraction of virtual experiment, and looks into the future of automatically correcting of virtual experiments.