处理模糊逻辑或缺失数据-az-900 exam
4.3处理模糊逻辑或缺失数据贝叶斯网络等统计推理技术由于其在处理不确定性问题及模糊逻辑方面的出色表现已经在工程上得到广泛应用,相比这些基于统计的方法,规则引擎依然有它的长处:易理解、易修改、计算量小等。然而它的缺点也逐渐明显:不能处理缺失的数据及模糊的逻辑。如何发挥两者的长处,即让Rete这样的确定性推理算法处理模糊缺失数据的问题近年来也相应地得到研究者的重要关注[3]并一度成为研究热点。3.4节说明了近年来在这个方面的研究趋势。在未来,这种对Rete的随机化、经验化进行改进以适应不确定性的推理问题依然需要更深入的研究。目前研究的关注点在于逻辑的泛化,包括增加参数等方法,未来的研究可以结合更多的模糊推理技术如贝叶斯网络、概率图推理等,使Rete算法既能发挥产生式推理引擎易理解、易修改的优点,又具备更强的表达能力和逻辑处理能力。结束语规则引擎是人工智能的一个重要分支,它的出现极大地便利了知识表示、数据推理、专家系统等。Rete算法的提出极大地提高了规则引擎的推理效率,并为其他相关领域的研究提供了新技术和新思路,研究前景广阔。近年来,虽然仍有很多研究人员致力于该方面的研究,发表了很多新的研究成果,但随着业务逻辑的日益复杂、数据的日益复杂庞大及新的计算平台如分布式和云计算平台的日益普遍,目前的方法将面临诸多问题:(1)处理海量数据与规则;(2)处理快速变化的数据;(3)模糊逻辑和缺失数据的推理。本文对产生式推理系统中一个具有代表性的算法Rete作了一个比较全面的综述,分析了Rete算法的各种改进方法、研究现状和待解决的问题。对这类问题的研究与解决必将进一步推动数据推理乃至人工智能领域的新发展。参考文献[1] Forg C L.Rete:a fast algorithm for the many pattern/many ob- ject pattern match problem[J].Artificial intelligence,1982,19: 17-37 [2] Apache Drools Project.Apache Drools Expert User Guide[OL]. http://docs.jboss.org/drools/release/5.3.0.Final/droolsjb- pm-integration-docs/html_single/index.html,2008-09-07 [3] Sottara D,Mello P,Proctor M.A configurable Rete-OO engine for reasoning with different types of imperfect information[J]. IEEE Transactions on knowledge and data engineering,2010,22 (11):1535-1548 [4] Wikipedia[OL].http://en.wikipedia.org/wiki/Rete_algori-thm, 2009-08 [5] Xiao Ding,Tong Yi,Yang Hai-tao,et al.The Improvement for Rete Algorithm[C]∥Proc of The International Conference on Information Science and Engineering(ICISE 2009).NJ:IEEE, 2009:5222-5226 [6] Cheng Fu-chiung,Chen Huei-huang,Perng J-H.Parallel Execu- tion Production Systems[C]∥Proc of the IEEE Second Sym- posium on Parallel and Distributed Processing.NJ:IEEE,1990: 463-470 [7] Walzer K,Breddin T,Groch M.Relative temporal constraints in the Rete algorithm for complex event detection[C]∥Proc of the Second International Conference on Distributed Eventbased Sys- tems.New York:ACM,2008:147-155 [8] Zhou Dong-dai,Fu Yi-fan,Zhong Shao-chun,et al.The Rete Al- gorithm Improvement and Implementation[C]∥Proc of the 2008International Conference on Information Management,In- novation Management and Industrial Engineering.NJ:IEEE, 2008:426-429 [9] Walzer K,Groch M,Breddin T.Time to the Rescue-Supporting Temporal Reasoning in the Rete Algorithm for Complex Event Processing[C]∥Proc of the 19th international conference on Database and Expert Systems Applications.Berlin:Springer, 2008:635-642 [10]Florian S,Nour S,Georg L.Adapting the Rete-algorithm to e- valuate F-Logic rules[C]∥LNCS 4824:Proc of the International Symposium on Rule Interchange and Applications(RuleML’ 07).Berlin:Springer,2007:166-173 [11]Lagun E.Evaluation and Implementation of match algorithms for rule-based multi-agent systems using the ecample of Jadex [D].Hamburg:University of Hamburg,2010 [12]Boyer J,Mili H.Agile Business Rule Development Process,Ar- chitecture,and JRulesExamples[M].Berlin:Springer,2011:161- 174 [13]Scals D J.Efficient Matching Algorithms for the SOARlOPS Production System[OR].ftp://reports.stanford.edu/pub/cstr. old/reports/cs/tr/86/1124/CS-TR-86-1124.pdf,1986 [14]Ishida T.An optimization algorithm for production systems[J]. IEEE Transactions on Knowledge and Data Engineering,1994,6 (4):549-558 [15]Brownston L,Farrell R,Kant E,et al.Programming Expert Sys- tems in OPS5:An Introduction to Rule Bused Programming [M].MA:Addison-Wesley,1985 [16]Tamasmeszaros,Vadasz B.An extension to the Rete match al- gorithm:supporting both forward and backward chaining[OL]. http://www.mit.bme.hu/~ meszaros/me/pubs/tempus94. ps.gz,2012-03 [17]Rangel P,Junior J G C,Ramirez M R,et al.Context reasoning through a multiple logic framework[C]∥Proc of the Sixth In- ternational Conference on Intelligent Environments.NJ:IEEE, 2010:116-121 (下转第33页) ·21·
725.89KB
文件大小:
评论区