Research On Discovering Knowledge Of Problem Solving Of Program Design Resources Based On Natural Language Processing

With the generalization and development of computer network education,programming education,which thought as nuclear course of computer basic scientific education,has been more and more concerned.recently,ACM International Collegiate Programming Contest(ACM/ICPC,or ACM for short)has evoked students great interst in learn programming,more and more students commite in the ACM,many experienced ACM participant and amateur share the reports of solving ACM problem on the internet,which have been formed a large number of programming resource.but regretfully,these problem solving resources haven't been well-organized so far,and they can't be used effectively by general programming amateur.The research of this article is a combination of Natural Science Foundation of China(NSFC),the phenomenon of "Isolated Island of Knowledge" on the web was put forward by this project,that is the knowledge concepts structure of these knowledge resources in our mind(composed by the internal relations among knowledge concept)was different from the physical structure in the Internet(connected by hyperlinks).So it leaded to a phenomenon on the Internet that we called"Isolated Island of Knowledge".The Research of this article is more specific,the research object is programming resource on the internet.There are two problem when we connect the programming resources.First,get the name of the algorithmic knowledge which contain by the programming resources.Second,recognize the algorithmic knowledge of the programming resources.After that,we can organize the programming resources by the structure of knowledge.This article proposes a method which is based on natural language process to get the name of algorithmic knowledge from programming resources.This method consists of discovering the patterns of strings which contain terms of algorithmic knowledge,extracting from programming resources the strings that probably contain terms of algorithmic knowledge according to the discovered patterns,finding the participles most likely appearing in the terms of algorithmic knowledge,and fetching the terms of algorithmic knowledge according to these participles.This method has incremented 11.2%algorithmic knowledge and 49.5%terms of algorithmic knowledge in comparison to the manual collection of terms of algorithmic knowledge which was obtained by previous work.The names of algorthmic knowledge points can't be used for algorthmic knowledge.Because a algorthmic knowledge point might have one or more designation,we have to classify these names of algorithmic knowledge points.The research group has summarized an expert document,which contains algorithmic knowledge points and one or more designation of the algorithmic knowledge point.the name of algorithmic knowledge we get can add to the expert document.This article proposes a method to expansion the exsit knolwdge structure use the knowledge name.This article also propose a method to recognize the algorithmic knowledge which the programming resources contaion.First,this method constructs the ontology model of the acm problem and its corresponding reports.Second,concreting ontology individuals according to the semantic when the reports contains name of algorithmic knowledge.Third,concluding the algorithmic knowledge from the ontology individuals.This methods proceeds from the semantic knowledge when the reports contains name of algorithmic knowledge,which improves the accuracy.