Automatic Grading System For Multi File Program Analysis With Flexible Grading

This thesis present method to handle multi file program analysis with flexible grading which are dynamic and static implemented practically in automatic grading system.Both dynamic and static grading can analyze multi file program for solving one programming assignment problem.It can be advantage to handle complex programming problem which requires more than one program file.The feature support modular or distributed programming such as creating user customized library without having to write all the code into main program.In here,user customized library can be referred to external sources.Main program can use external sources by call written in preprocessor or import statement similarly like using another class.The approach for multi file program in each grading method is distinguished by following their principle concept.Dynamic analysis use compiler feature in compilation process.It takes advantage of object file linker in compilation to link multi file program.Static analysis parse the program into Abstract Syntax Tree(AST)as chosen program representation.The AST will be mapped into AST data map for standardization and program matching.The information of preprocessor is also used for linking external sources called in main program by multi file program linkerfusion algorithm.Both present techniques give instant scoring and feedback supporting multiple file program submission for solving one programming problem.In static analysis,standardization process used for problematic code removal by problem binder tag,unused function removal and function sequence ordering based on function call.The process resulting transformed program with correct function sequence based on call and more efficient program nodes if unused or problematic code existed in original program.In program matching,it successful tackle structure variance problem by previous standardization process and code variance by AST simple tree matching using tag classifier to make variance tolerances for similarity nodes of both compared program.Finally,the result show improvement in grading precision accuracy which gives reliable grading score delivered with intuitive system.This is original contribution to handle multi file program analysis with flexible grading as consideration of modularity and big scale of programming problem complexity.