| 摘要 |
<p>A method of grading the performance of a student in a test sat by a plurality of students, the method comprising: 5 receiving an answer from each of said plurality of students; generating a knowledge representation network associated with each said answer, each said knowledge representation network defining semantic relationships between elements in said associated answer; generating a master knowledge representation network associated with a 10 plurality of said answers, said master knowledge representation network defining semantic relationships between elements in said plurality of said answers; determining a master ranking indicative of the performance of each of said plurality of students relative to the other students by comparing each of said knowledge representation networks with said master knowledge representation network; 15 distributing to at least some of said plurality of students a pair of answers to determine a pair ranking for said pair of answers, wherein each answer in said pair is from different students and wherein said pair ranking indicates which of said answers in said pair of answers is better than the other answer in said pair; receiving said pair ranking for each said pair of said answers, 20 comparing each of said pair rankings with said master ranking to determine an agreement value indicative of consistency between said pair ranking and said master ranking; redistributing pairs of said answers dependent upon said agreement values; receiving additional ranking information following said redistributing step; and 25 determining an overall ranking of said students dependent on said pair rankings and said additional ranking information. (Fig 2) TEACHER DEFINES TEST (STORE IN DATABASE) STUDENTS SIT TEST ANSWER SCRIPTS STORED IN DATABASE EACH SCRIPT CODIFIED INTO AN INDIVIDUAL SEMNET SINGLE MASTER SEMNET CREATED FROM ALL INDIVIDUAL SEMNETS COMPARE EACH INDIVIDUAL SEMNET AGAINST MASTER SEMNET TO DERIVE INITIAL RANKING ALLOCATE RANDOM PAIRS OF SCRIPTS TO STUDENTS WHO HAVE TAKEN TEST & SEND THEM SCRIPTS COMPARE STUDENT RANKING WITH INITIAL RANKING TO GENERATE AN AGREEMENT MEASURE FOR EACH SCRIPT YES IS AGREEMENT MEASURE GREATER THAN THRESHOLD NO IS THERE IFS220 TIME LEFT NO OUTPUT RANKING DERIVED BY STUDENTS S218 ,S222 REALLOCATE SCRIPTS (OPTIONAL) OUTPUT - FOR RANKING BY FIG 2 MARKS/GRADES</p> |
