Student Evaluation System Project Proposal Student Evaluation System -Create a student evaluation system that accurately assesses the level of students’ understanding of a certain subject. -Because the system will identify the specific areas that students don’t understand, teachers will be able to better meet the needs of their students. -The difficulty in creating such a test system lies in being able to accurately identify the topics that are not understood from just a few test questions.

Approach Previous approaches: -Computer-Adaptive Testing (tailored testing) : -adapts to testee’s level -uses an iterative algorithm -provides uniformly precise scores for most test-takers but all new items in the test must be pre-tested (thus, not new) My approach: -The test engine will figure out what question to ask as it traverses through a directed graph that maps out the subtopics of the subject. -The algorithm should also be able to calculate the student’s level of mastery as well as its confidence in that estimate. -This approach will not only take into account the correctness of the answer, it will also be able to evaluate how well the student knows the principles behind the question.

Challenges Things that might prove to be hard: Figuring out the best and most effective way to evaluate student answers Creating an effective directed graph How to test students without asking too many questions Applying for approval and testing actual test subjects

Fall back plan / Testing -Since I will be testing as I go along, I can adjust goals and plans according to success/failure -One aspect of the fall back plan concerns the subject testing. Since it is human testing, the research has to go through OPRS for approval. Several possible ways to conduct testing is as follows: 1) Conduct testing among high school or middle school students who are currently taking topic of choice at school. 2) Using college students (and more advanced topic) so that only one set of consent forms is required 3) Doing an internet survey so that consent is less of a hassle -Current plan might include applying for all three possibilities so that I can just pick the most ideal course of action later

Methodology, Milestones & Deliverables By end of Fall quarter: Create initial version of directed graph Begin amassing data base of questions and problems Create a working algorithm that can create a test and evaluate answers. Apply for approval from UCLA Office for Protection of Research Subjects By end of Winter quarter: Refine both directed graph and algorithm Begin testing part of research: get some feedback from test subjects By end of Spring quarter: 6) Conduct more tests Analyze experimental results Write final project thesis

Conclusion This research is cool because -It hasn’t been done before -America doesn’t focus nearly enough on education -This could help and make education more effective