Seattle, Washington
June 14, 2015
June 14, 2015
June 17, 2015
978-0-692-50180-1
2153-5965
Computers in Education
Diversity
12
26.1588.1 - 26.1588.12
10.18260/p.24924
https://peer.asee.org/24924
250
Dr. Jamieson is an assistant professor in the Electrical and Computer Engineering department at Miami University. His research focuses on Education, Games, and FPGAs.
Towards a Better Graphlet-based Mind Map Metric for Automating Student FeedbackAbstractIn this work, we evaluate a new mind map analysis metric that compares an experts mind map(called the criterion map) to a students map to evaluate how similar the two maps are. In previouswork, we evaluated graph metrics such as graph density, node degree, RGF-distance 1 , and anedge to edge math metric that we created. The later two metrics, when applied in a longitudinalexperiment, showed promising results in capturing student improvement.The context of this research is the idea of a computer-based system that allows students to testtheir understanding of terminology in a particular subject or ﬁeld and receive immediate feedbackon how they compare to experts.This work focuses on graphlets, which are small graphs of with 3 to 5 nodes. The RGF-distancemetric, in our previous studies, uses a count of each type of graphlet which then can be comparedwith the criterion maps graphlet count to determine if graphs are similar, and this technique wasdeveloped to compare proteins with computational efﬁciency, where proteins are signiﬁcantlylarger graphs compared to our mind maps. Mind maps, which are a connection of terms withlines, are small graphs and it is possible to directly compare all graphlets.To determine our new direct matched graphlet based metric, we analyzed the same data from the 2years of longitudinal based experiments previously studied in 2011 and 2012. We built a tool thatallowed us to extract and analyze data on which graphlets had sufﬁcient frequency to be useful forstudent feedback, and we create a new metric based on this analysis. In the results, we ﬁnd thatthis metric has similar behavior to both RGF-distance and our edge based match metric whereimprovement of a student’s mind maps is observed over the period of a semester course. This newmetric is better than RGF-distance since it compares actual matched graphlets allowing us to usethe presence and absence of these graphlets as direct feedback to the student. Additionally, similarto our previous studies, there is no correlation between grades and any of our metrics.Figure 1 shows one result that is a comparison of the RGF-distance metric and the match metricin both 2011 and 2012 data set. The assumption made for this work is, ”students are learning overthe semester”, and therefore, a metric that shows learning will also show improvement over ourlongitudinal studies. For RGF-distance, the metric will move downwards towards 1 as the studentmap and criterion map become more similar, and for the GraphletMatch metric the value willmove upwards towards 1 where 0 reﬂects no matching. Figure 1: Results for students in the 80 to 90 range for RGF metric and our new metric.References[1] Natasa Przulj, Derek G. Corneil, and Igor Jurisica. Modeling interactome: scale-free or geometric? Bioinformatics, 20(18):3508–3515, 2004.
Jamieson, P., & Eaton, J. (2015, June), Towards a Better Graphlet-based Mind Map Metric for Automating Student Feedback Paper presented at 2015 ASEE Annual Conference & Exposition, Seattle, Washington. 10.18260/p.24924
ASEE holds the copyright on this document. It may be read by the public free of charge. Authors may archive their work on personal websites or in institutional repositories with the following citation: © 2015 American Society for Engineering Education. Other scholars may excerpt or quote from these materials with the same citation. When excerpting or quoting from Conference Proceedings, authors should, in addition to noting the ASEE copyright, list all the original authors and their institutions and name the host city of the conference. - Last updated April 1, 2015