Abstract:

University classes are often populated with students of different majors and consequently different levels of background knowledge. It is necessary to perceive the knowledge level of the class as a whole and also of individuals to create an effective lesson series. Additionally it is essential to create a learning environment to foster in-depth learning and reflection on the subject matter.

Concept Maps (CM) are a proven tool to on the one hand get a measure of the knowledge the students have learned, and on the other hand to have the students reflect on the learned subject matter and to conceive coherencies.

A CM is a structured graph where nodes are labeled with terms or concepts and connections between the nodes (edges) are labeled with a verb as a relation. The creation of these networks takes effort and helps with reflecting and understanding the subject matter.

With our study we aim to determine how much and in what way students’ knowledge is affected and processed through the use of Concept Maps. Additionaly if there are differences in knowledge processing between different majors.

We let students create three consecutive concept maps over time in the semester. Our Principles of Artificial Intelligence course is attended by engineering, computer science and data analysis majors. In total 15 students attended the study but only roughly half of them sent in all three CMs.

The overall majority of CMs were written on paper. Several people wrote on their tablets and one student used dedicated software to create their CM.

To analyze the data, we classified the used terms and their relationships in the CMs. Due to the small sample size, we conducted a manual qualitative analysis of individual CMs only. We conducted follow-up interviews with students which showed that they liked the exercise and planned to do CMs to review their knowledge in other fields. The collected CMs show an expected distribution of background knowledge in the first elicitation with computer science majors being more precise in wording, but data analysis students having a broader concept knowledge in machine learning. The second and third elicitation shows a knowledge consolidation overall. But with the low attendance conclusions are hard to make, as students that couldn’t follow the subject matter are likely to have not worked on CMs.

Even with the rather low number of participants that completed all three CMs, the data shows a growing understanding of the subject matter. Differences between students from different fields could not be determined. Concept Maps are advantageous for assessing students’ background knowledge and promoting reflection on the subject matter simultaneously. Some students seem to like the method according to personal communications. We plan to incorporate the method in further classes to get more data on learning behavior and CM usage and benefit.

Keywords:

Concept Map, Reflection Techniques, Knowledge Level Measurements.