Problem-Based Learning, Complexity, and Similarity Assessment in Case-Based Reasoning

• In Event: 53.081 - Roundtable Session 21
  In Roundtable Session: 53.081-12 - All About Problems: Considerations of Problem-Based Learning Problems for Learning Processes and Outcomes

Abstract

Purpose
One of the criticisms of problem-based learning (PBL) and other inquiry-based methods is that students lack the previous experience required to solve complex, ill-structured problems. Proponents argue that these methods better reflect the conditions that practitioners often face; therefore, they are better prepared to problem solving in these contexts. However, critics argue the complexity and gap in experience precludes meaningful learning (Kirschner, Sweller, & Clark, 2006). Additional discussions are needed to understand how learners can overcome the lack of experience.

Theoretical Framework
Implicit within the argument is that the complexity stems from the lack of previous experience needed to information a solution to a presented problem within the classroom. Similarly, CBR researchers have argued that prior problems are key to understanding and solving new problems (Schank, 1999). Case-based reasoning theory (CBR) specifically suggests that when presented with a new problem, learners will retrieve prior experience based on a set of indices (labels) from memory. If applicable, s/he will then reuse the lessons learned in order to inform a solution. CBR suggests two important implications for PBL. First, an important aspect is how learners retrieve and reuse previous experience. CBR also suggests that one way to ascertain the complexity of a problem is by understanding the gap in prior experience and the main PBL problem to solve (Tawfik & Kolodner, 2016) .

CBR theorists suggest that one way to bridge the experience gap is by providing a representative set of cases that introduce learners to the problem space germane to the problem. When using case libraries, an important element of retrieval and reuse is how learners see cases as relevant, a process known as similarity assessment. Based on previous literature (De Mantaras et al., 2005), we argue use case library design should consider the following approaches to better support similarity assessment :
A. K-D trees: K-d trees suggests that cases should be congregated into groups of cases such that groups contain cases based on a similarity measure. When cases are retrieved, the retrieval processes is conducted based on a set of similarity bounds to determine which cases should be considered first.
B. Footprint cases: In footprint cases, the footprint is most similar case (reference) to the problem to be solved. Another smaller set of cases are attached to the footprint (reference) cases and retrieved once the similarity between the main problem and footprint (reference) case has been identified. This approach allows learners to efficiently eliminate unnecessary case during retrieval.
C. Validation procedures: When undergoing similarity assessment, CBR also suggests that validation measures are embedded as part of the retrieval process. Rather than retrieve just based on indices, the retrieved case is only identified as beneficial for reuse if a series of validation measures are achieved.

Scholarly Significance
Although similarity assessment mediates the retrieval and reuse, much of the case library literature has focused on learning outcomes as opposed to the learning process. As these similarity assessment processes are explored, educators may better address the complexity concerns from critics of PBL.

Author

• Andrew Tawfik, University of Memphis