A hybrid approach for composing groups in collaborative learning contexts.

Area of Science:

• Educational Technology
• Computer Science

Background:

• Existing methods for group composition in collaborative learning lack flexibility.
• There is a need for efficient algorithms that can handle varying group sizes and attributes.
• Current approaches often struggle with balancing homogeneity and heterogeneity criteria.

Purpose of the Study:

• To present a novel, flexible, and efficient algorithm for group composition in collaborative learning.
• To combine local brute-force search with a randomized heuristic for improved algorithmic performance.
• To provide a computationally efficient solution for grouping large numbers of students.

Main Methods:

• Developed a flexible operational framework allowing for diverse group sizes and attributes.
• Implemented an algorithmic approach combining local brute-force search within an iterative, randomized heuristic process.
• Tested the algorithm on datasets with student numbers ranging from 20 to 3500.

Main Results:

• The proposed algorithm demonstrates reasonable performance and execution times across various dataset sizes.
• Computational complexity is non-polynomial (NP) with respect to group size but polynomial (P) with respect to the number of students.
• Execution times were found to be lower than those reported in 19 related works under comparable conditions.

Conclusions:

• The proposed group composition method offers significant flexibility and efficiency for collaborative learning environments.
• The algorithm's performance is suitable for practical applications involving large student cohorts.
• Availability of datasets and source code facilitates objective comparisons and future research.