Team Anneal - team formation

Professionals working as an effective team are the key to success across disciplines. While student team work is becoming increasingly important in education, it frequently leads to frustration and unhappiness for students.

Purposeful formation of student teams ensures groups have the desired characteristics for particular learning activities, be they semester-long projects or active-learning classes in collaborative spaces.  TeamAnneal is a team formation software tool which is used to create student groups based on demographic information, academic performance, class sign-on information or other constraints.

How does it work?

TeamAnneal is based on a simulated annealing algorithm, a probabilistic optimisation approach suitable for large-scale global optimisation problems.  It is modelled on metallurgical annealing, in that it finds a low cost (energy) state as a system is cooled.  Simulated annealing requires a suitable cost function – some function of the properties of the system that is to be minimised.  In TeamAnneal, this cost function is defined by the user in the form of constraints on various student characteristics and their relative importance. 

Examples of constraints currently supported by TeamAnneal and in use are:

• Size constraints – including support for hierarchical groups, e.g.
  • “Team” size is ideally 6, minimum 5, maximum 7 members
  • “Group” size is ideally 4, minimum 3, maximum 4 members
  • “Table” size is ideally 3, minimum 2, maximum 3 “groups” and “group” size is ideally 3, minimum 2, maximum 3 members
• Count constraints – with a desired min/max/exact number of members meeting some criteria, e.g.
  • each “Team” must have at least 2 members whose “Major” is “Electrical Engineering”
  • no “Team” should have exactly one member whose “Gender” is “Female”
    • this means we want 0 or 2 or more members who are female in each team
  • each “Team” of size 3 must have exactly 3 members whose “GPA” is ≥ 5.0
    • constraints can vary depending on the team size, in this case, where the ideal team size was 4, only better performers (GPA wise) were put in smaller teams of 3
  • each “Group” ideally has at least two members whose “CSSE2002 result” is “Pass”
• Similarity constraints – where certain characteristics of members should be similar or different
  • each “Team” should have members with “GPA” as similar as possible
  • each “Table” ideally has members with “GPA” as different as possible
  • each “Table Group” should have “Previous Table Group” as different as possible
    • this constraint ensures that new groups are different from old groups

The weighting of constraints is based on the words used – “must have” constraints are weighted more heavily than “should have” constraints which are weighted more heavily than “ideally has” constraints.

Project team/Contacts

Project Status

The web interface for TeamAnneal has been completed and is available for use.

Resources

TeamAnneal can be accessed at https://teamanneal.uqcloud.net

TeamAnneal presentation - eLIPSE eLearning Tools Showcase 25 May 2022 - video.

The following resources for users of TeamAnneal are planned:

• Shareable Library of Constraints

• Documentation within TeamAnneal to support users in their use of the tool. This will include:

  • help on using the tool itself (e.g. how to construct constraints),
  • information on the types of data available to build constraints, and
  • guidelines on the selection of appropriate constraints in a variety of situations.

Technologies Used

C++, JSON

Acknowledgements

The project was initially funded by a Faculty of EAIT Teaching and Learning Grant in 2015.

Publications

L. Kavanagh, D. Neil, and J. Cokley (2011). Developing and disseminating team skills capacities using interactive online tools for team formation, learning, assessment and mentoring. Project Report, Australian Learning and Teaching Council, 2011.