Posted by In a previous post, I showed a lecture by Prof. Michael Wooldridge of The Alan Turing Institute on the development of data analysis methods for artificial intelligence. Not so long ago, 7 years ago, a team from Siegen University and I analyzed the behavior of managers using classical pattern recognition methods, and it was very effective.
I will briefly describe the research we conducted and recorded the behavior of managers using TransistorsHead.com’s managerial tools.
We involved 150 volunteers in the experiment for 15 months. The participants worked in small groups, usually 4-5 people each, at different universities in Upper Silesia. Each group consisted of a team leader and several colleagues. For data analysis, we used data on the behavior of 56 managers, each of whom performed between 200 and 400 managerial activities in goal setting and task planning during the study. Each managerial activity was described by a 24-dimensional feature vector.
As a result of the study, we obtained the data and published it in one of our conference series articles https://icpram.scitevents.org/
In this article, we proposed a new manager representation and matching algorithm based on manager activities and their characteristics. For the manager representation, we first represented the manager’s behavior by sequences of actions collected by existing online manager tools. Then, each action was described by multiple features in the time domain under flexible feature groups. In this case, the proposed representation method is flexible and general enough to cover most types of managers. For managers, based on manager action sequences, we first use a partial matching method to search for matched blocks in manager action sequences. Then global similarities between managers were calculated based on the matched blocks of the sequence.
See the entire article here:
Flak, O., Yang, C., Grzegorzek, M. Action Sequence Matching of Team Managers