Novel Machine Learning Approach for Automatic Employees' Soft Skills Assessment: Group Collaboration Analysis Case Study

Machine Learning (ML) based assessment of soft skills is a challenging domain in social computing. In today's enterprises, soft skills are regarded as one of the most crucial components for success. Based on data obtained from internal organizational sources, we evaluated the viability of utilizing ML to assess Group Communication Analysis (GSA) skills. We leveraged Latent Semantic Analysis (LSA) to understand and learn cognitive phenomena extracted from the communication extracted from the issue tracker of several popular open-source projects, with a specific input representation that allows the model to look for contextual cues in consecutive utterances. We looked at the data of 1520 engineers from three open-source software development projects that worked on an internal line of business operational systems. The ML models revealed five interpersonal and intrapersonal socio-cognitive GSA metrics in 10583 utterances and procedures, which were integrated into five clusters that assign socially engaged employee roles. Because there has not been any research on assessing employees' Soft Skills utilizing ML, our proposed method is a first in this field, relying on internal organizational datasets to obtain reliable Soft Skills assessment evaluation.