This repository host codes from the paper XXX (Pelletier and Wirtz, 2023).

The aim of our study was to gain insights on the team composition that supports successful novelty. To do this, we developed a measure that allowed us to identify the characteristics of researchers and examine the team composition that foster potential breakthrough novelty. Our analysis was based on the Pubmed Knowledge Graph, a digital library of scientific publications in health science. We investigated the relationship between our measure and expert perceptions of novelty (From Faculty Opinion), combinatorial novelty, and impact measures. Our findings suggest that teams with a highly diverse background fosters combinatorial novelty but that the relation follows an inverse u-shape. Furthermore a mix of highly exploratory and highly exploitative individuals are more likely to produce breakthrough novelty. Our study highlights the importance of team composition in facilitating successful novelty in scientific research. By identifying the factors that contribute to breakthroughs, we can better support research proposal with potential to advance the boundary of science but also create environments that facilitates the collaboration between exploratory and exploitative authors.

├── requirements.txt
│ 
├── Data
│   └── regression.csv
│
└── Paper
    ├── cleaner
    ├── download
    ├── Regression
    ├── Run_indicators
    └── Stats

regression.csv is available on zenodo. If you are interested in working with pkg we left the cleaner,download and Run_indicators folder that we used to create this regression.csv

The table and plots were done using the folder Regression.