Github Collaboration Network

Non-Backtracking Walk

The renewal non-backtracking walk method offers a fresh perspective on network analysis, particularly in understanding GitHub's developer communities. Behnaz Moradi-Jamei explains that traditional methods failed to capture the true nature of collaboration on GitHub, leading to the development of this innovative approach. This method assigns weights to collaborations based on their likelihood of forming sustained mutual connections, providing a more realistic depiction of community formation 1.

Imagine you are exploring a city, but with three special rules. First, you can't immediately go back to where you just came from. Second, if you complete a circle back to where you started, you begin a new walk. And third, you keep track of how often you use each street and mainly the last street that circles back to where you started.

--- Behnaz Moradi-Jamei

This approach, when combined with the Louvain Community Detection algorithm, reveals a more nuanced understanding of how developer communities form and function 2.

   

Network Methods

Innovative network methods are transforming how we analyze complex data sets, with Behnaz Moradi-Jamei highlighting the importance of ethical considerations in these analyses. Her work includes developing algorithms that introduce inclusivity parameters, balancing group inclusion with individual representation 3. This approach has been applied to various networks, including friendship and social networks, demonstrating the versatility of these methods.

I'm really interested in developing, you know, algorithms for community detection or in general for network analysis through a, you know, an ethic lens.

--- Behnaz Moradi-Jamei

Behnaz is also exploring network data analytics for social good, aiming to create algorithms that address complex structures while maintaining ethical integrity 4.