GIAI's primary research objective with the coming cycle's of MSc AI/Data Science is to build a graph-based Shapley Value for HR contribution analysis. In case you are not familiar with Shapley Value, it is a game-theory concept for properly allocating group project's gains/costs, which was first introduced in 1951 and awarded Nobel Prize in 2012.

The idea for this model originally came from one of the business case study classes(BUS501) in the MBA AI/Big Data program. In the class, students were given the task of testing a model to measure each student's contribution to group projects. Some students wanted to extend the model by incorporating participation in forum discussions as an additional metric.

This idea gained traction and has since been integrated into all course evaluations at the Swiss Institute of Artificial Intelligence (SIAI). Now, we aim to take this model beyond the classroom and make it more general and business-friendly. The goal is to refine it into a structured, scalable framework that can address a key challenge in corporate HR analytics: how to accurately measure multi-stage and indirect contributions in large organizations.

Understanding Team Contribution in Multi-Staged Work Environments

Traditional regression-based models for performance evaluation assign proportional credit based on direct contributions. While useful, they assume that all contributions are immediate and directly observable within a single stage of work. However, in real-world workplaces:

• Projects are multi-staged and often take months or years to complete.
• Some contributions emerge over time, rather than being immediately visible.
• Key individuals may act as connectors or enablers, rather than direct output producers.

To address these challenges, I am developing a new model that leverages graph-based Shapley value calculations. Unlike conventional models, this approach:

• Captures contributions that unfold over multiple project cycles.
• Identifies knowledge-sharing roles that support long-term success.
• Quantifies the impact of ‘helpers’ who enable others to succeed without always producing measurable outputs themselves.

Leveraging Communication Data to Measure Contribution

To make this model applicable in business settings, I plan to incorporate email and chat data as key sources of information. These internal communication networks serve as vital indicators of:

• How knowledge flows within an organization.
• Who provides critical insights, guidance, and solutions.
• Which employees are silent contributors who strengthen a team’s efficiency over time.

This naturally raises concerns about privacy, and I want to emphasize that ethical implementation is a key priority. While companies may find it reasonable to analyze work-related communication, employees must also have the right to:

• Opt out if they do not wish to be evaluated using this model.
• Maintain separate communication channels—one strictly for business, another for personal interactions.

---

Building on Traditional Contribution Models

This model does not aim to replace existing HR analytics but rather to complement them. Traditional evaluation methods already track:

✅ Task completion and project logs (Jira, Trello, Asana) ✅ Document collaboration (Google Docs, Notion, Confluence) ✅ Meeting participation and scheduling (Google Calendar, Outlook) ✅ Code commits and technical contributions (GitHub, GitLab)

However, these approaches primarily measure direct, immediate contributions. By integrating a graph-based structure, this model adds an extra rung on the ladder, allowing us to:

• Identify individuals whose contributions emerge across multiple projects.
• Detect key connectors and enablers within an organization.
• Assign Shapley value-based credit to those who facilitate success beyond direct outputs.

Why Does This Matter? The Role of 'Helpers' in Teams

Many workplaces unintentionally overlook contributors who are not direct project leaders. These individuals—whom I call 'helpers'—are vital in ensuring long-term efficiency, knowledge-sharing, and problem-solving.

• Traditional performance metrics reward project leaders, often missing those who facilitate success behind the scenes.
• A graph-based evaluation helps reveal these hidden contributors, ensuring fair recognition.
• Large-scale organizations rely on cross-team knowledge flow, which is difficult to quantify with traditional models.

By refining this methodology, we aim to provide a more balanced and fair assessment of who truly drives organizational success.

A Practical Application: Fairer Bonus Allocation

A major application of this research is in corporate HR, where annual bonus allocation is often based on direct deliverables. However:

❌ Employees who create long-term strategic advantages often go unnoticed. ❌ Those who enable cross-team collaboration are rarely rewarded. ❌ Many companies struggle to identify silent contributors who significantly impact multiple projects.

Our model seeks to address this by providing data-driven, fairer evaluations that recognize both direct and indirect contributions. This could help businesses:

• Improve bonus distribution fairness.
• Identify emerging leaders within the company.
• Strengthen team efficiency and collaboration.

Join the Research: MSc AI/Data Science at SIAI

This project is one of the key research opportunities in the MSc AI/Data Science program for the 2025-2026 cycle. This project demands more than just enthusiasm for AI—it requires the ability to navigate complex, multi-layered problems where business reality meets mathematical precision.

If you are passionate about:

🔹 Applying cutting-edge machine learning techniques to real-world business challenges. 🔹 Exploring AI-driven approaches to performance evaluation. 🔹 Using graph theory, game theory (Shapley value), and NLP for corporate applications.

Then this could be the perfect research opportunity for you.

💡 Exceptional students who demonstrate strong analytical skills and a commitment to AI-driven research may be considered for scholarships and funding opportunities.

However, I want to be clear—this is not a program for those seeking an easy credential. The MSc AI/Data Science at SIAI is for students who:

✔️ Want to work on serious, high-impact AI research. ✔️ Are ready to challenge traditional methods with new AI-driven approaches. ✔️ Aspire to develop solutions that companies can implement in real-world settings.

I welcome smart, ambitious, and research-driven students to join me in pushing the boundaries of AI for business.

Not sure if a year work will be enough to build a fully robust, easily modifiable, and conceptually intuitive model, but application of the work-in-progress model will be periodically shared as a form of case studies.

Necessary knowledge

• Game Theory
• Network Theory
• Machine Learning
• Large Language Model
• (Some level of) Panel data

Key concepts are discussed in PreMSc (or MBA AI), and deeper ones to come in MSc AI/Data Science.

Most AI-driven HR analytics focus on traditional models. We are developing an advanced, multi-stage contribution evaluation framework—something that could redefine how businesses measure and reward employees' true impact. This is not about minor improvements; this is about setting a new industry standard. Likely mind-set is also strongly emphasized.

If interested, feel free to ask questions in comments through GIAI Square.