Analysis Logic Explained

A Deeper Look into the Analysis Engine

This document explains the methodology behind each step of the analysis pipeline.

1. Semantic Profile Analysis

• Objective: To move beyond resume keywords and quantitatively capture the true essence of an individual's expertise and experience.
• Methodology: We use a pre-trained AI language model (Sentence-Transformer) to convert each person's entire profile text into a high-dimensional "embedding vector." This is analogous to placing each profile on a vast "map of meaning," where profiles with similar contexts and expertise are located closer to one another.
• Why this approach? Simple keyword counting is brittle. It cannot distinguish between deep experience and a passing mention. By embedding the full text, we capture the context, nuance, and richness of a profile, providing a far more robust foundation for all subsequent analysis.

2. Synergy Calculation

• Objective: To compute a single, holistic "synergy score" for each possible pair within the team.
• Methodology: This score is a weighted composite of four key factors:

• Skill Complementarity: Calculated from the distance between the semantic vectors of two individuals. A greater distance implies more diverse, complementary skills, which results in a higher score.
• Experience Synergy: Assumes that individuals with similar years of experience can operate on a more equal footing, leading to smoother collaboration.
• Network Effects: Models the exponential value created by combining the professional networks of two well-connected individuals.
• Cultural Diversity: Applies a bonus to pairs with different cultural backgrounds, acknowledging the value of diverse perspectives in problem-solving.

• Why this approach? Synergy is not one-dimensional. A successful partnership requires a mix of complementary skills, shared context, external reach, and diverse viewpoints. This model attempts to quantify that multifaceted reality.

3. Dynamic System Modeling

• Objective: To forecast the team's performance over a 24-month period, demonstrating their growth potential.
• Methodology: We employ a coupled logistic growth model, a staple of system dynamics used to model population growth. Each individual's performance grows based on their own abilities and is accelerated by the synergy they share with their teammates. Crucially, this growth is constrained by a "Carrying Capacity (K)," a realistic performance ceiling.
• Why this approach? Teams don't grow infinitely. They learn, adapt, and eventually reach a state of peak effectiveness. The S-shaped growth curve produced by this model is a realistic representation of a team's journey from formation to maturity, making it a far more credible forecasting tool than a simple linear projection.

4. Game Theory Analysis (Shapley Values)

• Objective: To fairly attribute the team's total generated value to each individual member.
• Methodology: We use the Shapley value, a Nobel Prize-winning concept from cooperative game theory. It calculates each member's contribution by averaging their marginal impact across every possible subgroup, or "coalition," that could be formed. The "base value" for each person is derived from the magnitude of their semantic vector, reflecting the richness of their profile.
• Why this approach? A person's value is not just their individual output, but how they elevate others. The Shapley value is a mathematically sound method for capturing this, identifying "lynchpin" members who may not have the most senior title but whose presence is critical for unlocking the team's full potential.