The Mechanics of Integrity

ASAI operates on a four-layer verification engine. Here is exactly how we calculate the A-Index.

Step 01

10:1 Leadership Symmetry

Unlike traditional metrics that dilute credit equally, ASAI recognizes the "Drivers" of research. We utilize a 100% distribution model where citations are allocated based on functional authorship roles. Traditional metrics give 100% credit to every author on a paper. ASAI applies a intelligent model and role-weighting based on CRediT (Contributor Roles Taxonomy).

Weight_i = (R_i / Σ R_j) * Citations *Where R=10 for First/Last authors, and R=1 for middle authors. Every citation is 100% accounted for.*

Step 02

Semantic Weighting

Our NLP engine scans the full text of citing papers to determine the intent of the citation. Not all citations are equal.

Foundational: Paper uses your specific methodology or data. Weight: 1.0
Comparative: Your work is used as a benchmark or comparison. Weight: 0.5
Incidental: Your work is listed in a general literature review. Weight: 0.1

Step 03

Network Density Penalty (K-Factor)

We map the global citation graph. If our algorithm detects a "Closed Loop"—where Author A and B cite each other at rates far higher than average—a penalty is applied. To neutralize citation cartels, ASAI calculates a topological penalty based on co-authorship overlap. The raw impact is first calculated as a Provisional Score, then normalized against the candidate's network density.

Provisional_Score = max { i | Weighted_Score_i ≥ i }
K_Factor = 1 + (Max_Overlap / Total_Works)

A-Index = Provisional_Score / K_Factor *Weighted_Score incorporates 10:1 Role Attribution and Semantic intent.*

Final Step

Technical Knowledge Audit (TKA)

The calculation is only finalized after the TKA. This 15-minute AI-led dialogue verifies that the author possesses the "Conceptual DNA" of the paper. A failed audit results in a "Non-Verified" status.

Curious how the interrogation works?

Launch Interactive TKA Demo →